National Ambient Air Quality Standards for Particulate ...38652 Federal Register/Vol. 62, No....

fede

ral r

egiste

r

38651

FridayJuly 18, 1997

Part II

EnvironmentalProtection Agency40 CFR Part 50National Ambient Air Quality Standardsfor Particulate Matter; Final Rule

38652 Federal Register / Vol. 62, No. 138 / Friday, July 18, 1997 / Rules and Regulations

ENVIRONMENTAL PROTECTIONAGENCY

40 CFR Part 50

[AD–FRL–5725–2]

RIN 2060–AE66

National Ambient Air QualityStandards for Particulate Matter

AGENCY: Environmental ProtectionAgency (EPA).ACTION: Final rule.SUMMARY: This document describesEPA’s decision to revise the nationalambient air quality standards (NAAQS)for particulate matter (PM) based on itsreview of the available scientificevidence linking exposures to ambientPM to adverse health and welfare effectsat levels allowed by the current PMstandards. The current primary PMstandards are revised in severalrespects: Two new PM2.5 standards areadded, set at 15 µg/m3 , based on the 3-year average of annual arithmetic meanPM2.5 concentrations from single ormultiple community-oriented monitors,and 65 µg/m 3 , based on the 3-yearaverage of the 98th percentile of 24-hourPM2.5 concentrations at eachpopulation-oriented monitor within anarea; and the current 24-hour PM10

standard is revised to be based on the99th percentile of 24-hour PM10

concentrations at each monitor withinan area. The new suite of primarystandards will provide increasedprotection against a wide range of PM-related health effects, includingpremature mortality and increasedhospital admissions and emergencyroom visits, primarily in the elderly andindividuals with cardiopulmonarydisease; increased respiratory symptomsand disease, in children and individualswith cardiopulmonary disease such asasthma; decreased lung function,particularly in children and individualswith asthma; and alterations in lungtissue and structure and in respiratorytract defense mechanisms. The currentsecondary standards are revised bymaking them identical in all respects tothe new suite of primary standards. Thenew secondary standards, inconjunction with a regional hazeprogram, will provide appropriateprotection against PM-related publicwelfare effects including soiling,material damage, and visibilityimpairment. In conjunction with thenew PM2.5 standards, a new referencemethod has been specified formonitoring PM as PM2.5 .EFFECTIVE DATE: This action is effectiveSeptember 16, 1997.ADDRESSES: A docket containinginformation relating to the EPA’s review

of the PM primary and secondarystandards (Docket No. A–95–54) isavailable for public inspection in theCentral Docket Section of the U.S.Environmental Protection Agency,South Conference Center, Rm. 4, 401 MSt., SW., Washington, DC. This docketincorporates the docket established forthe air quality Criteria Document(Docket No. ECAO–CD–92–0671). Thedocket may be inspected between 8 a.m.and 3 p.m., Monday through Friday,except legal holidays, and a reasonablefee may be charged for copying. Theinformation in the docket constitutesthe complete basis for the decisionannounced in this document. For theavailability of related information, see‘‘SUPPLEMENTARY INFORMATION.’’FOR FURTHER INFORMATION CONTACT: JohnH. Haines, MD–15, Air QualityStrategies and Standards Division,Office of Air Quality Planning andStandards, U.S. EnvironmentalProtection Agency, Research TrianglePark, NC 27711; telephone: (919) 541–5533; e-mail:[email protected] INFORMATION:

Related Final Rules on PM MonitoringIn a separate document published

elsewhere in this issue of the FederalRegister, EPA is amending its ambientair quality surveillance requirements (40CFR part 58) and its ambient airmonitoring reference and equivalentmethods (40 CFR part 53) for PM.

Availability of Related InformationCertain documents are available from

the U.S. Department of Commerce,National Technical Information Service,5285 Port Royal Road, Springfield, VA22161. Available documents include:

(1) Air Quality Criteria for ParticulateMatter (Criteria Document) (threevolumes, EPA/600/P–95–001aF thruEPA/600/P–95–001cF, April 1996, NTIS#PB–96–168224, $234.00 paper copy).

(2) Review of the National AmbientAir Quality Standards for ParticulateMatter: Policy Assessment of Scientificand Technical Information (Staff Paper)(EPA–452/R–96–013, July 1996, NTIS#PB–97–115406, $47.00 paper copy and$19.50 microfiche). (Add a $3.00handling charge per order.)

A limited number of copies of otherdocuments generated in connectionwith this standard review, such astechnical support documents pertainingto air quality, monitoring, and healthrisk assessment, can be obtained from:Environmental Protection AgencyLibrary (MD–35), Research TrianglePark, NC 27711, telephone (919) 541–2777. These and other relateddocuments are also available for

inspection and copying in the EPAdocket at the address under‘‘ADDRESSES,’’ at the beginning of thisdocument.

Electronic AvailabilityThe Staff Paper and human health

risk assessment support documents areavailable on the Agency’s Office of AirQuality Planning and Standards’(OAQPS) Technology Transfer Network(TTN) Bulletin Board System (BBS) inthe Clean Air Act Amendments area,under Title I, Policy/GuidanceDocuments. To access the bulletinboard, a modem and communicationssoftware are necessary. To dial up, setyour communications software to 8 databits, no parity and one stop bit. Dial(919) 541–5742 and follow the on-screen instructions to register for access.After registering, proceed to choice‘‘<T> Gateway to TTN TechnicalAreas’’, then choose ‘‘<E> CAAA BBS’’.From the main menu, choose ‘‘<1> TitleI: Attain/Maint of NAAQS’’, then ‘‘<P>Policy Guidance Documents.’’ To accessthese documents through the WorldWide Web, click on ‘‘TTN BBSWeb’’,then proceed to the Gateway to TTNTechnical areas, as above. If assistanceis needed in accessing the system, callthe help desk at (919) 541–5384 inResearch Triangle Park, NC.

Implementation Strategy For RevisedAir Quality Standards

On Wednesday, July 16, 1997,President Clinton signed amemorandum to the Administratorspecifying his goals for theimplementation of the O3 and PMstandards. Attached to the President’smemorandum is a strategy prepared byan interagency Administration groupoutlining the next steps that would benecessary for implementing thesestandards. The EPA will prepareguidance and proposed rules consistentwith the President’s memorandum.Copies of the Presidential document areavailable in paper copy by contactingthe U.S. Environmental ProtectionAgency Library at the address under‘‘Availability of Related Information’’and in electronic form as discussedabove in ‘‘Electronic Availability.’’

The following topics are discussed inthis preamble:I. Background

A. Legislative RequirementsB. Related Control RequirementsC. Review of Air Quality Criteria and

Standards for PMD. Summary of Proposed Revisions to the

PM StandardsII. Rationale for the Primary PM Standards

A. IntroductionB. Need for Revision of the Current

Primary PM Standards

38653Federal Register / Vol. 62, No. 138 / Friday, July 18, 1997 / Rules and Regulations

C. Indicators of PMD. Averaging Time of PM2.5 StandardsE. Form of PM2.5 StandardsF. Levels for the Annual and 24-Hour PM2.5

StandardsG. Conclusions Regarding the Current PM10

StandardsH. Final Decisions on Primary PM

StandardsIII. Rationale for the Secondary Standards

A. Need for Revision of the CurrentSecondaryStandards

B. Decision on the Secondary StandardsIV. Other Issues

A. Consideration of CostsB. Margin of SafetyC. Data AvailabilityD. 1990 Amendments

V. Revisions to 40 CFR Part 50, AppendixK—Interpretation of the PM NAAQS

A. PM2.5 Computations and Data HandlingConventions

B. PM10 Computations and Data HandlingConventions

C. Changes that Apply to Both PM2.5 andPM10 Computations

VI. Reference Methods for the Determinationof Particulate Matter as PM10 and PM2.5

in the AtmosphereA. Revisions to 40 CFR Part 50, Appendix

J—Reference Method for PM10

B. 40 CFR Part 50, Appendix L—NewReference Method for PM2.5

VII. Effective Date of the Revised PMStandards and Applicability of theExisting PM10 Standards

VIII. Regulatory and Environmental ImpactAnalyses

A. Executive Order 12866B. Regulatory Flexibility AnalysisC. Impact on Reporting RequirementsD. Unfunded Mandates Reform ActE. Environmental JusticeF. Submission to Congress and the

Comptroller GeneralIX. Response to Petition for Administrator

Browner’s RecusalX. References

I. Background

A. Legislative RequirementsTwo sections of the Clean Air Act

(Act) govern the establishment, review,and revision of NAAQS. Section 108 ofthe Act (42 U.S.C. 7408) directs theAdministrator to identify certainpollutants which ‘‘may reasonably beanticipated to endanger public healthand welfare’’ and to issue air qualitycriteria for them. These air qualitycriteria are to ‘‘accurately reflect thelatest scientific knowledge useful inindicating the kind and extent of allidentifiable effects on public health orwelfare which may be expected from thepresence of [a] pollutant in the ambientair * * *.’’

Section 109 of the Act (42 U.S.C.7409) directs the Administrator topropose and promulgate ‘‘primary’’ and‘‘secondary’’ NAAQS for pollutantsidentified under section 108 of the Act.Section 109(b)(1) of the Act defines a

primary standard as one ‘‘the attainmentand maintenance of which in thejudgment of the Administrator, based on[the] criteria and allowing an adequatemargin of safety, are requisite to protectthe public health.’’ The margin of safetyrequirement was intended to addressuncertainties associated withinconclusive scientific and technicalinformation available at the time ofstandard setting, as well as to provide areasonable degree of protection againsthazards that research has not yetidentified. Both kinds of uncertaintiesare components of the risk associatedwith pollution at levels below those atwhich human health effects can be saidto occur with reasonable scientificcertainty. Thus, by selecting primarystandards that provide an adequatemargin of safety, the Administrator isseeking not only to prevent pollutionlevels that have been demonstrated to beharmful but also to prevent lowerpollutant levels that she finds may posean unacceptable risk of harm, even if therisk is not precisely identified as tonature or degree. The Act does notrequire the Administrator to establish aprimary NAAQS at a zero-risk level, butrather at a level that reduces risksufficiently so as to protect publichealth with an adequate margin ofsafety. The selection of any particularapproach to providing an adequatemargin of safety is a policy choice leftspecifically to the Administrator’sjudgment. Lead Industries Ass’n v. EPA,647 F.2d 1130, 1161–1162 (D.C.Cir.1980).

A secondary standard, as defined insection 109 (b)(2) of the Act, must‘‘specify a level of air quality theattainment and maintenance of which inthe judgment of the Administrator,based on [the] criteria, [are] requisite toprotect the public welfare from anyknown or anticipated adverse effectsassociated with the presence of [the]pollutant in the ambient air.’’ Welfareeffects as defined in section 302(h) ofthe Act (42 U.S.C. 7602(h)) include, butare not limited to, ‘‘effects on soils,water, crops, vegetation, manmadematerials, animals, wildlife, weather,visibility, and climate, damage to anddeterioration of property, and hazards totransportation, as well as effects oneconomic values and on personalcomfort and well-being.’’

Section 109(d)(1) of the Act requiresperiodic review and, if appropriate,revision of existing air quality criteriaand NAAQS. Section 109(d)(2) of theAct requires appointment of anindependent scientific reviewcommittee to review criteria andstandards and recommend newstandards or revisions of existing

criteria and standards, as appropriate.The committee established undersection 109(d)(2) of the Act is known asthe Clean Air Scientific AdvisoryCommittee (CASAC), a standingcommittee of EPA’s Science AdvisoryBoard.

B. Related Control Requirements

States are primarily responsible forensuring attainment and maintenance ofambient air quality standards once EPAhas established them. Under section 110of the Act (42 U.S.C. 7410) and relatedprovisions, States are to submit, for EPAapproval, State implementation plans(SIP’s) that provide for the attainmentand maintenance of such standardsthrough control programs directed tosources of the pollutants involved. TheStates, in conjunction with EPA, alsoadminister the prevention of significantdeterioration program (42 U.S.C. 7470–7479) for these pollutants. In addition,Federal programs provide fornationwide reductions in emissions ofthese and other air pollutants throughthe Federal Motor Vehicle ControlProgram under Title II of the Act (42U.S.C. 7521–7574), which involvescontrols for automobile, truck, bus,motorcycle, nonroad engine, and aircraftemissions; the new source performancestandards under section 111 of the Act(42 U.S.C. 7411); and the nationalemission standards for hazardous airpollutants under section 112 of the Act(42 U.S.C. 7412).

C. Review of Air Quality Criteria andStandards for PM

Particulate matter is the generic termfor a broad class of chemically andphysically diverse substances that existas discrete particles (liquid droplets orsolids) over a wide range of sizes.Particles originate from a variety ofanthropogenic stationary and mobilesources as well as from natural sources.Particles may be emitted directly orformed in the atmosphere bytransformations of gaseous emissionssuch as sulfur oxides (SOx), nitrogenoxides (NOx), and volatile organiccompounds (VOC). The chemical andphysical properties of PM vary greatlywith time, region, meteorology, andsource category, thus complicating theassessment of health and welfare effects.

The last review of PM air qualitycriteria and standards was completed inJuly 1987 with notice of a final decisionto revise the existing standardspublished in the Federal Register (52FR 24854, July 1, 1987). In that decision,EPA changed the indicator for PM fromtotal suspended particles (TSP) to


1 PM10 refers to particles with an aerodynamicdiameter less than or equal to a nominal 10micrometers. Technical details further specifyingthe measurement of PM10 are contained in 40 CFRpart 50, Appendices J and M.

2 A more complete history of the PM NAAQS ispresented in section II.B of the OAQPS Staff Paper,Review of National Ambient Air Quality Standardsfor Particulate Matter: Assessment of Scientific andTechnical Information (U.S. EPA, 1996b).

3 A court order entered in American LungAssociation v. Browner, CIV–93–643–TUC–ACM (D.Ariz.,October 6, 1994), as subsequently modified,requires publication of EPA’s final decision on thereview of the PM NAAQS by July 19, 1997.

4 The Staff Paper evaluates policy implications ofthe key studies and scientific information in theCriteria Document, identifies critical elements thatEPA staff believes should be considered, andpresents staff conclusions and recommendations ofsuggested options for the Administrator’sconsideration.

5 PM2.5 refers to particles with an aerodynamicdiameter less than or equal to a nominal 2.5micrometers, as further specified in 40 CFR part 50,Appendix L in this document.

6 PM10–2.5 refers to those particles with anaerodynamic diameter less than or equal to anominal 10 micrometers but greater than 2.5micrometers. In other words, it refers to theinhalable particles that remain if fine (PM2.5)particles are removed from a sample of PM10

particles.

PM10.1 Identical primary and secondaryPM10 standards were set for twoaveraging times: 50 µg/m3, expectedannual arithmetic mean, averaged over3 years, and 150 µg/m3, 24-hour average,with no more than one expectedexceedance per year.2

The EPA initiated this current reviewof the air quality criteria and standardsfor PM in April 1994 by announcing itsintention to develop a revised AirQuality Criteria Document forParticulate Matter (henceforth, the‘‘Criteria Document’’). Thereafter, theEPA presented its plans for review ofthe criteria and standards for PM undera highly accelerated, court-orderedschedule3 at a public meeting of theCASAC in December 1994. Severalworkshops were held by EPA’s NationalCenter for Environmental Assessment(NCEA) to discuss important new healtheffects information in November 1994and January 1995. External review draftsof the Criteria Document were madeavailable for public comment and werereviewed by CASAC at public meetingsheld in August and December 1995 andFebruary 1996. The CASAC came toclosure in its review of the CriteriaDocument, advising the Administratorin a March 15, 1996 closure letter(Wolff, 1996a) that ‘‘although ourunderstanding of the health effects ofPM is far from complete, a revisedCriteria Document which incorporatesthe Panel’s latest comments will providean adequate review of the availablescientific data and relevant studies ofPM.’’ CASAC and public commentsfrom these meetings, and fromsubsequent written comments and theclosure letter, were incorporated asappropriate in the final CriteriaDocument (U.S. EPA, 1996a).

External review drafts of a Staff Paperprepared by the Office of Air QualityPlanning and Standards (OAQPS),Review of the National Ambient AirQuality Standards for Particulate Matter:Assessment of Scientific and TechnicalInformation (henceforth, the ‘‘StaffPaper’’), were made available for publiccomment and were reviewed by CASACat public meetings in December 1995

and May 1996.4 The CASAC came toclosure in its review of the Staff Paper,advising the Administrator in a June 13,1996 closure letter (Wolff, 1996b) that‘‘the Staff Paper, when revised, willprovide an adequate summary of ourpresent understanding of the scientificbasis for making regulatory decisionsconcerning PM standards.’’ CASAC andpublic comments from these meetings,subsequent written comments, and theCASAC closure letter were incorporatedas appropriate in the final Staff Paper(U.S. EPA, 1996b).

On November 27, 1996, EPAannounced its proposed decision torevise the NAAQS for PM (61 FR 65638,December 13, 1996) (hereafter‘‘proposal’’) as well as its proposeddecision to revise the NAAQS for ozone(O3)(61 FR 65716, December 13, 1996).In the proposal, EPA identifiedproposed revisions, based on the airquality criteria for PM, and solicitedpublic comments on alternative primarystandards and on the proposed forms ofthe standards.

To ensure the broadest possiblepublic input on the PM and O3

proposals, EPA took extensive andunprecedented steps to facilitate thepublic comment process beyond thenormal process of providing anopportunity to request a hearing andreceiving written comments submittedto the rulemaking docket. The EPAestablished a national toll-freetelephone hotline to facilitate publiccomments on the proposed revisions tothe PM and O3 NAAQS, and on relatednotices dealing with the implementationof revised PM and O3 standards, as wellas a system for the public to submitcomments on the proposalselectronically via the Internet. Over14,000 calls and over 4,000 electronicmail messages were received throughthese channels. The public could alsoaccess key supporting documents(including the Criteria Document, StaffPaper, related technical documents andfact sheets) via the Internet.

The EPA also held several publichearings and meetings across thecountry to provide direct opportunitiesfor public comment on the proposedrevisions to the PM and O3 NAAQS andto disseminate information to the publicabout the proposed standard revisions.On January 14 and 15, 1997, EPA heldconcurrent, 2-day public hearings inBoston, MA, Chicago, IL, and Salt Lake

City, UT. A fourth public hearing,which focused primarily on PMmonitoring issues, was held in Durham,NC on January 14, 1997. Over 400citizens and organizations testifiedduring these public hearings. EPA alsoheld two national satellite telecasts toanswer questions on the standards andparticipated in meetings sponsored bythe Air and Waste ManagementAssociation on the proposed revisionsto the standards at more than 10locations across the country. Beyondthat, several EPA regional offices heldpublic meetings and workshops andparticipated in hearings that States andcities held around the country.

As a result of this intensive effort tosolicit public input, over 50,000 writtenand oral comments were received on theproposed revisions to the PM NAAQSby the close of the public commentperiod on March 12, 1997. Major issuesraised in the comments are discussedthroughout the preamble of this finaldecision. A comprehensive summary ofall significant comments, along withEPA’s response to such comments(hereafter ‘‘Response to Comments’’),can be found in the docket for thisrulemaking (Docket No. A–95–54).

The principal focus of this currentreview of the air quality criteria andstandards for PM is on recentepidemiological evidence reportingassociations between ambientconcentrations of PM and a range ofserious health effects. Particularattention has been given to several size-specific classes of particles, includingPM10 and the principal fractions ofPM10, referred to as the fine (PM2.5)5 andcoarse (PM10–2.5)6 fractions. Asdiscussed in the Criteria Document, fineand coarse fraction particles can bedifferentiated by their sources andformation processes and their chemicaland physical properties, includingbehavior in the atmosphere. Detaileddiscussions of atmospheric formation,ambient concentrations, and health andwelfare effects of PM, as well asquantitative estimates of human healthrisks associated with exposure to PM,can be found in the Criteria Documentand in the Staff Paper.


D. Summary of Proposed Revisions tothe PM Standards

For reasons discussed in the proposal,the Administrator proposed to revise thecurrent primary standards for PM (asindicated by PM10), by adding two newprimary PM2.5 standards set at 15 µg/m3,annual mean, and 50 µg/m3, 24-houraverage. The proposed annual PM2.5

standard would be based on the 3-yearaverage of the annual arithmetic meanPM2.5 concentrations, spatially averagedacross an area. The proposed 24-hourPM2.5 standard would be based on the3-year average of the 98th percentile of24-hour PM2.5 concentrations at eachpopulation-oriented monitor within anarea. The proposal solicited commenton two alternative approaches forselecting the levels of PM2.5 standards.The Administrator also proposed torevise the current 24-hour primary PM10

standard of 150 µg/m3 by replacing the1-expected-exceedance form with a 98th

percentile form, averaged over 3 years ateach monitor within an area, solicitedcomment on an alternative proposal torevoke the 24-hour PM10 standard, andproposed to retain the current annualprimary PM10 standard of 50 µg/m3. Theproposal also solicited comment onproposed revisions to 40 CFR part 50,Appendix K to establish new datahandling conventions for calculating98th percentile values and spatialaverages, revisions to 40 CFR part 50,Appendix J to modify the referencemethod for monitoring PM as PM10, anda proposed new reference method formonitoring PM as PM2.5 (40 CFR part 50,Appendix L).

With regard to the secondarystandards, the Administrator proposedto revise the current secondarystandards by making them identical tothe suite of proposed primary standards,in conjunction with the establishment ofa regional haze program under section169A of the Act.

II. Rationale for the Primary Standards

A. Introduction

1. Overview. This document presentsthe Administrator’s final decisionsregarding the need to revise the currentprimary ambient air quality standardsfor PM, and, more specifically,regarding the establishment of newannual and 24-hour PM2.5 primarystandards and revisions to the form ofthe current 24-hour PM10 primaryNAAQS. These decisions are based ona thorough review, in the CriteriaDocument, of the latest scientificinformation on known and potentialhuman health effects associated withexposure to PM at levels typically found

in the ambient air. These decisions alsotake into account:

(1) Staff Paper assessments of themost policy-relevant information in theCriteria Document, upon which staffrecommendations for new and revisedprimary standards are based.

(2) CASAC advice andrecommendations, as reflected indiscussions of drafts of the CriteriaDocument and Staff Paper at publicmeetings, in separate written comments,and in the CASAC’s closure letters tothe Administrator.

(3) Public comments received duringthe development of these documents,either in connection with CASACmeetings or separately.

(4) Extensive public commentsreceived on the proposed decisionsregarding the primary PM standards.

After taking this information andcomments into account, and for thereasons discussed below in this unit, theAdministrator concludes that revisionsto the current primary standards toprovide increased public healthprotection against a variety of healthrisks are appropriate. More specifically,the Administrator has determined that itis appropriate to establish new annualand 24-hour PM2.5 standards, to revisethe current 24-hour PM10 standard, andto retain the current annual PM10

standard. As discussed more fully belowin this unit, the rationale for the finaldecisions regarding the PM primaryNAAQS includes consideration of:

(1) Health effects information, andalternative views on the appropriateinterpretation and use of theinformation, as the basis for judgmentsabout the risks to public healthpresented by population exposures toambient PM.

(2) Insights gained from a quantitativerisk assessment conducted to provide abroader perspective for judgments aboutprotecting public health from the risksassociated with PM exposures.

(3) Specific conclusions regarding theneed for revisions to the currentstandards and the elements of PMstandards (i.e., indicator, averagingtime, form, and level) that, takentogether, would be appropriate toprotect public health with an adequatemargin of safety.

As with virtually any policy-relevantscientific research, there is uncertaintyin the characterization of health effectsattributable to exposure to ambient PM.As discussed in the proposal, however,there is now a greatly expanded body ofhealth effects information as comparedwith that available during the lastreview of the PM standards. Moreover,the recent evidence on PM-relatedhealth effects has undergone an

unusually high degree of scrutiny andreanalysis over the past several years,beginning with a series of workshopsheld early in the review process todiscuss important new information. Anumber of opportunities were providedfor public comment on successive draftsof the Criteria Document and StaffPaper, as well as for intensive peerreview of these documents by CASAC atseveral public meetings attended bymany knowledgeable individuals andrepresentatives of interestedorganizations. In addition, there havebeen a number of important scientificconferences, symposia, and colloquia onPM issues, sponsored by the EPA andothers, in the U.S. and abroad, duringthis period. While significantuncertainties exist, the review of thehealth effects information has beenthorough and deliberate. In thejudgment of the Administrator, thisintensive evaluation of the scientificevidence has provided an adequatebasis for regulatory decision making atthis time, as well as for thecomprehensive research needsdocument recently developed by EPA,and reviewed by CASAC and others, forimproving our future understanding ofthe relationships between ambient PMexposures and health effects.

The health effects information andhuman risk assessment weresummarized in the proposal and areonly briefly outlined below in this unit.Subsequent units provide a morecomplete discussion of theAdministrator’s rationale, in light of keyissues raised in public comments, forconcluding that it is appropriate torevise the current primary standards(Unit II.B. of this preamble) and torevise the specific elements of thestandards including indicator (Unit II.C.of this preamble); averaging time, form,and level of new PM2.5 standards (UnitsII.D., II.E., and II.F. of this preamble);and averaging time, form, and level ofrevised PM10 standards (Unit II.G. ofthis preamble).

2. Summary of the health effectsevidence. In brief, since the last reviewof the PM criteria and standards, themost significant new evidence on thehealth effects of PM is the greatlyexpanded body of communityepidemiological studies. The CriteriaDocument stated that these recentstudies provide ‘‘evidence that serioushealth effects (mortality, exacerbation ofchronic disease, increased hospitaladmissions, etc.) are associated withexposures to ambient levels of PMfound in contemporary U.S. urbanairsheds even at concentrations belowcurrent U.S. PM standard’’ (U.S. EPA,1996a; p. 13-1). Although a variety of


7 The risk assessment results that appear in theStaff Paper and are summarized in the proposalhave been updated to include analyses of theparticular forms of standard alternatives containedin the proposal and to correct estimates for oneeffects category (mortality from long-term exposure)to reflect the actual statistics used in the study uponwhich they were based (Pope et al., 1995). Thecorrections, which cumulatively reduce estimates ofmortality associated with long-term exposures by 20to 35%, have no effect on risk estimates formortality associated with short-term exposures orthe estimates for any other effects. Because the keysensitivity analyses that provide additional insightsregarding thresholds, copollutants, averaging timeand related issues involved the short-term exposurestudies, none of these results are affected bychanges to the long-term exposure risk estimates.

responses to constituents of ambient PMhave been hypothesized to contribute tothe reported health effects, the relevanttoxicological and controlled humanstudies published to date have notidentified any accepted mechanism(s)that would explain how such relativelylow concentrations of ambient PMmight cause the health effects reportedin the epidemiological literature.

Unit II.A. of the proposal furtheroutlines key information contained inthe Criteria Document, Chapters 10-13,and the Staff Paper, Chapter V, on theknown and potential health effectsassociated with airborne PM, alone andin combination with other pollutantsthat are routinely present in the ambientair. The information highlighted theresummarizes:

(1) The nature of the effects that havebeen reported to be associated withambient PM, which include prematuremortality, aggravation of respiratory andcardiovascular disease (as indicated byincreased hospital admissions andemergency room visits, school absences,work loss days, and restricted activitydays), changes in lung function andincreased respiratory symptoms,changes to lung tissues and structure,and altered respiratory defensemechanisms.

(2) Sensitive subpopulations thatappear to be at greater risk to sucheffects, specifically individuals withrespiratory disease and cardiovasculardisease and the elderly (prematuremortality and hospitalization), children(increased respiratory symptoms anddecreased lung function), and asthmaticchildren and adults (aggravation ofsymptoms).

(3) An integrated evaluation of thehealth effects evidence, with anemphasis on the key issues raised inassessing community epidemiologicalstudies, including alternativeinterpretations of the evidence, both forindividual studies and for the evidenceas a whole.

(4) The PM fractions of greatestconcern to health.

The summary in the proposal will notbe repeated here. EPA emphasizes thatthe final decisions on these standardstake into account the morecomprehensive and detailed discussionsof the scientific information on theseissues contained in the CriteriaDocument and Staff Paper, which werereviewed by the CASAC and the public.

3. Key insights from the riskassessment. The Staff Paper presents theresults of a quantitative assessment ofhealth risks for two example cities,including risk estimates for severalcategories of health effects associated

with: existing PM air quality levels,projected PM air quality levels thatwould occur upon attainment of thecurrent PM10 standards, and projectedPM air quality levels that would occurupon attainment of alternative PM2.5

standards. The risk assessment isintended as an aid to the Administratorin judging which alternative PMNAAQS would reduce risks sufficientlyto protect public health with anadequate margin of safety, recognizingthat such standards will not be risk-free.The risk assessment is described morefully in the Staff Paper and summarizedin the proposal. Related technicalreports and updates7 have been placedin the docket (Abt Associates, 1996a,b;1997a,b).

EPA emphasizes that it places greaterweight on the overall conclusionsderived from the studies—that PM airpollution is likely causing orcontributing to significant adverseeffects at levels below those permittedby the current standards—than on thespecific concentration-responsefunctions and quantitative risk estimatesderived from them. These quantitativerisk estimates include significantuncertainty and, therefore, should notbe viewed as demonstrated healthimpacts. EPA believes, however, thatthey do represent reasonable estimatesas to the possible extent of risk for theseeffects given the available information.Keeping in mind the importantuncertainties inherent in any suchanalyses, the key insights from the riskassessment that are most pertinent tothe current decision include:

(1) Fairly wide ranges of estimates ofthe incidence of PM-related mortalityand morbidity effects and riskreductions associated with attainment ofalternative standards were calculated forthe two locations analyzed when theeffects of key uncertainties andalternative assumptions wereconsidered. Significantly, the combinedanalysis for these two cities alone foundthat the risk remaining after attainingthe current PM10 standards was on the

order of hundreds of premature deathseach year, hundreds to thousands ofrespiratory-related hospital admissions,and tens of thousands of additionalrespiratory related symptoms inchildren.

(2) Based on the results from thesensitivity analyses of key uncertaintiesand the integrated uncertainty analyses,the single most important factorinfluencing the uncertainty associatedwith the risk estimates is whether or nota threshold concentration exists belowwhich PM-associated health risks arenot likely to occur.

(3) Over the course of a year, the fewpeak 24-hour PM2.5 concentrationsappear to contribute a relatively smallamount to the total health risk posed bythe entire air quality distribution ascompared to the aggregated risksassociated with the low to mid-rangeconcentrations.

(4) There is greater uncertainty aboutboth the existence and the magnitude ofestimated excess mortality and othereffects associated with PM exposures asone considers increasingly lowerconcentrations approaching backgroundlevels.

B. Need for Revision of the CurrentPrimary PM Standards

1. Introduction. The overarching issuein the present review of the primaryNAAQS is whether, in view of theadvances in scientific knowledgereflected in the Criteria Document andStaff Paper, the existing PM standardsshould be revised and, if so, whatrevised or new standards would beappropriate. The concluding section ofthe integrative synthesis of healtheffects information in the CriteriaDocument, which CASAC characterizedas EPA’s ‘‘best ever example of a trueintegrative summary of the state ofknowledge about the health effects ofairborne PM,’’ (Wolff, 1996b) providesthe following summary of the sciencewith respect to this issue:

The evidence for PM-related effects fromepidemiological studies is fairly strong, withmost studies showing increases in mortality,hospital admissions, respiratory symptoms,and pulmonary function decrementsassociated with several PM indices. Theseepidemiological findings cannot be whollyattributed to inappropriate or incorrectstatistical methods, misspecification ofconcentration-effect models, biases in studydesign or implementation, measurementerrors in health endpoint, pollutionexposure, weather, or other variables, norconfounding of PM effects with effects ofother factors. While the results of theepidemiological studies should beinterpreted cautiously, they nonethelessprovide ample reason to be concerned thatthere are detectable health effects attributable


8 As discussed more fully below in this unit,epidemiological studies alone cannot be used todemonstrate mechanisms of action, but they canprovide evidence useful in making inferences withregard to causal relationships (U.S. EPA, 1996b, p.V-9).

9 As noted in the proposal, the kinds of effectsobserved in the epidemiological studies arelogically related. For example, the association ofPM with mortality is mainly linked to respiratoryand cardiovascular causes, which is coherent withobserved PM associations with respiratory andcardiovascular hospital admissions and respiratorysymptoms. Further, similar categories of effects areseen in long- and short-term exposure studies.

to PM at levels below the current NAAQS.[U.S. EPA, 1996a, p. 13-92]

Given the nature of the health effectsin question, this finding, which is basedon a large number of studies that usedPM10 measurements, as well as studiesusing other indicators of PM, clearlyindicates that revision of the current PMNAAQS is appropriate. Quite apart fromthe issue of whether PM10 should be thesole indicator for the PM NAAQS, theextensive PM epidemiological data baseprovides evidence of serious healtheffects (e.g., mortality, exacerbation ofchronic disease, increased hospitaladmissions) in sensitive populations(e.g., the elderly, individuals withcardiopulmonary disease), as well assignificant adverse health effects (e.g.,increased respiratory symptoms, schoolabsences, and lung functiondecrements) in children. Moreover,these effects associations are observedin areas or at times when the levels ofthe current PM10 standards are met.Although the increase in relative risk issmall for the most serious outcomes,EPA believes it is significant from anoverall public health perspective,because of the large number ofindividuals in sensitive populations thatare exposed to ambient PM, as well asthe significance of the health effectsinvolved (U.S. EPA, 1996a, p. 1-21). Theresults of the two-city PM riskassessment reinforce these conclusionsregarding the significance of the publichealth risk—even under a scenario inwhich the current PM10 standards areattained.

While the lack of demonstratedmechanisms that explain the extensivebody of epidemiological findings is animportant caution, which presentsdifficulties in providing an integratedassessment of PM health effectsresearch, a number of potentialmechanisms have been hypothesized inthe recent literature (U.S. EPA, 1996b; p.V-5 to V-8; appendix D). Moreover,qualitative information from laboratorystudies of the effects of particlecomponents at high concentrations anddosimetry considerations suggest thatthe kinds of effects observed incommunity studies (e.g., respiratory-and cardiovascular-related responses)are at least plausibly related toinhalation of PM.8 Indeed, as discussedin the Criteria Document and sectionV.E of the Staff Paper, the consistencyof the results of the epidemiologicalstudies from a large number of different

locations and the coherent nature of theobserved effects9 are suggestive of alikely causal role of ambient PM incontributing to the reported effects.

2. Comments on scientific basis forrevision. A majority of the publiccomments received on the proposalagreed that, based on the availablescientific information, the current PM10

standards are not of themselvessufficient to protect public health and itwould be appropriate to revise them.Included in those calling for revisions tothe current standards are many publichealth professionals, includingnumerous medical doctors andacademic researchers. For example, agroup of 27 members of the scientificand medical community recognized ashaving substantial expertise inconducting research on the healtheffects of air pollution stated:

Health studies conducted in the U.S. andaround the world have demonstrated thatlevels of particulate and ozone air pollutionbelow the current U.S. National Air QualityStandards exacerbate serious respiratorydisease and contribute to early death. A largebody of scientific and medical evidenceclearly indicates that the current NAAQS arenot sufficiently protective of public health.[Thurston, 1997]

Similar conclusions were reached in aletter signed by more than 1,000scientists, clinicians, researchers, andother health care professionals (Dickey,1997). The cosigners to this letter arguedthat tens of thousands of hospital visitsand premature deaths could beprevented with the proposed air qualitystandard revisions. In fact, thesecommenters argued that even strongerstandards than those proposed by EPAare needed to protect the health of themost vulnerable residents of ourcommunities.

A number of State and localgovernment authorities also submittedcomments in support of adopting newair quality standards for fine particulatematter. The commenters concurred withconclusions reached through the EPA’speer review process that the PMstandards should be revised to protectpublic health. A number of thesecommenters suggested that thestandards proposed by EPA should beeven stronger, while several other Stateagencies recommended that EPA adoptPM2.5 standards, but at less stringentlevels. A number of the comments from

states supporting even strongerstandards acknowledged the lack ofdemonstrated mechanism(s) and otheruncertainties but stressed the strength ofthe other evidence in urging EPA to setprotective standards.

Many comments were also receivedfrom representatives of environmentalor community health organizations thatsupported the adoption of air qualitystandards for PM2.5. These commentersagreed with EPA’s finding that a largebody of compelling evidencedemonstrates that exposure toparticulate matter pollution, in general,is associated with premature death,aggravation of heart and lung diseases,increased respiratory illness andreduced lung function. They agreedwith EPA that these studies present aconsistent and coherent relationshipbetween exposure to PM and bothmortality and various measures ofmorbidity. However, the majority ofthese commenters argued that EPA’sproposed standards for PM2.5 wereinadequate and recommended adoptionof more stringent levels of the 24-hourand/or annual air quality standards forPM2.5. Many of these commenters alsourged EPA to revise the NAAQS forPM10 to be more protective of publichealth. These commenters based theirrecommendations on the findings of thestudies that were reviewed in thepreparation of the Criteria Documentand Staff Paper. One commenter usedresults from five of these studies as thebasis for recommending PM2.5 standardsof 10 µg/m3 (annual) and 18 µg/m3 (24-hour) (Dockery et al., 1993; Pope et al.,1995; Schwartz et al., 1996; Schwartz etal., 1994; Thurston et al., 1994). Thecommenters agreed with EPA on thesignificance of these studies’ results andthe need to revise the PM standards,while differing with EPA’sinterpretation of the findings forpurposes of developing the proposedPM standards.

Several commenters made referenceto the conclusions of a number ofinternational scientific panels regardingthe health effects of exposure toairborne particulate matter—the BritishExpert Panel on Air Quality Standards,the British Committee on the MedicalEffects of Air Pollutants, the WorldHealth Organization, the CanadianMinistry of Environment, Lands andParks, and the Health Council of theNetherlands -- and argued that all thesepanels found that PM concentrationsequivalent to the current U.S. standardsfor PM10 are not protective of humanhealth and made recommendations forgreater protection. One commenternoted that the findings of the BritishHealth Panel have resulted in a British


proposal to adopt a 24-hour PM10

standard of 50 µg/m3, which is one-thirdthe level of the current U.S. NAAQS.

In these comments, sometoxicological studies were cited asproviding evidence for toxicity ofparticulate pollution. These commentersdisagreed with arguments that PMstandards cannot be adopted due to alack of a sufficient understanding of thebiological mechanism of injury. Thecommenters argued that there issufficient evidence that particulatepollution is associated with adversehealth effects to make it inappropriate todelay the establishment of standardswhile further studies are undertaken.This group of commenters was alsocritical of arguments against theestablishment of additional PMstandards based on the possibility ofconfounding by other pollutants, andurged that more attention be paidinstead to the possible additive orsynergistic effects of multiple pollutantexposures.

In general, the EPA agrees with thesecommenters’ arguments regarding theneed to revise the PM standards. Thescientific studies cited by thesecommenters were the same studies usedin the development of the CriteriaDocument and the Staff Paper, and theEPA agrees that there is a sufficientbody of evidence that the currentNAAQS for PM are not adequatelyprotective of the public health. Forreasons detailed in Unit II.F. of thispreamble and in the Response toComments, EPA disagrees with aspectsof these commenters’ views on the levelof protection that is appropriate andsupported by the available scientificinformation.

Another body of commenters,including almost all commentersrepresenting businesses and industryassociations, many local governmentalgroups and private citizens, and someStates opposed revising the standards.Many of these commenters argued thatthe available scientific evidence doesnot provide an adequate basis forrevising the current standards. Thecentral arguments made by thesecommenters can be divided into twocategories: (1) General comments on theappropriateness of relying on theepidemiological evidence for makingregulatory decisions, and (2) morespecific comments challenging EPA’sappraisal of the consistency andcoherence of the available information,EPA’s conclusions regarding causality,and the use of these studies for riskassessment and decisions on whether torevise the standards. While EPA hasincluded comprehensive responses tothese comments in the Response to

Comments, certain key points aresummarized below in this unit.

a. General comments on the use ofepidemiological studies. The firstcategory of comments was largelyderived from ad hoc panels ofoccupational and other epidemiologicalexperts, consulting groups, andindividual consultants. Most of theseindividuals and groups commented onthe use of epidemiology in reachingscientific and policy conclusionsprimarily from an occupational orhazard assessment perspective, incontrast to the perspective of the reviewof ambient PM criteria and standards,where the use of community airpollution epidemiological studies arecentral. Citing accepted criteria used inevaluating epidemiological studies toassess the likelihood of causality (mostnotably those of Sir Austin BradfordHill, 1965), these commenters arguedthat in the absence of a demonstratedbiological mechanism, the relative risksof effects in the PM epidemiologicalstudies are too low (less than valuesvariously cited as 1.5 to 2.0) to reachany conclusions regarding causality orto form the basis for regulations. Ingeneral, the commenters applied thesecriteria to a subset of studies evaluatedin the Criteria Document, including asfew as two long-term exposure studies(EOP Group) (API, 1997), a group of 9selected studies (Greenland panel) (API,1997), those studies cited in theproposal (AIHC, 1997), or as many as 23selected short-term exposure studiesexamined in a recently publishedreview paper (Gamble and Lewis, 1996).

Based on a careful review of thesecomments, EPA notes a number oflimitations in these commenters’evaluations of the epidemiologicalstudies that they considered, asdiscussed in detail in the Response toComments. In summary, EPA notes thatthese commenters provided scientificadvice and conclusions that are insubstantial disagreement with theconclusions of the review reflected inthe Criteria Document and Staff Paper.EPA stands behind the scientificconclusions reached in these documentsregarding the appropriate use of theavailable community epidemiologicalstudies. These documents were theproduct of an extended public processthat included conducting publicworkshops involving the leadingresearchers in the field, drafts of theCriteria Document and Staff Paperproviding opportunities for publicscrutiny and comment on, and, notleast, receiving the advice of anindependent panel of air pollutionexperts, including epidemiologists.

EPA clearly specified the key criteriaby which it evaluated the availableepidemiological studies in section12.1.2 of the Criteria Document, withsubstantial reliance on those specifiedby Hill (1965). In rejecting results withrelative risks less than 1.5 to 2 asmeaningful absent demonstratedbiological mechanisms, the commentersfail to note that Hill and other expertgroups (U.S. DHEW, 1964) haveemphasized that no one criterion isdefinitive by itself, nor is it necessarythat all be met in order to support adetermination of causality (U.S. EPA,1996a, p. 12-3).

With respect to biological plausibility,Hill noted that ‘‘this is a feature I amconvinced we cannot demand. What isbiologically plausible depends upon thebiological knowledge of the day’’ (Hill,1965). This statement is clearlypertinent to the toxicological andmechanistic understanding of the effectsof PM and associated air pollutants,especially at lower concentrations. It isalso important to stress that while themechanistic evidence published as ofthe time the Criteria Document closeddoes not provide quantitative supportfor the epidemiological results, neithercan such limited evidence refute thesefindings. It is also important to stressthat our understanding of biologicalmechanisms for PM pollution effects isnot sufficient to explain the effectsobserved at much higher concentrationsin air pollution episodes, for whichcausality is generally accepted.Moreover, the toxicological literaturehas only recently begun to examineanimal models (or controlled humanstudies) that might reflect the sensitivepopulations in question (the elderly,individuals with chronic respiratoryand cardiovascular disease) or thatadequately reproduce all of the physico-chemical properties of particles in theambient atmosphere. In short, theabsence of evidence of a particularmechanism is hardly proof that there areno mechanisms that could explain theeffects observed so consistently in theepidemiological studies. The absence ofbiological mechanisms did not deterCASAC from recommending revisionsto the PM standards in 1982, 1986, andagain in 1996.

While Hill appropriately emphasizedthe strength of the association asimportant (e.g., size of the relative risk),he also pointed out that ‘‘We must notbe too ready to dismiss a cause-and-effect hypothesis merely on the groundthat the observed association appears tobe slight. There are many occasions inmedicine when this in truth is so’’ (Hill,1965). EPA believes that the effects ofair pollution containing PM is such a


case. Unlike the ‘‘textbook’’ examples ofunlikely significant associationsprovided by some commenters (e.g., icecream consumption correlated with heatstroke), the abundant epidemiologicalliterature on combustion particlesdocuments numerous occasions inwhich single short-term episodes ofhigh air pollution producedunequivocally elevated relative risks.For the week of the well documented1952 London air pollution episode, forexample, the relative risk of mortalityfor all causes was 2.6, while the relativerisk for bronchitis mortality was as highas 9.3 (Ministry of Health, 1954).Hospital admissions also increased bymore than a factor of two. Britishepidemiologists in the 1950s concludedthat increased mortality was likelywhen PM (as mass calibrated BritishSmoke <4.5 µm in aerodynamicdiameter) exceeded 500 µg/m3 (Martinand Bradley, 1960). This is only abouta factor of 3 higher than that allowed bythe current PM standard. Unlike the‘‘textbook’’ and other unlikely statisticalassociations noted by some commenters,where the only evidence is for lowrelative risk, clear and convincing linksbetween high-level PM concentrationsand mortality and morbidity buttress thefindings of similar associations at muchlower PM concentrations as suggested inthe more recent epidemiologicalliterature.

These commenters also appear toignore several epidemiological studiesconducted at low PM concentrations inU.S. and European cities, including bothshort- and long-term exposures to PMair pollution, that find statisticallysignificant relative risks of respiratorysymptom categories in children in therange of 1.5 to 5 (Schwartz et al., 1994;Pope and Dockery, 1992; Braun-Fahrlander et al., 1992; Dockery et al.,1989; Dockery et al., 1996).Concentrations in these studies extendfrom moderately above to well belowthose permitted by the current PM10

standards. While, as noted in theproposal, most of the recentepidemiological studies of mortality andhospital admissions reportcomparatively small relative risks, thefindings of relative risks well in excessof the 1.5 to 2 criterion noted bycommenters for earlier studies of highPM episodes, as well as the relativerisks of 1.5 to 5 reported in more recentstudies of less serious, but stillimportant effects categories, lendcredibility to EPA’s interpretation of theresults.

In addition to basing theirconclusions primarily on their ownassessment of a limited set of studies,this group of commenters reached

different conclusions about theconsistency of the observed associationsbecause of their assumptions that allmodel building strategies by all authorsare equally valid. Even the mostthorough of these treatments (Gambleand Lewis, 1996) shared this flaw,particularly in the discussion of theseries of Philadelphia mortality studiesand in the discussion of modelingapproaches. The authors’ treatment ofmodeling and confounding issues wasfurther limited because they did notinclude the most recent Philadelphiaresults (Samet et al., 1996a,b) sponsoredby the Health Effects Institute (HEI,1997). One of the important functions ofthe Criteria Document is to evaluate thestrengths and limitations of variousstudies. As discussed more fully belowin this unit, the Criteria Documentfound that some of the studies cited bycommenters as suggesting a lack ofconsistency had important limitations.In general, these commenters’ analysessuffered by ignoring the much morethorough critical review of these studiesand issues contained in the CriteriaDocument, notably that in section 12.6on alternative modeling approaches.

EPA also rejects the notion advancedby these commenters thatepidemiological studies must usepersonal exposure monitoring to beconsidered for regulatory purposes. Inparticular, commenters ignore thesignificant strengths of the time-seriesstudies and prospective cohort studiesrelied on by EPA as compared to cross-sectional epidemiological studies. Time-series studies, such as the dailymortality studies, look at changes inresponse rate in relation to changes inweather and air pollution over timeintervals of a few days. This controls forother factors such as smoking andsocioeconomic status, which are littlechanged during such short intervals.Prospective cohort studies (e.g., Pope etal., 1995; Raizenne et al., 1996), on theother hand, look at changes in healthstatus in a selected cohort ofindividuals, which allows directadjustment for smoking status,socioeconomic status, and other subject-specific factors. The commenters alsoignore the Criteria Documentconclusions on how properly conductedmonitoring can provide an adequateindex of population exposure toambient air pollution in such studiesthat, as detailed below, is more relevantto establishing ambient air qualitystandards (U.S. EPA 1996a, chapter 7).Although personal monitoring may bepractical for some occupational andepidemiological studies, and has beenemployed in some past studies of air

pollution, it is not realistic to requirepersonal monitors in air pollutionstudies of daily mortality, which requireurban scale population data over aperiod of years. Furthermore, the use ofcommunity monitoring-basedepidemiological studies as a basis forestablishing standards and guidelineshas a long history in air pollution,including the British authorities’response to the London episodes andthe establishment of the original U.S.NAAQS in 1971. Rejecting the use of thevast array of such studies on this basisalone would also go against the adviceof the independent scientific experts onevery CASAC panel that has addressedthe subject of PM pollution through theyears, each of which has recommendedgeneral PM standards based primarilyon the results of communityepidemiological studies (Friedlander,1982; Lippmann, 1986; Wolff, 1996b).As noted above in this unit, EPA hasincluded a more detailed discussion ofits responses to these comments in theResponse to Comments.

b. Specific comments onepidemiologic studies. The secondgroup of commenters noted above mademore specific challenges to EPA’sassessment of the epidemiologicalstudies. These comments, althoughoverlapping some of those made by thefirst group, were generally made bycommenters who have taken a moreactive role in the review of the CriteriaDocument and Staff Paper. Thesecommenters asserted that theepidemiological evidence on PM is notas consistent and coherent as EPA hasclaimed, and, in particular, charged thatEPA ignored or downplayed a numberof studies that the commenters arguecontradict the evidence the Agencycited as supporting the consistency andcoherence of PM effects. The studies, allof which commenters contend do abetter job of addressing one or more keyissues, such as confounding pollutants,weather, exposure misclassification, andmodel specification, than earlierstudies, include several that wereavailable during preparation of theCriteria Document, and a number thatappeared after the Criteria Documentand Staff Paper were completed.Because the status of the later studiesdiffer from that of the earlier ones forpurposes of decisions under section 109of the Act, the two categories arediscussed separately below in this unit.Additional responses to commentsrelating to both sets of studies have beenincluded in the Response to Comments.In addition to the inclusion of specificstudies, commenters also raised otherissues regarding the limitations of the


10 The term ‘‘negative’’ studies, as used in thesecomments, should not be construed to mean thosein which there is a negative effects estimate (eithersignificant or non-significant) for the nominalcause. As used by these commenters, the term alsoincludes statistically non-significant positive effectestimates. In other words, the commenters define‘‘positive’’ studies as including only those in whichthe effect estimate is both positive and statisticallysignificant.

11 Data sets were those used in the originalstudies by Dockery et al. (1992) for St. Louis andEastern Tennessee; Pope et al. (1992) for UtahValley; Schwartz and Dockery (1992a) forPhiladelphia; Schwartz (1993) for Birmingham; anda portion of the Santa Clara data from Fairley(1990). The data set from the Moolgavkar et al.(1995a) Philadelphia reanalysis was also included(Samet et al., 1995).

12 The HEI Board of Directors appointed an eightmember Oversight Committee consisting of leadingscientists in several disciplines relevant to airpollution epidemiology to oversee key aspects ofthe project and to prepare HEI’s assessment of theresults.

epidemiological information and the useof these studies in EPA’s two-city riskassessment. Both of these topics are alsodiscussed below in this unit.

(i) Studies available for inclusion inthe criteria review. With someexceptions, most of the abovecommenters cited somewhat similarlists of ‘‘negative’’ studies that theyargue EPA ignored or downplayed inarriving at conclusions on consistencyand coherence. Of the most commonlycited studies, the following wereavailable for inclusion in the CriteriaDocument: daily mortality studies byStyer et al. (1995), Lyon et al. (1995), Liand Roth (1995), Moolgavkar (1995a,b),Wyzga and Lipfert (1995), Lipfert andWyzga (1995), and Samet et al. (1995,1996a,b); the long-term exposuremortality study by Abbey et al. (1991);and the re-examination of the Six-Citymortality results (Dockery et al., 1993)by Lipfert (1995).

The written record of EPA’sevaluations of these studies effectivelyrefutes the claim that the Agencyignored any of these studies andsupports the treatment the Agencyaccorded to each of them. All of thestudies available to EPA at the time ofCASAC closure on the PM CriteriaDocument (March 1996) were examinedfor inclusion in the Criteria Documentand Staff Paper, which form the basisfor the PM proposal. ‘‘Negative’’10

studies were evaluated in detail alongwith ‘‘positive’’ studies when they werefound to have no critical methodologicaldeficiencies, or to point out strengthsand limitations. Studies that had moreserious problems were generallydiscussed in less detail, whetherpositive or negative, than studies withfewer or small deficiencies. The EPAassessments were evaluated by peerreviewers, by CASAC, and by thepublic.

Most of the short-term exposurestudies cited above in this unit arereanalyses and extensions of PM/mortality studies that had beenpublished by other investigators. Ingeneral, the Criteria Documentconcluded that the most comprehensiveand thorough reanalyses were those inthe series conducted for the HEI, whichreanalyzed data sets used in studiesfrom six urban areas in Phase I.A (Samet

et al., 1995)11, with extended analysesfor Philadelphia in Phase I.B (Samet etal., 1996a,b). The most importantfinding in the HEI Phase I.A reanalysesof the six areas is ‘‘the confirmation ofthe numerical results of the earlieranalyses of all six data sets’’ (HEI,1995)12. After replicating the originalinvestigators’ analyses, Samet et al.(1995) also found similar resultsanalyzing the data using an improvedstatistical model. The HEI OversightCommittee found

[I]t is reasonable to conclude that, in thesesix data sets, daily mortality from all causescombined, and from cardiovascular andrespiratory causes in particular, increases aslevels of particulate air pollution indexesincrease. [HEI, 1995]

It is important to note that thesereanalyses by respected independentscientists confirm the reliability andreproducibility of the work of theoriginal investigators, particularly inview of the concerns some commentershave expressed about EPA’s reliance ona number of PM studies published bythese authors.

The Phase I.A HEI results forPhiladelphia also found that it wasdifficult to separate the effects of PMfrom those of co-occurring SO2, inagreement with the Moolgavkar etal.(1995a) analysis. Subsequent HEIwork, and several of the other so-called‘‘negative’’ studies cited above in thisunit, further examined this issue interms of confounding or effectsmodification by one or more co-occurring gaseous pollutants or weather.Contrary to commenters’ claims, thisissue and these studies receivedconsiderable attention in the CriteriaDocument and Staff Paper, and theoverall implications and conclusionsfrom these assessments weresummarized in the proposal. Inparticular, the so-called ‘‘negative’’ andother findings of Moolgalvkar et al.(1995a,b) in their Philadelphia andSteubenville studies were discussed ingreat detail in section 12.6 of the PMCritera Document and compared tothose of the original investigators(Schwartz and Dockery, 1992a,b) and

other investigators (Li and Roth, 1995;Wyzga and Lipfert, 1995). Furtheranalytical studies of the Philadelphiadata set were carried out by HEI (Sametet al., 1996a,b) and have largely resolvedmany of the uncertainties in the earlieranalyses; in EPA’s opinion, thesestudies supersede the results of theoriginal investigators (Schwartz andDockery, 1992a) and the several earlierreanalyses, including Moolgavkar(1995a), Moolgavkar and Luebeck(1996), Li and Roth (1995), Wyzga andLipfert (1995), and Samet et al. (1995).Even though TSP is not the best PMindicator for health effects, since itincludes a substantial fraction of non-thoracic particles, the extended CriteriaDocument assessment (U.S. EPA, 1996a,pp. 12-291 to -299; 12-327) of the PhaseI.B HEI analyses in Philadelphia (Sametet al., 1996a,b) serves to support thefollowing findings:

(1) The mortality effects estimates forTSP do not depend heavily on statisticalmethods when appropriate models areused.

(2) Estimated PM effects are nothighly sensitive to appropriate methodsfor adjusting for time trends and forweather.

(3) Air pollution has significant healtheffects above and beyond those ofweather.

(4) Copollutants such as ozone, CO,and NO2 may be important predictors ofmortality, but their effects can besubstantially separated from those ofTSP and SO2.

(5) The health effects of TSP inPhiladelphia cannot be completelyseparated from SO2, which is itself aprecursor of fine particles, based solelyon the epidemiological analyses in thissingle city.

The most recent HEI OversightCommittee comments on these studies(HEI, 1997), which were submitted tothe docket by HEI, state that:

Although individual air pollutants (TSP,SO2, and ozone) are associated withincreased daily mortality in these data, thelimitations of the Philadelphia data make itimpossible to establish that particulate airpollution alone is responsible for the widelyobserved associations between increasedmortality and air pollution in that city. Allwe can conclude is that it appears to play arole. [HEI, 1997; p.38.]

While recognizing the limitations in theconclusions that can be made based onstudies in a single city, the OversightCommittee endorses the approach takenby EPA in evaluating a broader set ofepidemiological studies:

Consistent and repeated observations inlocales with different air pollution profilescan provide the most convincingepidemiological evidence to support


13 Their March 20, 1996 letter to theAdministrator concludes that the HEI analysis ofPhiladelphia supersedes earlier analyses,specifically Moolgavkar et al. (1995a), Lipfert andWyzga (1995), and Li and Roth (1995), and pointsout the limitations of Styer et al. (1995).

14 In response to comments on this rulemaking,some papers submitted by industry commenters

Continued

generalizing the findings from these models.This has been the approach reported by theEPA in its recent Criteria Document and StaffPaper. [HEI, 1997; p. 38.]

As noted in the proposal, based onthis approach, EPA’s assessment ofnumerous mortality studies concludesthat when studies are evaluated on anindividual basis, the PM-effectsassociations are valid and, in a numberof studies, not seriously confounded byco-pollutants (U.S. EPA, 1996a; p. 13-57); and when a collection of studiesfrom multiple areas with differingconcentrations of PM and co-pollutantsare examined together, the associationwith PM10 remains reasonablyconsistent across a wide range ofconcentrations of these potentiallyinfluential pollutants (U.S. EPA, 1996a;p. 12-33; U.S. EPA, 1996b; p. V-55).

In addition to relying on the mostcomprehensive and best analyses inevaluating the reanalysis inPhiladelphia and other areas, theCriteria Document gave less weight toboth so-called ‘‘negative’’ and‘‘positive’’ studies with methodogicallimitations. In particular, EPA agreedwith the epidemiological experts onCASAC (Lippmann et al., 1996; Samet,1995) that the Li and Roth (1995) studyapproach of using a ‘‘panoply’’ ofdifferent modeling strategies to produceseemingly conflicting findings provideslittle useful insight and is superseded bythe HEI report. The attempt by Lipfertand Wyzga (1995) to address relativeeffects of different pollutants wasconsidered inconclusive (Lippmann etal., 1996) and flawed by the use of ametric (elasticity) that ignores theabsolute concentrations of thepollutants being compared (seeResponse to Comments).

Further, the Steubenville studies andreanalyses (Schwartz and Dockery,1992b; Moolgavkar, 1995b) werediscussed in detail to examinemethodologies, and the differences inrelative risks between the two wereregarded as small (U.S. EPA, 1996a, p.12-280 to 283). Both studies used TSPas the PM indicator variable, and theyare augmented by the more recentfindings of Schwartz et al. (1996) thatexamine PM10 and its components. Themixed results by Lyon et al. (1995) inUtah Valley are compromised by loss ofinformation related to the methodology(U.S. EPA, 1996a, p. 12-58). As notedabove, subsequent reanalyses of theUtah Valley study by HEI (Samet et al.,1995) as well as by Pope and Kalkstein(1996) confirmed the original findings ofPope et al. (1992) using different modelspecifications. The Salt Lake City studyby Styer et al. (1995) was mentioned inthe PM Criteria Document, but received

little discussion because aspects of themethodological approach limited itsstatistical power to detect effects. Theanalysis of Chicago mortality data in thesame paper shared these problems,particularly for seasonal analyses; inthis larger city, they nonetheless foundsignificant associations on an annualbasis between PM10 and mortality thatare consistent with other studies. Inshort, the record shows that EPA did notignore these short-term exposure studiescited by commenters; moreover, EPA’sassessment of these studies is consistentwith the views of four researchers onthe CASAC panel who have extensiveinvolvement in conducting populationstudies of air pollution (Lippmann et al.,1996).13

Similarly, EPA believes thatappropriate treatment and weight weregiven to studies of long-term exposureand mortality. EPA concluded that thelack of associations in the Abbey et al.(1991) prospective cohort study werenot inconsistent with two other suchstudies because the use of days of peakTSP levels as the PM indicator (insteadof PM10 or PM2.5) is inappropriate forCalifornia cohorts exposed to bothurban smog and fugitive dust episodes,and the overall sample size may havebeen too small to detect significanteffects (U.S. EPA, 1996b; pp. V-17 to-18). The inadequacy of Lipfert’s (1995)application of state-wide averagesedentary lifestyle data to adjustmortality for the six cities studied byDockery et al. (1993), in which superiorsubject-specific body mass index datahad already been considered, was alsonoted and addressed in the Staff Paper(U.S. EPA, 1996b; p. V-16). Again, EPAdid not ignore these studies; therationale for giving them less weightwas clearly articulated in the documentsreviewed by CASAC and judgedappropriate for use in standard setting.

While the proposal presents only asummary discussion of key CriteriaDocument and Staff Paper findings, EPAbelieves that discussion is fullyconsistent with the state of the science.Furthermore, the proposal highlights thenature of alternative viewpoints on theepidemiology in a quotation from theCriteria Document (61 FR 65644,December 13, 1996) and cites explicitlythe views of most of the authors notedabove in this unit (Moolgavkar et al.,1995b; Moolgavkar and Luebeck, 1996;Li and Roth, 1995; Samet et al., 1996;Wyzga and Lipfert, 1995). The proposal

also summarizes EPA conclusions basedon all of the literature as assessed in theCriteria Document and Staff Paper withrespect to issues raised in these andother studies, including potentialconfounding by independent risk factorssuch as weather and other pollutants,choice of statistical models, use ofoutdoor monitors, and exposuremisclassification.

More specifically, in the proposalEPA has not ignored the view advancedby some that the results of individualstudies of multiple pollutants, such asthe HEI Philadelphia studies, are moresuggestive of an ‘‘air pollution’’ effectthan an effect of PM alone. Indeed, theproposal notes that it is reasonable toexpect that other pollutants may play arole in modifying the magnitude of theestimated effects of PM on mortality,either through pollutant interactions orindependent effects (61 FR 65645,December 13, 1996). Based on the largebody of evidence at hand, however, EPAcannot accept the suggestion that suchmulti-pollutant studies are in any way‘‘negative’’ with respect to EPA’sconclusions that PM, alone or incombination with other pollutants, isassociated with adverse effects at levelsbelow those allowed by the currentstandards. This conclusion is based notonly on the consistency of PM effectsacross areas with widely varyingconcentrations of potentiallyconfounding copollutants, but also onthe extended analyses of thePhiladelphia studies in the CriteriaDocument and Staff Paper.

Because commenters have tended toignore the latter analyses, it isappropriate to summarize them herebriefly. As noted above in this unit, theCriteria Document assessment of thePhiladelphia studies finds that PM canreasonably be distinguished frompotential effects of all pollutants exceptSO2. The Staff Paper builds on thisanalysis through an integratedassessment that draws on informationfrom atmospheric chemistry, humanexposure studies, and respiratory tractpenetration results to provide insight asto which of these two pollutants is morelikely to be responsible for mortality inthe elderly and individuals withcardiopulmonary disease (U.S. EPA1996b; pp. V-46 to -50). That assessmentnotes that the inhalable (PM10),including the fine (PM2.5), componentsof TSP are more likely than SO2 topenetrate and remain indoors where thesensitive population resides most of thetime.14 In addition, these PM


make statements that are in substantial agreementwith these staff conclusions with respect to thelikelihood of SO2 penetrating to indoorenvironments and the lesser likelihood of affectingsensitive populations indoors (Lipfert and Wyzga,1997; Lipfert and Urch, 1997).

15 Since the 1970 amendments, the EPA has takenthe view that NAAQS decisions are to be based onscientific studies that have been assessed in airquality criteria [see e.g., 36 FR 8186 (April 30, 1971)(EPA based original NAAQS for six pollutants onscientific studies discussed in the air qualitycriteria and limited consideration of comments tothose concerning validity of scientific basis); 38 FR25678, 25679-25680 (September 14, 1973) (EPArevised air quality criteria for sulfur oxides toprovide basis for reevaluation of secondaryNAAQS)]. This longstanding interpretation wasstrengthened by new legislative requirementsenacted in 1977 (section 109(d)(2) of the Act;section 8(c) of the Environmental Research,Development, and Demonstration AuthorizationAct of 1978) for CASAC review of air qualitycriteria and reaffirmed in EPA’s decision not torevise the ozone standards in 1993. 58 FR 13008,13013-13014 (March 9, 1993). Some of thecommenters now criticizing EPA for notconsidering the most recent PM studies stronglysupported the Agency’s interpretation in the 1993decision (UARG, 1992).

16 As discussed in EPA’s 1993 decision not torevise the NAAQS for ozone, new studies may

sometimes be of such significance that it isappropriate to delay a decision on revision ofNAAQS and to supplement the pertinent air qualitycriteria so the new studies can be taken intoaccount. 58 FR at 13014, March 9, 1993. In thepresent case, EPA’s provisional examination ofrecent studies suggests that reopening the airquality criteria review would not be warranted evenif there were time to do so under the court ordergoverning the schedule for this rulemaking.Accordingly, EPA believes that the appropriatecourse of action is to consider the newly publishedstudies during the next periodic review cycle.

17 For example, commenting on the Rothexamination of alternative model specifications, Dr.Stolwijk noted ‘‘If you select out of his [Roth’s]matrix the things that other people have done, hecomes to a different conclusion than when he takeshis whole matrix * * *. [Y]ou are going to get arandom effect that shows that there is no effect. He[Roth] did this, I think, on purpose in this case.Most epidemiologists, I think, have been trained tolimit their observations to something that they canstate or would have stated before they started andobserve that and base their conclusions on it’’ [U.S.EPA 1996(c); May 17, 1996 Transcript, pages 45-46].

components, especially PM2.5, penetratefar more effectively to the airways andgas exchange regions of the lung thandoes SO2. Furthermore, in Philadelphia,it is possible that SO2 is a surrogate forfine particulate acid sulfates. For thesereasons, even though statistical analysesof the Philadelphia data set cannot fullydistinguish between these two highlycorrelated pollutants, EPA believes thatthe weight of the available evidencefrom an integrated assessment morestrongly supports the notion that PM isplaying an important direct role in theobserved mortality effects associationsin Philadelphia. Moreover, as notedabove in this unit, in some otherlocations with significant PM-mortalityassociations, ambient SO2 levels are toolow to confound PM.

(ii) Recent studies available aftercompletion of criteria review. As notedabove in this unit, other studies cited bysome commenters as so-called‘‘negative’’ evidence ignored by EPAwere published or otherwise madeavailable only after completion of thePM Criteria Document. EPA agrees thatit did not rely on these studies, based onits long-standing practice of basingNAAQS decisions on studies andrelated information included in thepertinent air quality criteria andavailable for CASAC review.15 AlthoughEPA has not relied on such studies inthis review and decision process, theAgency nevertheless has conducted aprovisional examination of these andother recent studies to assess theirgeneral consistency with the muchlarger body of literature evaluated in theCriteria Document.16 EPA has placed its

examination of recent studies in therulemaking docket.

Among the most frequently cited newstudies relied on by commenters wereDavis et al. (1996), Moolgavkar et al.(1997), and Roth and Li (1997). Davis etal. (1996) conducted a reanalysis of theBirmingham mortality data setoriginally investigated in Schwartz(1993). At the time of the close of thepublic comment period, the paper basedon this manuscript had not beenaccepted for publication in a peerreviewed journal (Sacks, 1997).Commenters nevertheless highlight theauthors’ claim that ‘‘when humidity isincluded among the meteorologicalvariables (it is excluded in the analysisby Schwartz [1993]), we find that thePM10 effect is not statisticallysignificant.’’ EPA’s review foundimportant factual errors in this study.Contrary to Davis et al., Schwartz didinclude humidity in his 1993 study, andhis finding of a hot-and-humid-dayeffect was reported there. In addition,the PM-related variables used by Daviset al. in their manuscript were not, asthe authors claimed, the same as thosein Schwartz (1993). Davis et al. alsoused a different humidity indicator,specific humidity. Reanalysis by one ofthe co-authors (R. Smith, personalcommunication, February 8, 1997)showed that when Schwartz’s PMmetric was used, the estimated PM10

effect was of about the same magnitude,and statistically significant at the 0.05level, even using the characterization ofhumidity effect proposed by Davis et al.It therefore appears that the Davis et al.PM10 result was, in fact, consistent withthat of Schwartz, and robust against avery different weather modelspecification.

Based on its examination of both thecontent and the publication status ofthis study, EPA believes the heavyreliance and attention given to it aremisguided. In contrast to commenters’assertions, this study does notcontradict EPA’s conclusions withrespect to consistency of theepidemiological evidence andconfounding by weather variables;indeed, the consideration of thecorrected results would actually support

EPA’s conclusions. EPA believes thisexample reinforces the importance ofrelying on peer reviewed studies andalso conducting the kind of criticalexamination of such studies that takesplace in the criteria and standardsreview process.

Several commenters note that Rothand Li (1997) also reexamined theBirmingham mortality data, as well ashospital admissions data from Schwartz(1994), and produced a number ofnegative and inconsistent results thatdepend on temperature effects andchoice of statistical model. Preliminaryfindings from this study were presentedby Roth at the May 1996 CASACmeeting. CASAC epidemiologists andstatisticians at the meeting pointed outa number of shortcomings, both in theanalytical strategy and in details of themodels being evaluated.17 As discussedin more detail in the Response toComments, the materials from Roth andLi (1997) recently provided to EPA asattachments to public comments showthat the deficiencies pointed out at theMay 1996 CASAC meeting have notbeen adequately addressed. EPAconcludes that this study does notsupport commenters’ claims.

The paper recently accepted forpublication by Moolgavkar et al. (1997)examines hospital admissions and airpollution in Minneapolis andBirmingham and comes to differentconclusions than earlier investigatorswith respect to the role of PM10. Whilethe paper is a useful addition to theliterature, the authors clearly do notattempt to replicate the original studies,making the kind of direct comparisonssuggested by commenters difficult. Thepaper finds an air pollution effect in onecity that implicates ozone but is unableto separate effects of PM from a groupof other pollutants. EPA’s provisionalexamination of this study raises somequestions about the methodology, whichmight usefully be supplemented tofurther separate pollutants as was doneby Samet et al. (1996a,b) inPhiladelphia, and about the authors’interpretation of the results in bothcities. In any event, EPA does notbelieve this study negates the PMassociations with hospital admissions


18 CASAC panelists recommended a discussion ofthis issue in the Staff Paper. The Staff Paper notes:‘‘While greater measurement error for the coarsefraction could depress a potential coarse particleeffect, this would not explain the results in Topekarelative to other cities. Even considering relativemeasurement error, these results provide no clearevidence implicating coarse particles in thereported effects.’’ (U.S. EPA, 1996b p. V-64). EPA’sprovisional examination of the Lipfert and Wyzga(1997) paper in the Response to Comments, findsthat it is implausible that most of the effectattributed to PM2.5 could in fact be due to PM10-2.5,since differential measurement error cannot make aweaker effect appear stronger than a stronger one,except under extremely unusual circumstances.

19 The APHEA (Air Pollution and Health: aEuropean Approach) project was supported by theEuropean Union Environment 1991-1994Programme to investigate the possible short-termhealth effects of exposure to low or moderate levelsof ambient air pollutants. Eleven European researchgroups carried out studies in 15 cities (Amsterdam,Athens, Barcelona, Bratislava, Cracow, Helsinki,Koln, Lodz, London, Lyon, Milan, Paris, Poznan,Rotterdam and Wroclaw) in which air pollutantconcentration data had been collected for at least5 years. Initial findings of studies on mortality andhospital admissions were published in a series ofpapers in Supplement 1 to the Journal ofEpidemiology and Community Health in 1996 anda meta-analysis of the mortality data from 12 citiesis currently in press (Katsouyanni et al., 1997).

20 The Roth et al. (1997) study in Prague used ameasurement termed ‘‘suspended particles’’ thatappears to be close to TSP. The relation of thisindicator to PM10 or PM2.5 in this city is notreported. Moreover, this study uses a variant of theproblematic methodology in the Roth analyses citedabove.

21 These concerns are consistent with EPA’streatment of a number of European and SouthAmerican studies that are included in the CriteriaDocument and contributed to the evaluation of theepidemiology in Chapter 12. Because of differencesin aerometry methods and characteristic sourceclasses between North America and other regions ofthe world, however, the integrative assessmentchapter reported results only from studiesconducted in the U.S. and Canada (cf. Tables 13-3 to 13-5) in reaching quantitative conclusions foreffects estimates.

22 See, for example, the United Kingdom AirQuality Strategy, 1997; Swiss Federal Commissionof Air Hygiene, 1996; World Health OrganizationRevised Air Quality Guidelines for Europe, InPress).

reported in a number of other studiescited in the Criteria Document.

Another recent paper by Lipfert andWyzga (1997) provides analysessuggesting that differential measurementerror might account for some or all ofthe observation by Schwartz et al. (1996)that daily mortality is more stronglyassociated with fine (PM2.5) than withcoarse (PM10-2.5) PM. EPA staff andCASAC accounted for this possibility,however, and it was factored into boththe Staff Paper and CASACrecommendations.18

Some commenters have highlightedselected individual papers or summariesfrom the APHEA19 project conducted inEurope, and from Roth (1996), callingattention particularly to negative resultsfound in heavily polluted regions ofEastern Europe. EPA notes that anumber of the recent APHEA and otherstudies in Western Europe have shownsignificant associations betweenmortality and air pollution includingPM, and that a meta-analysis of 12Western and Central-eastern Europeanstudies ‘‘is supportive of a causalassociation between PM and SO2

exposure and all-cause mortality’’(Katsouyanni et al., 1997). The Easternand Western European studies useddiffering measurement methods for PM,including PM10, gravimetric ‘‘suspendedparticles,’’ and the British Smokemethod.20 The differences in aerometry

and the substantial differences inlocation and strength of primary PMemissions sources in central and easternEurope as compared to western Europeor the U.S. might well explain thedifferent results in these unique areas.Consequently, integration of theseresults would involve comprehensiveexamination of the various PMinstruments used, monitor siting inrelation to sources, mass calibrationprocedures and other aspects of thesestudies.21 EPA notes that a number ofEuropean authorities, who are familiarwith this recent literature, haveproceeded with recommendations tostrengthen their health guidelines, riskassessments, or regulations for PM.22

Aside from the recent literature citedby these commenters, there are anumber of other recent epidemiologicalstudies that, if considered in today’sdecision, would tend to support EPA’sconclusions about the effects of PM atlower concentrations, assuming theirresults were accepted following a fullreview in the criteria and CASACprocess. For example, in addition to theAPHEA studies, several other recentepidemiologic studies have reportedsignificant positive associationsbetween PM and health effects (Lipsettet al., 1997; Peters et al., 1997; Borja-Aburto et al., 1997; Delfino et al., 1997;Scarlett et al., 1996; Woodruff et al.,1997; Wordley et al., 1977). In addition,a number of recent toxicologic papershave been accepted or appear inproceedings (Costa and Dreher, 1997;Killingsworth et al., 1997; Godleski etal., 1997) that involve exposure toconcentrated ambient fine particles orPM constituents and appear to providesupportive evidence as to theplausibility of the effects that have beenreported epidemiologically. Ifconsidered in this decision, thesestudies would also provide biologicalsupport for the epidemiologicalobservation that certain susceptiblegroups (notably those withcardiopulmonary disease) are mostlikely to be affected by PM, againassuming the results were sustained in

the full criteria and CASAC reviewprocess.

In summary, EPA has conducted aprovisional assessment of the morerecent scientific literature. Based on thisprovisional assessment, EPA disagreeswith commenters’ assertion that fullconsideration of selected new studies inthis decision would materially changethe Criteria Document and Staff Paperconclusions on the consistency andcoherence of the PM data, or on theneed to revise the current standards.

(iii) Other specific comments on theepidemiological studies. Aside fromtheir assertion that EPA ignored ordownplayed particular studies, thissecond group of commenters raiseadditional objections, based on thestatistical modeling strategies used andthe potential importance of personalexposure misclassification, to EPA’sconclusions regarding the consistency ofthe epidemiological evidence. EPAconclusions on these topics weresummarized in the proposal andsupported by extensive treatments inthe Criteria Document and Staff Paper.With respect to the first issue,commenters argued that sufficientflexibility exists in the analyses of largedata sets that it may be possible toobtain almost any result desired throughchoice of statistical method. Analyticalchoices include the specific statisticalmodel; methods used to adjust forseasonal variation and the trends in thedata; treatment of other variables (e.g.,other pollutants, weather, and day ofweek); ‘‘lag’’ structure; and studypopulation.

A more detailed discussion of thisissue, which expands on the assessmentsummarized in the Criteria Document, isincluded in the Response to Comments.In summary, EPA must rejectcommenters’ contention that legitimatealternative analyses can obtain ‘‘almostany result.’’ As outlined above in thisunit, EPA’s detailed reviews ofindividual studies have shown that notall methods are equally valid orlegitimate. Moreover, strong argumentscan be made that the methods andanalytical strategies in the studies EPArelied upon are more appropriateapproaches than those cited bycommenters (e.g., Li and Roth, 1995;Lipfert and Wyzga, 1995; Davis et al.,1996; Roth and Li, 1997). While not allstudies have addressed each of theabove issues in this unit equally well,the most comprehensive analyses ofthese issues (e.g., Samet et al., 1995,1996a,b; Pope and Kalkstein, 1996), aswell as the EPA analyses comparingstudy results for each issue (U.S. EPA,1996a, pp. 12-261 to 12-305) found thatthe authors of studies on which EPA


23 Paradoxically, some commenters have argued(e.g., Valdberg, 1997) that the PM results areconfounded because the weather and other factorsthat cause daily variations in outdoor pollution willcause similar daily variations in indoor generatedair pollution. For this to be true, outdoor ambientpollution concentrations would have to becorrelated with personal exposure to indoorgenerated air pollution such as that from smoking,cleaning, and cooking. This argument is logicallyinconsistent with the other comments on the lackof any such correlation with personal exposure, andthese commenters have offered no scientificevidence to support their claim. In response, EPAhas performed and included in the Response toComments a numerical analysis of the relevantinformation from the PTEAM exposure study thatfinds no evidence for such a correspondence in theactual data.

24 As documented in Chapter 7 of the CriteriaDocument, time-series community studies observethe effects of varying levels of ambient air pollution;therefore the effects of indoor-generated airpollution would be independent of and in additionto the effects found in these epidemiologicalstudies. Commenters apparently believe EPA isclaiming such studies are detecting the effects ofdaily variations in total PM personal exposure fromindoor and outdoor sources. This misunderstandingis evidenced, for example, by Wyzga and Lipfert’s(1995) treatment of the difference between ambientmonitors and actual personal exposures as‘‘exposure errors’’ and Brown’s comment for APIthat ‘‘if (ambient) PM is causally related tomortality/morbidity, then it is personal PMexposure that must be reduced to have an effect.’’On the contrary, it is personal exposure to ambientPM that must be reduced to address the risk

identified in community air pollution studies. Anylack of significant correlation between outdoor PMconcentrations and personal exposure to total PMfrom all sources is irrelevant, except to the extentit may decrease the power of time-series studies todetect the effects of ambient pollution.

25 The EPA analysis finds that in order formeasurement errors in one pollutant variable tosignificantly bias the estimated effect of anotherpollutant, three conditions are necessary: (1) Themeasurement error in the poorly measuredpollutant must be very large, roughly at least thesame size as the population variability in thatpollutant; (2) the poorly measured pollutant mustbe highly correlated with the other pollutant, eitherpositively or negatively; and (3) the measurementerrors for the two pollutants must be highlynegatively correlated (Response to Comments,Appendix D). This important factor was notconsidered in Lipfert and Wyzga (1995) or bycommenters.

chiefly relied made appropriatemodeling choices. The CriteriaDocument concludes that: ‘‘[T]he largelyconsistent specific results, indicative ofsignificant positive associations ofambient PM exposures and humanmortality/morbidity effects, are notmodel specific, nor are they artifactualyderived due to misspecification of anyspecific model. The robustness of theresults of different modeling strategiesand approaches increases ourconfidence in their validity [U.S. EPA,1996a, p. 13-54].’’ While it is true, asevidenced in Li and Roth (1995), thatPM-effects data can be randomlymanipulated to produce apparentlyconflicting results, commenters haveprovided no evidence that differentplausible model specifications couldlead to markedly different conclusions.

Some commenters have expressedconcerns about the reliability of theepidemiological results because somestudies showed a lack of correlation incross-sectional comparisons betweenoutdoor PM measured at centrallocations and indoor or personalexposures to PM (which includes PMfrom the outdoor, indoor and personalenvironments).23 EPA acknowledgedand responded to this issue in chapter7 of the Criteria Document and theproposal (61 FR 65645, December 13,1996). The major premise underlyingcommenters’ arguments on this issue isincorrect.24 The question is not whether

central monitoring site measurementscontain a signal reflecting actualexposures to total PM from both outdoorand indoor sources at the individuallevel; the relevant question is whethercentral monitoring site measurementscontain a signal reflecting actualexposures to ambient PM for the subjectpopulation, including both ambient PM,while individuals are outdoors, andambient PM that has infiltrated indoors,while individuals are indoors. The PMstandards are intended to protect thepublic from exposure to ambient PM,not PM generated by indoor or personalsources. There is ample evidence, asdiscussed in chapter 7 of the CriteriaDocument, that personal exposure toambient PM, while outdoors and whilein indoor micro-environments, doescorrelate on a day-to-day basis withconcentrations measured at properlysited central monitors (U.S. EPA, 1996a,p. 1-10). EPA has, therefore, concludedthat it is reasonable to presume that areduction in ambient PM concentrationswill reduce personal exposure toambient PM, and that this will protectthe public from adverse healthoutcomes associated with personalexposure to ambient PM.

Commenters have also restatedtheoretically based concerns on arelated issue, namely errors in themeasurement of the concentrations ofair pollutants, that was summarized inthe proposal. In multiple pollutantanalyses, measurement error or, moregenerally, exposure misclassification,could theoretically bias effects estimatesof PM or co-pollutants in eitherdirection, introducing furtheruncertainties in the estimatedconcentration-response relationships forall pollutants (U.S. EPA, 1996b, pp. V-39 to V-43). Relevant insights on thisissue in material appended to publiccomments (Ozkaynak and Spengler,1996) have prompted an expandedstatistical analysis of the conditionsunder which such errors could inflatethe magnitude of the effects estimates orthe significance of PM relative togaseous pollutants, as has beensuggested by Lipfert and Wyzga (1995).This analysis, which is summarized inthe Response to Comments, finds thatthe conditions under whichmeasurement error could inflate theeffects estimates or significance of PMrelative to other pollutants are restrictedto a limited set of statisticalrelationships. Commenters have not

provided evidence that suggest suchconditions are likely to occur withrespect to the measurement of ambientPM in relation to those for gaseous co-pollutants commonly used inepidemiological studies.25 Therefore, itappears unlikely that measurement andexposure errors for PM and otherpollutants have inflated the estimatedeffects of PM, even in multivariateanalyses. More importantly, theavailable evidence on the consistency ofthe PM-effects relationships in multipleurban locations, with widely varyingindoor/outdoor conditions and a varietyof monitoring approaches, makes it lesslikely that the observed associations ofPM with serious health effects at levelsallowed under the current NAAQS arean artifact of errors in measurement ofpollution or of exposure (U.S. EPA1996b, pp. V-39 to V-43).

(iv) Comments on the PM riskassessment. As noted in the proposal,uncertainties about measurement errors,exposure misclassification, and therelative effects of copollutants are moreimportant to the quantitative estimatesof risk associated with PM than to theexistence of valid PM-effectsassociations at levels found in recentstudies. A number of commentersargued that EPA’s risk assessment isflawed and incomplete. Chief among thereasons they advanced is that theassessment is based on the sameepidemiological studies thesecommenters argued are inadequate forthe reasons summarized and respondedto above. Specific comments alsoaddressed the extent to which the riskassessment might overstate riskestimates because it assumes a linearno-threshold relationship and the use ofstudies that might inflate PM risk due toinadequate consideration of co-pollutants and other potentialconfounders. The full risk assessmentacknowledges these issues anduncertainties, however, and it illustratesthe potential influence of suchuncertainties in sensitivity analyses(U.S. EPA 1996b; chapter 6, appendix F;Abt Associates, 1996a,b; 1997a,b). Forexample, Figure 2c in the proposal (61FR 65653, December 13, 1996)


illustrates the potential influence ofwhat appears to be the most significantuncertainty in current information,whether a population threshold existsbelow which the effects of PM no longeroccur (61 FR 65653, December 13,1996). EPA notes that a fullconsideration of the uncertainties,including the analysis summarizedabove on measurement error, suggeststhat the epidemiological studies mightwell have understated the total effects ofair pollution; thus, both the directionand the extent of any bias in the riskestimates are less clear than commenterssuggest.

EPA believes that, even recognizingthe large uncertainties, the keyqualitative insights derived from therisk assessment and summarized in UnitII.A.3. of this preamble remainappropriate. While not placing greatweight on the specific numericalestimates, EPA believes that the riskanalysis confirms the generalconclusions drawn primarily from theepidemiological results themselves, thatthere is ample reason to be concernedthat exposure to ambient PM at levelsallowed under the current air qualitystandards presents a serious publichealth problem.

3. Key considerations informing thedecision. Having carefully consideredthe public comments on the abovematters, EPA believes the fundamentalscientific conclusions on the effects ofPM reached in the Criteria Documentand Staff Paper, and restated in theintroduction to this unit, remain valid.That is, the epidemiological evidencefor ambient PM, alone or in combinationwith other pollutants, showsassociations with premature mortality,hospital admissions, respiratorysymptoms, and lung functiondecrements. Despite extensive criticalexamination in the criteria andstandards review, these findings cannotbe otherwise explained by analytical,data, or other problems inherent in theconduct of such studies. Although theevidence from toxicological studiesavailable during the criteria review hasnot revealed demonstrated mechanismsthat explain the range of effects reportedin epidemiological studies, it does notand cannot refute the observation ofsuch effects in exposed populations.Moreover, the effects observed in therecent epidemiological studies at lowerPM concentrations are both coherentwith each other and plausible based onthe categories of effects observed atmuch higher concentrations in historicair pollution episodes, laboratorystudies of PM effects at high doses, andparticle dosimetry studies. Theconsistency of the results from a large

number of locations and the coherentnature of the observed results suggest alikely causal role of ambient PM incontributing to the reported effects (U.S.EPA, 1996a; p. 13-1). Many of thestudies showing PM effects wereconducted in areas where the currentPM10 standards are largely met, andboth the studies and EPA’s riskassessment suggest that the collectivemagnitude of the effects reflects asignificant public health problem.

For these reasons, and havingconsidered public comments on thisissue, the Administrator concludes thatthe review of the criteria and standardsprovides strong evidence that thecurrent PM10 standards do notadequately protect public health, andthat revision of the standards is not onlyappropriate, but necessary.

Aside from that conclusion, theappropriateness of continuing to rely onthe use of PM10 as the sole indicator forrevised PM standards is also relevanthere. While the basis for decisions onspecific indicators is discussed morefully in Unit II.C. of this preamble, thisissue is related to the Administrator’sdecision on the need to revise thestandards. Based on both the staffreview (U.S. EPA, 1996b, p. VII-3) andthe recommendations of somecommenters (e.g., California EPA), thereare two alternative approaches forproviding additional health protectionin revising the standards: Adopt tighterPM10 standards and/or recognize thefundamental differences between fineand coarse particles and developseparate standards for the majorcomponents of PM10, including fineparticles. Conceptually, the firstapproach would give weight tocomments that standards should bebased on pollutant indicators for whichthe most data have been collected, withless consideration of the evidence thatsuggests that the current standardsprovide adequate protection against theeffects of coarse particles, and thattightening the current PM10 standards inan attempt to control fine particleswould place unnecessary requirementson coarse particles. Because the PM10

network is in place, a more stringentPM10 standard would also respond tocommenters who have expressed adesire for more immediateimplementation of revised standards.The second approach is based on theview that, in the long run, moreeffective and efficient protection can beprovided by separately targetingappropriate levels of controls to fine andcoarse PM.

The Staff Paper examined this issue indetail (U.S. EPA 1996b, pp. VII-3 to VII-11), and concluded that the available

information was sufficient to developseparate indicators for fine and coarsefractions of PM10, based on the recenthealth evidence, the fundamentaldifferences between fine- and coarse-fraction particles, and implementationexperience with PM10. Further, the staffconcluded that:

[C]onsideration of comparisons betweenfine and coarse fractions suggests that finefraction particles are a better surrogate forthose particle components linked to mortalityand morbidity effects at levels below thecurrent standards. In contrast, coarse fractionparticles are more likely linked with certaineffects at levels above those allowed by thecurrent PM10 standards. In examiningalternative approaches to increasing theprotection afforded by PM10 standards, thestaff concludes that reducing the levels of thecurrent PM10 standards would not providethe most effective and efficient protectionfrom these health effects. [U.S. EPA 1996b; p.7-45]

As discussed in Unit II.C. of thispreamble, the Administrator believesthat it is more appropriate to provideadditional protection against the riskposed by PM by adding new standardsfor the fine fraction of PM10, as opposedto tightening the current PM10

standards. Although fewerepidemiological studies have used PM2.5

and other fine particle indicators (e.g.,sulfates, acids), there are nonethelesssignificant indications from thescientific evidence - drawn from thephysicochemical studies of PM, airquality and exposure information,toxicological studies, and respiratorytract deposition data - that this approachwill provide the most effective andefficient protection of public health.

Several commenters have argued thatthe decision on whether to revise thePM standards should be deferred,particularly with regard to fine particlestandards, pending establishment andoperation of a national monitoringnetwork to characterize fine PM and aresearch program to reduceuncertainties in the effects information.These commenters expressed concernsthat establishing fine PM standards nowmight result in needless regulation ofPM components that may be unrelatedto observed health effects. As discussedmore fully in Unit II.F. of this preamble,such commenters recommended, atmost, that if fine PM standards wereestablished, they should be set at a level‘‘equivalent’’ to the current PMstandards.

EPA strongly disagrees that thedecision on revising the standardsshould be delayed to await the resultsof new PM monitoring and researchprograms. Under section 109(d) of theAct, EPA’s obligation after reviewing the


existing criteria and standards for PM isto make such revisions in the standardsand to promulgate such new standardsas are appropriate under section 109(b)of the Act. Based on her review of thecriteria and standards for PM, theAdministrator has concluded that thecurrent standards are not adequate toprotect public health and that revisionsare appropriate. In the face of theavailable evidence, a delay in revisingthe standards would not only beinconsistent with the statute but -- evenunder the optimistic assumption thatthe same extensive monitoring andstrategy assessment as nowcontemplated would occur in theabsence of a revised standard -- wouldadd approximately 2 years to the timewhen significant health benefits can berealized, resulting in potentiallysignificant numbers of additionalpremature deaths and even largernumbers of children and individualswith air pollution-related illness andsymptoms. On the other hand,establishing standards now will set intomotion the development ofimplementation programs andmonitoring that can be conducted inparallel with additional scientificresearch, without undue delays inherentin waiting for the research.

The question of which pollutantcomponents to regulate has been anissue since the inception of the first PMstandards. Other ambient pollutants(e.g., NO2 or CO) are uniquely definedas individual chemicals, whether or notthey serve as proxies for a larger classof substances (e.g., ozone as an index ofphotochemical oxidants). Regulatinggeneral PM, as opposed to multiplechemical components of PM, raises thespectre of a host of particulate materialsof varying composition, size, and otherphysicochemical properties, not all ofwhich are likely to produce identicaleffects.

Both EPA’s past and presentregulatory experience with PM controlprograms and its successive reviews ofthe standards have reaffirmed thewisdom of retaining standards thatcontrol particles as a group, rather thaneliminating such standards and waitingfor scientific research to developinformation needed to identify moreprecise limits for the literally thousandsof particle components. Each suchdecision recognized the possibility thatpotentially less harmful particles mightbe included in the mix that wasregulated, but concluded that the needto provide protection against serioushealth effects nonetheless requiredaction under section 109 of the Act. Thesuccess of this approach is evident inearly U.S. control programs that

dramatically reduced ‘‘smoke’’ and‘‘TSP’’ in major cities in the 1960’s and1970’s and in the continuedimprovement in air quality through thecurrent PM standards. The majorrefinements that have beenrecommended through the course ofreviews of PM standards have been toimprove the focus of control efforts bydefining scientifically based size classes(i.e., moving from TSP to PM10 and now,PM2.5) that will permit more effectiveand efficient regulation of thosefractions most likely to presentsignificant risks to health and theenvironment.

As discussed in Unit II.C. of thispreamble, the current review hasexamined the available evidence todetermine whether it would tend tosupport inclusion or exclusion of anyphysical or chemical classes of PM, forexample sulfates, nitrates, or ultra-fineparticles. That examination concludesthat, while both fine and coarseparticles can produce health effects, thefine fraction appears to contain more ofthe reactive substances potentiallylinked to the kinds of effects observedin the recent epidemiological studies(U.S. EPA 1996b, section V.F.).However, the available scientificinformation does not rule out any oneof these components as contributing tofine particle effects. Indeed, it isreasonable to anticipate that no singlecomponent will prove to be responsiblefor all of the effects of PM.

EPA recognizes that whether thestandards are set for PM10 only or alsofor fine particles, there are uncertaintieswith respect to the relative riskpresented by various components ofPM. In this regard, the Administratorplaces greater weight on the concernthat by failing to act now, the PMNAAQS would not control adequatelythose components of air pollution thatare most responsible for serious effects,than on the possibility they might alsocontrol some component that is not.EPA believes that movingsimultaneously to establish standardsbased on the best available scientificevidence and to conduct an aggressivemonitoring and scientific researchprogram designed to help resolvecurrent uncertainties is a prudent andresponsible approach for addressingboth the risks and the uncertaintiesinherent in this important public healthissue.

In summary, given the evidence thatPM-related health effects appear likelyto occur at levels below the currentstandards, the serious nature andpotential magnitude of the public healthrisks involved, and the need to considerthe fine and coarse fractions as distinct

classes of particles, the Staff Paper andthe CASAC (Wolff, 1996b) concludedthat revision of the current standards isclearly appropriate. Moreover, at theirMay 1996 public meeting (U.S. EPA,1996c), and in separate writtencomments (including Lippmann et al.,1996), a majority of CASAC panelmembers recommended revisions thatwould strengthen the health protectionprovided by the current PM standards.Based on the rationale andrecommendations contained in the StaffPaper and the advice of CASAC, andtaking into account public comments,the Administrator concludes that it isappropriate at this time to revise thecurrent PM standards to increase thepublic health protection providedagainst the known and potential effectsof PM identified in the air qualitycriteria.

C. Indicators of PM

In establishing adequately protective,effective, and efficient PM standards, itis necessary to specify the fraction ofparticles found in the ambient air thatshould be used as the indicator(s) forthe standards. In this regard, EPAconcludes that the most recentassessment of scientific information inthe Criteria Document, summarized inchapters IV and V of the Staff Paper,continues to support past staff andCASAC recommendations regarding theselection of size-specific indicators forPM standards. More specifically, EPAcontinues to find that the followingconclusions reached in the Staff Paperand in the 1987 review remain valid:

(1) Health risks posed by inhaledparticles are influenced both by thepenetration and deposition of particlesin the various regions of the respiratorytract and by the biological responses tothese deposited materials.

(2) The risks of adverse health effectsassociated with deposition of ambientfine and coarse fraction particles in thethoracic (tracheobronchial and alveolar)regions of the respiratory tract aremarkedly greater than for deposition inthe extrathoracic (head) region.Maximum particle penetration to thethoracic region occurs during oronasalor mouth breathing.

(3) The risks of adverse health effectsfrom extrathoracic deposition of generalambient PM are sufficiently low thatparticles which deposit only in thatregion can safely be excluded from thestandard indicator.

(4) The size-specific indicator(s)should represent those particles capableof penetrating to the thoracic region,including both the tracheobronchial andalveolar regions.


26 As discussed above, a number of commentersexpressed concerns that various portions of fineparticles might not be responsible for any observedeffects. One group (PG&E, 1997) recommended thatnitrates should be excluded from fine PM masscollected on the basis of their assessment ofavailable effects literature on particulate and gasphase inorganic nitrates. Based on an examinationof this information as well as the earlier staffassessment, EPA maintains its conclusion that theavailable evidence is not sufficient to excludenitrates or any other class of fine particles that arecollected by PM monitors comparable to those usedin the recent epidemiological studies.

These conclusions, together withinformation on the dosimetry ofparticles in humans, were the basis forthe promulgation in 1987 of a new size-specific indicator for the PM NAAQS,PM10, that includes particles with anaerodynamic diameter smaller than orequal to a nominal 10 µm. The recentinformation on human particledosimetry contained in the CriteriaDocument provides no basis forchanging 10 µm as the appropriate cutpoint for particles capable of penetratingto the thoracic regions.

As noted in Unit II.B. of thispreamble, however, the Staff Paperconcludes that continued use of PM10 asthe sole indicator for the PM standardswould not provide the most effectiveand efficient protection from the healtheffects of PM (U.S. EPA, 1996b, pp. VII-4 to VII-11). Based on the recent healtheffects evidence and the fundamentalphysical and chemical differencesbetween fine and coarse fractionparticles, the Criteria Document andStaff Paper conclude that fine andcoarse fractions of PM10 should beconsidered separately (U.S. EPA, 1996a,p. 13-93; 1996b, p. VII-18). Taking intoaccount such information, CASACfound sufficient scientific and technicalbases to support establishment ofseparate standards relating to these twofractions of PM10. Specifically, CASACadvised the Administrator that ‘‘there isa consensus that retaining an annualPM10 NAAQS * * * is reasonable at thistime’’ and that there is ‘‘also aconsensus that a new PM2.5 NAAQS beestablished’’ (Wolff, 1996b).

Some commenters have noted that itis often difficult to distinguish theeffects of either fine or coarse fractionparticles from those of PM10; this is tobe expected because both fractions arethemselves components of PM10, andhence not fully independent. EPAbelieves that it is more meaningful toexamine comparisons between the fineand coarse fraction components. Suchcomparisons presented in the StaffPaper suggest that fine particles are abetter surrogate for those components ofPM that are linked to mortality andmorbidity effects at levels below thecurrent standards (U.S. EPA, 1996b, p.VII-18). Moreover, a regulatory focus onfine particles would likely also result incontrols on gaseous precursors of fineparticles (e.g., SOx, NOx, VOC), whichare all components of the complexmixture of air pollution that has mostgenerally been associated with mortalityand morbidity effects. The Staff Paperconcludes that, in contrast to fineparticles, coarse fraction particles aremore clearly linked with certainmorbidity effects at levels above those

allowed by the current 24-hourstandard.

Public comments received on theproposed indicators wereoverwhelmingly in favor of EPA’sproposal to maintain PM10 as anindicator for PM, whether as anindicator of coarse particles inconjunction with a fine PM standard, oras the sole PM indicator. This nearunanimity shows strong support forretaining general PM standards. While asubstantial number of commenterssupported EPA’s proposal to add anindicator for fine PM, a number of othercommenters objected to any standardrevisions, including addition of a finePM indicator. Beyond the general pointsabout the basis for any revisionsdiscussed in Unit II.B. of this preamble,these commenters argued either that theavailable epidemiological data did notprovide a basis for separating fine andcoarse fraction particles, or that therewere not enough fine particle studies tosupport selecting standard levels. Mostof these commenters also expressedconcerns that there were insufficientambient fine particle data by which toevaluate the relative protection affordedby new standards.

EPA notes that issues relating to thebasis for separating PM10 fractions wereaddressed in the Criteria Document and/or Staff Paper assessments, and theseperspectives were also available forCASAC consideration in developing itsrecommendations. The proposal statesthat the main basis for separating thefine and coarse fractions of PM10 is that,because they are fundamentallydifferent PM components withsignificantly different physico-chemicalproperties and origins (U.S. EPA 1996b,section V.D), separate standards wouldpermit more effective and efficientregulation of PM. While the difficulty inseparating these classes in theepidemiological studies is noted above,the preponderance of the availableevidence suggests that strategies tocontrol fine particles will moreeffectively reduce population exposureto substances associated with healtheffects in the recent epidemiologicalstudies. Although the number of studiesusing fine PM indicators is more limitedthan for PM10, there are more than 20community studies showing significantassociations for a consistent set ofmortality and morbidity effects. Asubstantial subset of these studies(Tables V-12 to V-13; U.S. EPA, 1996b)provides a sufficient quantitative basisfor selecting standard levels, withoutthe need to rely on estimates based onPM2.5/PM10 ratios.

Having considered the publiccomments on this issue, the

Administrator concurs with staff andCASAC recommendations to controlparticles of health concern (i.e., PM10)through separate standards for fine andcoarse fraction particles. The followingunits outline the basis for theAdministrator’s decision on specificindicators for fine and coarse fractionparticle standards.

1. Indicators for the fine fraction ofPM10. The Administrator continues toconclude that it is appropriate to controlfine particles as a group, as opposed tosingling out particular components orclasses of fine particles. The morequalitative scientific literature,evaluated in Chapter 11 of the CriteriaDocument and summarized in sectionV.C of the Staff Paper, has reportedvarious health effects associated withhigh concentrations of a number of fineparticle components (e.g., sulfates,nitrates, organics, transition metals),alone or in some cases in combinationwith gases. Community epidemiolgicalstudies have found significantassociations between fine particles orPM10 and health effects in various areasacross the U.S. where such fine particlecomponents correlate significantly withparticle mass. As noted above in thisunit, it is not possible to rule out anyone of these components as contributingto fine particle effects.26 Thus, theAdministrator finds that the presentdata more readily support a standardbased on the total mass of fine particles.EPA will conduct additional research,in cooperation with other Federalagencies and in partnership with Stateand local agencies and the privatesector, to better identify which speciesare of concern for human health, andthe sources and relative magnitude ofsuch species.

In specifying a precise size range fora fine particle standard, both the staffand CASAC recommended PM2.5 as theindicator of fine particles (Wolff,1996b). The particle diameter reflectingthe mass minimum between the fineand coarse modes typically lies between1 and 3 µm, and the scientific datasupport a sampling ‘‘cut point’’ todelineate fine particles somewhere inthis range. Because of the potential


27 The National Mining Association (NMA) andrelated companies submitted comments favoringultimate selection of a smaller cutpoint of 1 µm(PM1) to further reduce coarse particle intrusion.EPA considered this approach in developing theStaff Paper and proposal. PM1 has not been used inhealth studies, although in most cases collectedmass should be similar to those for cutpoints of 2.1or 2.5 µm. While a PM1 indicator could reduceintrusion of coarse particles, it might also omitportions of hygroscopic PM components such asacid sulfates, nitrates, and some organic compoundsin higher humidity environments picked up byPM2.5 measurements. PM1 sampling technologieshave been developed, but have not been widelyused in the field to date; there are some concernsabout loss of certain organic materials in availablemodels relative to an instrument with a larger sizecut. NMA has also recommended consideration ofa methodology that could subtract coarse mass fromPM2.5 measurements where undue coarse particleintrusion resulted in fine standard violations. EPAwill evaluate this recommendation in the context ofimplementation policies.

overlap of fine and coarse particle massin this intermodal region, EPArecognizes that any specific samplingcut point would result in only anapproximation of the actual fine-modeparticle mass. Thus, the choice of aspecific diameter within this size rangeis largely a policy judgment. The staffand CASAC recommendations for a 2.5µm sampling cut point were based onconsiderations of consistency with thecommunity health studies, the limitedpotential for intrusion of coarse fractionparticles into the fine fraction, andavailability of monitoring technology.27

PM2.5 encompasses all of the potentialagents of concern in the fine fraction,including most sulfates, acids, fineparticle transition metals, organics, andultrafine particles, and includes most ofthe aggregate surface area and particlenumber in the entire distribution ofatmospheric particles.

The Administrator concurs with thestaff and CASAC recommendations andconcludes that PM2.5 is the appropriateindicator for fine particle standards. Asdiscussed in Unit VI.B. of this preamble,technical details of how PM2.5 is to bemeasured in the ambient air arespecified in the Federal ReferenceMethod (40 CFR part 50, Appendix L).

2. Indicators for the coarse fraction ofPM10. The Criteria Document and StaffPaper conclude that epidemiologicalinformation, together with dosimetryand toxicological information, supportthe need for a particle indicator thataddresses the health effects associatedwith coarse fraction particles withinPM10 (i.e., PM10-2.5). As noted above,coarse fraction particles can deposit inthose sensitive regions of the lung ofmost concern. Although the role ofcoarse fraction particles in much of therecent epidemiological results isunclear, limited evidence from studieswhere coarse fraction particles are the

dominant fraction of PM10 suggest thatsignificant short-term effects related tocoarse fraction particles includeaggravation of asthma and increasedupper respiratory illness. In addition,qualitative evidence suggests thatpotential chronic effects may beassociated with long-term exposure tohigh concentrations of coarse fractionparticles.

In selecting an indicator for coarsefraction particles, the Administratortook into account the views of severalCASAC panel members who suggestedusing the coarse fraction directly (i.e.,PM10-2.5) as the indicator. However, theAdministrator notes that the existingambient data base for coarse fractionparticles is smaller than that for fineparticles, and that the only studies ofclear quantitative relevance to effectsmost likely associated with coarsefraction particles have usedundifferentiated PM10. In fact, it was theconsensus of CASAC that it isreasonable to consider PM10 itself as asurrogate for coarse fraction particles,when used together with PM2.5

standards. The monitoring networkalready in place for PM10 is large.Therefore, in conjunction with thedecision to have separate standards forPM2.5, the Administrator concludes,consistent with CASACrecommendations and publiccomments, that it is appropriate toretain PM10 as the indicator for PMstandards intended to protect againstthe effects most likely associated withcoarse fraction particles.

D. Averaging Time of PM2.5 StandardsAs discussed above in this unit, the

Administrator has concluded that PM2.5

is an appropriate indicator for standardsintended to provide protection fromeffects associated primarily with fineparticles. The recent health effectsinformation includes reportedassociations with both short-term (fromless than 1 day to up to 5 days) andlong-term (from a year to several years)measures of PM.

On the basis of this information,summarized in chapter V of the StaffPaper and in the rationale presented inthe proposal, the Administrator hasconsidered both short- and long-termPM2.5 standards.

1. Short-term PM2.5 standard. Thecurrent 24-hour averaging time isconsistent with the majority ofcommunity epidemiological studies,which have reported associations ofhealth effects with 24-hourconcentrations of various PM indicatorssuch as PM10, fine particles, and TSP.Such health effects, includingpremature mortality and increased

hospital admissions, have generallybeen reported with same-day, previousday, or longer lagged single-dayconcentrations, although some studieshave reported stronger associations withmultiple-day average concentrations. Inany case, the Administrator recognizesthat a 24-hour PM2.5 standard caneffectively protect against episodeslasting several days, since attainment ofsuch a standard would provideprotection on each day of a multi-dayepisode, while also protecting sensitiveindividuals who may experience effectsafter even a single day of exposure.

Although most reported effects havebeen associated with daily or longermeasures of PM, evidence also suggeststhat some effects may be associated withPM exposures of shorter durations. Forexample, controlled human and animalexposures to specific components offine particles, such as acid aerosols,suggest that bronchoconstriction canoccur after exposures of minutes tohours. Some epidemiological studies ofexposures to acid aerosols have alsofound changes in respiratory symptomsin children using averaging times lessthan 24 hours. However, such reportedresults do not provide a satisfactoryquantitative basis for setting a fineparticle standard with an averaging timeof less than 24 hours, nor do currentgravimetric mass monitoring devicesmake such shorter durations generallypractical at present. Further, theAdministrator recognizes that a 24-houraverage PM2.5 standard which leads toreductions in 24-hour averageconcentrations is likely to lead as wellto reductions in shorter-term averageconcentrations in most urbanatmospheres, thus providing somedegree of protection from potentialeffects associated with shorter durationexposures.

2. Long-term PM2.5 standard.Community epidemiological studieshave reported associations of annualand multi-year average concentrationsof PM10, PM2.5, sulfates, and TSP withan array of health effects, notablypremature mortality, increasedrespiratory symptoms and illness (e.g.,bronchitis and cough in children), andreduced lung function. The relativerisks associated with such measures oflong-term exposures, although highlyuncertain, appear to be larger than thoseassociated with short-term exposures.Based on the available epidemiology,and consistent with the limited relevanttoxicological and dosimetricinformation, the Administratorconcludes that significant, andpotentially independent, healthconsequences are likely associated withlong-term PM exposures.


28 Of the 19 panel members who joined in theconsensus for PM2.5 standards, 17 (90 percent)recommended a 24-hour standard and 13 (70percent) recommended an annual standard (Wolff,1996b).

The Administrator has consideredthis evidence, which suggests that somehealth endpoints reflect the cumulativeeffects of PM exposures over a numberof years. In such cases, an annualstandard would provide effectiveprotection against persistent long-term(several years) exposures to PM.Requiring a much longer averaging timewould also complicate andunnecessarily delay control strategiesand attainment decisions.

The Administrator has alsoconsidered the seasonality of emissionsof fine particles and their precursors insome areas (e.g., wintertime smoke fromresidential wood combustion,summertime regional acid sulfate andozone formation), which suggests thatsome effects associated with annualaverage concentrations might be theresult of repeated seasonally highexposures. However, different seasonsare likely of concern in different partsof the country, and the current evidencedoes not provide a satisfactoryquantitative basis for setting a nationalfine particle standard in terms of aseasonal averaging time.

In addition, the Administratorrecognizes that an annual standardwould have the effect of improving airquality broadly across the entire annualdistribution of 24-hour PM2.5

concentrations, although such astandard would not as effectively limitpeak 24-hour concentrations as would a24-hour standard. The risk assessmentsummarized above found that becausesuch 24-hour peaks contribute muchless to the total health risk over a yearthan the more numerous low- to mid-range PM2.5 levels, an annual standardcould also provide effective protectionfrom health effects associated withshort-term exposures to PM2.5 as well asthose associated with long-termexposures (see figure 2; 61 FR 65652-65653, December 13, 1996).

3. Combined effect of annual and 24-hour standards. For the reasonsoutlined in Units II.C.1. and 2. of thispreamble, the Administrator concludedin the proposal that a short-term PM2.5

standard with a 24-hour averaging timecan serve to control short-term ambientPM2.5 concentrations, thus providingprotection from health effects associatedwith short-term (from less than 1-day toup to 5-day) exposures to PM2.5. Further,a long-term PM2.5 standard with anannual averaging time can serve tocontrol both long- and short-termambient PM2.5 concentrations, thusproviding protection from health effectsassociated with long-term (seasonal toseveral years) and, to some degree,short-term exposures to PM2.5.

EPA received comparatively fewpublic comments on these proposedaveraging times. Those supporting PM2.5

standards also strongly supportedadopting both annual and 24-houraveraging times. Many of thoseopposing PM2.5 standards, for thereasons discussed in Unit II.B. of thispreamble, provided contingentcomments that variously supported bothaveraging times for PM2.5 standards inthe event the Administrator disagreedwith their overall recommendations.Other opponents of PM2.5 standardsdisagreed with having two standards onadministrative grounds, or becausesome CASAC members did not supportboth averaging times.

The relationship between standardsfor the two averaging times is discussedbelow in this unit. In essence, based onits examination of the effects data andair quality relationships, EPA believesthat a single PM2.5 standard (24-hour orannual) either would not provideadequate protection against effects ofconcern for all averaging times, orwould be inefficient in the sense that itwas more stringent than necessary for atleast one averaging time. Contrary tocommenters who focused on minorityCASAC opinions, EPA notes that a clearmajority of CASAC supported both 24-hour and annual standards28. Afterconsidering public comments onaveraging time and the rationaleoutlined above, the Administrator hasconcluded that both 24-hour and annualPM2.5 standards are appropriate.

The Administrator next consideredthe potential combined effects of suchstandards on PM concentration levelsand distributions. The existing healtheffects evidence could, of course, beused to assess the form and level of eachstandard independently, with short-term exposure health effects evidencebeing used as the basis for a 24-hourstandard and the long-term exposurehealth effects evidence as the entirebasis for an annual standard. SomeCASAC panel members apparently usedthis approach as a basis for their viewson appropriate averaging times andstandard levels. In particular, a fewmembers focused only on a 24-hourPM2.5 standard in light of the relativestrength of the short-term exposurestudies. On the other hand, twomembers focused only on an annualstandard, recognizing that strategies tomeet an annual standard would provideprotection against effects of both short-and long-term exposures.

As noted above in this unit,attempting to provide protection for allof the effects identified in long- andshort-term PM exposure studies with asingle averaging time would result ineither inadequate protection for someeffects, or unnecessarily stringentcontrol for others. The Administratorhas, instead, emphasized a policyapproach that considers the consistencyand coherence, as well as thelimitations, of the body of evidence asa whole, and recognizes that there arevarious ways to combine two standardsto achieve an appropriate degree ofpublic health protection. Such anapproach to standard setting, whichintegrates the body of health effectsevidence and air quality analyses, andconsiders the combined effect of thestandards, has the potential to result ina more effective and efficient suite ofstandards than an approach that onlyconsiders short- and long-term exposureevidence, analyses, and standardsindependently.

In considering the combined effect ofsuch standards, the Administrator notesthat while an annual standard wouldfocus control programs on annualaverage PM2.5 concentrations, it wouldalso result in fewer and lower 24-hourpeak concentrations. Alternatively, a 24-hour standard that focuses controls onpeak concentrations could also result inlower annual average concentrations.Thus, either standard could be viewedas providing both short- and long-termprotection, with the other standardserving to address situations where thedaily peaks and annual averages are notconsistently correlated.

The Administrator proposed that thesuite of PM2.5 standards could mosteffectively and efficiently be defined bytreating the annual standard as thegenerally controlling standard forlowering both short- and long-termPM2.5 concentrations. In conjunctionwith the annual standard, the 24-hourstandard would serve to provideprotection against days with high peakPM2.5 concentrations, localized ‘‘hotspots,’’ and risks arising from seasonalemissions that would not be wellcontrolled by a national annualstandard.

Relatively few public comments wereaddressed specifically to the proposalthat the annual standard be directedtoward controlling both 24-hour andannual levels (thereby basing the annualstandard on an evaluation of both theshort- and long-term health effectsinformation), with the 24-hour standardbeing used to address more localizedshort-term peaks. A number ofcommenters, notably some among thegroups opposing any revised PM


29 A related comment criticized the riskassessment conclusion that peak 24-hourconcentrations contribute much less to the total riskover a year as inconsistent with the experience inhistoric air pollution episodes. EPA disagrees.While the historic London episodes werequantitatively different from those assumed in therisk assessment, the record over 14 London wintersindicates a continuum of effects down to the lowestlevels. It is therefore likely that the cumulativeincrease in mortality calculated for all the days inthe whole 14-year period would not be dominatedby the more limited number of episode days.

30 This point is buttressed by studies that havetaken out a limited number of higher PMconcentration days with little effect on the effectsestimates or significance of the association (e.g.,Schwartz et al., 1996; Pope and Dockery, 1992).

standards, appeared to have ignored thisfundamental aspect of the proposal,judging by their assertions that the solebasis for EPA’s proposed annualstandards was two long-term exposurestudies (Dockery et al., 1993; Pope et al.1995). This is incorrect; as the proposalstates, EPA based the proposed annualstandard level on a wider range of short-and long-term exposure studies. Othercommenters, including someenvironmental groups, reservedcomment on this specific issue, butexpressed concerns that the specificlevels for both standards were notstringent enough, regardless of whichstandard is intended to be controlling.Issues regarding specific levels arediscussed below in Unit II.F. of thispreamble.

Some commenters, however,disagreed with the proposition thatEPA’s proposed approach wouldnecessarily provide the most effectiveand efficient standards. In the view ofsome who opposed PM2.5 standards, thelikelihood that there are thresholdsbelow which no effects occur meansthat a 24-hour standard would be moreefficient than an annual standard. Inthis view, the reductions made on daysthat were below the threshold wouldprovide no protection.29 Somecommenters also noted that while amajority of CASAC members favoredboth annual and 24-hour standards,more recommended 24-hour standards.

While the available epidemiologicalstudies provide strong evidencesuggesting that PM causes or contributesto health effects at levels below thecurrent standards, EPA agrees, as statedpreviously, that uncertainties increasemarkedly at lower concentrations.Nevertheless, the level or even existenceof population thresholds below whichno effects occur cannot be reliablydetermined by an examination of theresults from the available studies.Analyses have placed some limits,however, and EPA has consideredhypothetical thresholds in its riskassessment. As noted in Unit II.A. ofthis preamble, even assuming anexample threshold of 18 µg/m3, the riskassessment (see Figure 2c; 61 FR 65653,December 13, 1996) finds that most of

the annual aggregate risk associatedwith short-term exposures still resultsfrom the large number of days at lowerto mid-range values above the mean.Given that neither the CriteriaDocument nor commenters haveprovided quantitative evidenceregarding the likelihood of a thresholdat levels much higher than the aboveexample, EPA believes that the evidenceprovided in the risk assessment does notsupport the commenters’ position. Asnoted above, EPA believes that mostCASAC opinions on averaging timereflect panelists’ judgments on therelative strength of the short-termexposure epidemiological studies, ajudgment that EPA shares. Althoughmost CASAC panel members did notoffer an opinion on the use of short-termexposure studies in specifying annualstandards, two panelists did supportthis notion. EPA therefore believes thisapproach is neither inconsistent withthe underlying science nor discordantwith the advice of CASAC.

Another concern was raised by someair pollution control officials whootherwise supported revised PMstandards. These commenters state that,from an implementation perspective, itis often easier to design controlstrategies for single short-term eventsthan for annual averages. Aside fromwhether this is a proper considerationin establishing NAAQS, the point in facthighlights one of the importantstrengths of an annual standard inaddressing short-term risks associatedwith PM2.5. As noted by thecommenters, risk management for ashort-term standard focuses on acharacteristic ‘‘design value’’ episoderesponsible for peak concentrations. ForPM, such peak values can be associatedwith single source contributions.Meteorology, relative sourcecontributions, and resulting particlecomposition for that day may or maynot be typical for the area or for theyear. Yet the short-term exposureepidemiological results are largelydrawn from studies that associatedvariations in area-wide effects withmonitor(s) that gauged the variation indaily levels over the course of up to 8years. The strength of the associations inthese data is demonstrably in thenumerous ‘‘typical’’ days in the upper tomiddle portion of the annualdistribution, not on the peak days.30 Forthese reasons, strategies that focus onlyon reducing peak days are less likely to

achieve reduction of the mix andsources of urban and regional-scale PMpollution most strongly associated withhealth effects. Although designingcontrol strategies to reduce annuallevels may be more difficult than for 24-hour standards, the available short- andlong-term epidemiological data suggestit is also likely to result in a greaterreduction in area-wide populationexposure and risk.

The Administrator concludes that themost effective and efficient approach toestablishing PM2.5 standards is to treatthe annual standard as the generallycontrolling standard for lowering bothshort- and long-term PM2.5

concentrations, while the 24-hourstandard would serve to provideprotection against days with high peakPM2.5 concentrations, localized ‘‘hotspots,’’ and risks arising from seasonalemissions that would not be wellcontrolled by a national annualstandard. In reaching this view, theAdministrator took into account thepublic comments and the factorsdiscussed below in this unit.

(1) Based on one of the keyobservations from the quantitative riskassessment summarized above (seeFigures 2a,b,c; 61 FR 65652-65653,December 13, 1996), the Administratornotes that much if not most of theaggregate annual risk associated withshort-term exposures results from thelarge number of days during which the24-hour average concentrations are inthe low- to mid-range, below the peak24-hour concentrations. As a result,lowering a wide range of ambient 24-hour PM2.5 concentrations, as opposedto focusing on control of peak 24-hourconcentrations, is the most effective andefficient way to reduce total populationrisk. Further, there is no evidencesuggesting that risks associated withlong-term exposures are likely to bedisproportionately driven by peak 24-hour concentrations. Thus, an annualstandard that controls an area’sattainment status is likely to reduceaggregate risks associated with bothshort- and long-term exposures withmore certainty than a 24-hour standard.

(2) The consistency and coherence ofthe health effects data base are,therefore, more directly related to themore frequently occurring PMexposures reflected in study periodmean measures of air quality (e.g., theannual distributions of 24-hour PMconcentrations), than to the potentiallysite-specific and/or otherwise infrequentPM exposures reflected in a limitednumber of peak 24-hour concentrations.More specifically, judgments about thequantitative consistency of the largenumber of short-term exposure studies


31 The notice of proposed revisions to 40 CFR part58 recognized that a single appropriately sitedmonitor could suffice for an area in place of anaverage of multiple monitors.

reporting associations with 24-hourconcentrations arise from comparing therelative risk results per PM increment asderived from analyzing the associationsacross the entire duration of the studies.These studies typically spanned at leastan annual time frame and the reportedassociations are most stronglyinfluenced by the large number of daystoward the middle of the distribution.

(3) An annual average measure of airquality is more stable over time than are24-hour measures. Thus, a controllingannual standard is likely to result in thedevelopment of more consistent riskreduction strategies over time, since anarea’s attainment status will be lesslikely to change due solely to year-to-year variations in meteorologicalconditions that affect the formation offine particles, than under a controlling24-hour standard.

Under this policy approach, theannual PM2.5 standard would serve inmost areas as the target for controlprograms designed to be effective inlowering the broad distribution of PM2.5

concentrations, thus protecting not onlyagainst long-term effects but also short-term effects as well. In combinationwith such an annual standard, the 24-hour PM2.5 standard would be set so asto protect against the occurrence of peak24-hour concentrations, particularlypeak concentrations that presentlocalized or seasonal exposures ofconcern in areas where the highest 24-hour-to-annual mean PM2.5 ratios areappreciably above the national average.

E. Form of PM2.5 Standards1. Annual standard. As discussed in

some detail during the last review of thePM NAAQS (see 49 FR 10408, March20, 1984; 52 FR 24634, July 1, 1987) andin the December 13, 1996 proposal, theannual arithmetic mean form of thecurrent annual PM10 standard (i.e., theannual arithmetic mean averaged over 3years) is a relatively stable measure ofair quality that reflects the totalcumulative dose of PM to which anindividual or population is exposed.Short-term peaks have an influence onthe arithmetic mean that is proportionalto their frequency, magnitude, andduration, and, thus, their contribution tocumulative exposure and risk. As aresult, the annual arithmetic mean formof an annual standard providesprotection across a wide range of the airquality distribution contributing toexposure and risk, in contrast to otherforms, such as the geometric mean, thatde-emphasize the effects of short-termpeak concentrations.

While almost no commenters tookspecific issue with use of an annualarithmetic mean, a number of

commenters disagreed with averagingover 3 years for both the annual and 24-hour standards because of their desirefor quick action in the initialimplementation of PM2.5 controls. TheAdministrator recognizes theimportance of promptly implementingappropriate control programs, but shedoes not believe that implementationstart-up concerns are an adequate basisfor adopting a form (e.g., a single yearannual average) that would provide lessstable risk reduction in the long-run.Therefore, the Administrator continuesto concur with the Staff Paperrecommendation, supported by CASAC,to use the annual arithmetic mean,averaged over 3 years, as the form for anannual PM2.5 standard consistent withthe current form of the annual PM10

standard. Nevertheless, EPA intends toaddress the concerns of those whocommented that the 3-year form mightprevent the public from being informedabout the air quality status of theircommunities. As outlined in Unit II.H.of this preamble, EPA plans to issuerevised Pollutant Standard Indexcriteria for PM2.5, to ensure the public isinformed promptly about air qualitystatus.

The Staff Paper and some CASACpanel members also recommended thatconsideration be given to calculating thePM2.5 annual arithmetic mean for anarea by averaging the annual arithmeticmeans derived from multiplemonitoring sites within a monitoringplanning area. In proposing acalculation method for annualarithmetic averages that involves spatialaveraging of monitoring data, theAdministrator reasoned as follows:

(1) Many of the community-basedepidemiological studies examined inthis review used spatial averages, whenmultiple monitoring sites wereavailable, to characterize area-wide PMexposure levels and the associatedpopulation health risk. In those studiesthat used only one monitoring location,the selected site was chosen to representcommunity-wide exposures, not thehighest value likely to be experiencedwithin the community. Thus, spatialaverages are most directly related to theepidemiological studies used as thebasis for the proposed revisions to thePM NAAQS.

(2) As a part of the overall policyapproach discussed in Unit II.D. of thispreamble, the annual PM2.5 standardwould be intended to reduce aggregatepopulation risk from both long- andshort-term exposures by lowering thebroad distribution of PM2.5

concentrations across the community.An annual standard based on spatiallyaveraged concentrations would better

reflect area-wide PM exposure levelsthan would a standard based onconcentrations from a single monitorwith the highest measured values.

(3) Under this policy approach, the24-hour PM2.5 standard would beintended to work in conjunction with aspatially averaged annual PM2.5

standard by providing protection againstpeak 24-hour concentrations, localized‘‘hot spots,’’ and higher PM2.5

concentrations arising from seasonalemissions and meteorology that wouldnot be as well controlled by an annualstandard. Accordingly, the 24-hourPM2.5 standard should be based on thesingle population-oriented monitoringsite within the monitoring planning areawith the highest measured values.

Based on these considerations, theAdministrator proposed that the form ofan annual PM2.5 standard be expressedas the annual arithmetic mean,temporally averaged over 3 years andspatially averaged over all designatedmonitoring sites,31 which, inconjunction with a 24-hour PM2.5

standard, was intended to provide themost appropriate target for reducingarea-wide population exposure to fineparticle pollution. Recognizing thecomplexities that spatial averagingmight introduce into risk managementprograms, in the proposal theAdministrator also requested commenton the alternative of basing the annualstandard for PM2.5 solely on the singlepopulation-oriented monitor site withinthe monitoring planning area with thehighest 3-year average annual mean.

The proposed approach to designatingsites that are appropriate for spatialaveraging was based on criteria andconstraints contained in the proposedrevision to the monitoring siting andnetwork planning requirements in 40CFR part 58. In proposing this approach,the Administrator noted concernsregarding the development andimplementation of appropriate andeffective criteria for the selection of sitesand designations of areas for spatialaveraging.

A number of commenters whootherwise favored setting PM2.5

standards objected to the concept ofpopulation-oriented monitors andexpressed the view that any monitorregardless of where it was sited shouldbe eligible for comparison to the annualPM2.5 standard. They further maintainedthat the proposed provisions for spatialaveraging would fail to provideadequate health protection because


32 The 40 CFR part 58 proposed rule identifiedthe proposed criteria for monitors to be averaged;namely, monitors must be properly sited to reflectpopulation-orientation, primarily influenced bysimilar sources, and within +/-20 percent of theaverage levels and a specific degree of correlation(or meet a ‘‘homogeneity’’ constraint). Additionalcriteria include demonstrations that the monitors tobe averaged are influenced primarily by similarsources (e.g., to prevent the placement of monitorsupwind in unrepresentative locations), EPAoversight of the monitoring program which includesregular review and approval of the State PMmonitoring network design, and other criteria toensure proper monitor siting. The final ruleincludes the addition of provisions that the StatePM monitoring network design be available forpublic inspection.

33 Daily mortality studies generally use urban ormetro-areawide effects statistics in conjunctionwith single or multiple monitors that index day-to-day pollution changes across the area. Ito et al.(1995) found that spatial averages from multiple PMmonitors in Chicago were better correlated withdaily mortality than were most single monitors, butthat single monitors were also associated. A numberof morbidity studies (e.g., Schwartz et al., 1994;Neas et al., 1995; Raizenne et al.; 1996) usedcommunity scale monitors and effects informationfrom a defined group of subjects from thecommunity, who were more closely represented bythe monitor.

34 Because the 24-hour standard is designed toaddress localized peaks, it would be inappropriateto extend spatial averaging forms to this standard.

‘‘clean areas’’ and ‘‘dirty areas’’ wouldbe averaged together. Some commentersexpressed concern that the proposedconstraints on spatial average would notbe sufficient to prevent use of suchaveraging to avoid pollution abatement.Others may not have fully understoodthe implications of the specificconstraints and siting requirementsdiscussed in the proposed revisions to40 CFR part 58, which were intended toensure that the population-orientedmonitors used for the annual standardwere actually reflective of community-wide exposures and that the spatialaverages did not include non-representative monitored values fromeither ‘‘clean areas’’ or ‘‘dirty areas.’’32

In order to clarify the intent that thespatially averaged annual standardprotect those in smaller communities, aswell as those in larger populationcenters, the final revisions to 40 CFRpart 58 adopt the term ‘‘community-oriented’’ monitors.

Other commenters, who supportedPM2.5 annual standards, endorsed theconcept of spatial averaging as beingmore reflective of the air quality dataused in the underlying health studiesand because there is general uniformityof fine particle concentrations across anarea. Opponents of the PM2.5 standardsexpressed contingent support for spatialaveraging in concept, again citing thelinkage to the underlying health studies.Indeed, they advocated the extension ofspatial averaging to the daily form of thestandard, and/or recommended lessconstrained spatial averaging to allowfor averaging across entire metropolitanareas.

The Administrator, of course, sharescommenters’ concerns that the form ofthe standards, in conjunction with othercomponents of the standards, mustprotect public health adequately againstrisks associated with PM. It was for thisreason that EPA proposed a policyapproach providing for greatest overallrisk reduction for all citizens in thecommunity from exposures to the mixof urban and regional scale PM

pollution most strongly associated withhealth effects. In specificallyconsidering whether to allow for the useof spatial averaging, the Administratorplaced great weight on consistency withthe underlying body of health effectsevidence. The Administrator is mindfulthat some community studies reliedinherently on exposure and effectsestimates that reflect comparativelybroad spatial scales, as highlighted bythose commenters desiring to extendpermissible averaging; however, thistype of exposure characterization maynot be appropriate for all circumstancesand might leave some areas withoutadequate protection.33

For these reasons, the 40 CFR part 58proposal package contained criteria andconstraints on spatial averaging. Thesecriteria and constraints were intended toensure that spatial averaging would notresult in inequities in the level ofprotection provided by the PMstandards. The Administrator againrecognizes that either a single properlysited community-oriented monitor, oran average of more than one suchmonitors, are both appropriate indicesof area-wide population exposures. Bothare consistent with monitoringapproaches used in communityepidemiological studies upon which thestandards are based. On the other hand,comparing the annual PM2.5 standard tothe maximum concentrations at a sitethat is not representative of communityexposures, as some have suggested,would be inconsistent with theAdministrator’s goal of using the annualstandard to reduce urban and regionalscale exposures and risks. Further, theAdministrator believes that the criteriaand, siting requirements contained in 40CFR part 58, provide adequatesafeguards against inappropriateapplication of spatial averaging.Therefore, the Administrator continuesto believe that an annual PM2.5 standardreflective of area-wide exposures, inconjunction with a 24-hour standarddesigned to provide adequate protectionagainst localized peak or seasonal PM2.5

levels, reflects the most appropriateapproach for public health against the

effects of PM reported in the scientificliterature.34

The majority of comments from Statesstressed the need for flexibility inspecifying network designs and spatialaveraging, given that the nature andsources of particle pollution vary fromone area to another. One State agencyspecifically requested the flexibility tochoose whether to use a singlecommunity-oriented monitor or aspatial average of several of suchmonitors, arguing that it is appropriateto provide this flexibility as PM2.5

monitoring networks evolve and toaddress the diversity of local conditions.

As a result of EPA’s evaluation ofthese comments, the requirements of 40CFR part 50, Appendix K, and 40 CFRpart 58 have been revised to clarify thatthe implementing agencies have theflexibility to compare the annual PM2.5

standard either to the measured value ata single representative community-oriented monitoring site, or to the valueresulting from an average of community-oriented monitoring sites that meet therevised criteria and constraintsenumerated in the 40 CFR part 58 finalrule.

In the Administrator’s view, the finalcriteria and siting requirementscontained in 40 CFR part 58 and in thenew 40 CFR part 50, Appendix N,address the concerns raised by thesecommenters about the protectionafforded by the form of the annualstandard. Therefore, the Administratorcontinues to believe that the form of aPM2.5 annual standard should beexpressed as an annual arithmetic mean,averaged over 3 years, from single ormultiple community-oriented monitors,in accordance with 40 CFR part 50,Appendix N and 40 CFR part 58. In herjudgment, an annual standard expressedin this manner and set at an appropriatelevel, in conjunction with a 24-hourPM2.5 standard, will adequately protectpublic health.

2. 24-hour standard. The current 24-hour PM10 standard is expressed in a ‘‘1-expected-exceedance’’ form. That is, thestandard is formulated on the basis ofthe expected number of days per year(averaged over 3 years) on which thelevel of the standard will be exceeded.The test for determining attainment ofthe current 24-hour standard ispresented in Appendix K to 40 CFR part50.

As discussed in the proposal, sincepromulgation of the current 24-hourPM10 standard in 1987, a number ofconcerns have been raised about the 1-


35 See sections 303, 110(a)(2)(y); 40 CFR part 51.EPA intends to establish a significant harm level forPM2.5 and associated guidance so States candevelop appropriate emergency episode plans. Thesignificant harm and episode criteria will beincluded in forthcoming proposed revisions to 40CFR part 51 and 40 CFR part 58 implementationguidance. In the interim, existing PM10 emergencyepisode plans should be triggered by events of thismagnitude.

expected-exceedance form. Theseinclude, in particular, the year-to-yearstability of the number of exceedances,the stability of the attainment status ofan area, and the complex data handlingconventions specified in 40 CFR part 50,Appendix K, including the proceduresfor making adjustments for missing dataand less-than-every-day monitoring.

In light of these concerns, the StaffPaper and several CASAC panelmembers (Wolff, 1996b) recommendedthat consideration be given to adoptionof a more stable and robust form for 24-hour standards. In considering thisrecommendation for the proposal, theAdministrator noted that the use of aconcentration-based percentile formwould have several advantages over thecurrent 1-expected-exceedance form:

(1) Such a concentration-based formwould be more directly related to theambient PM concentrations that areassociated with health effects. Giventhat there is a continuum of effectsassociated with exposures to varyinglevels of PM, the extent to which publichealth is affected by exposure toambient PM is related to the actualmagnitude of the concentration, not justwhether the concentration is above aspecified level. With an exceedance-based form, days on which the ambientconcentration is well above the level ofthe standard are given equal weight tothose days on which the concentrationis just above the standard (i.e., each dayis counted as one exceedance), eventhough the public health impact on the2 days is significantly different. With aconcentration-based form, days onwhich higher concentrations occurwould weigh proportionally more thandays with lower concentrations for thedesign value, since the actualconcentrations would be used directlyin determining whether the standard isattained.

(2) A concentration-based percentileform would also compensate for missingdata and less-than-every-daymonitoring, thereby reducing oreliminating the need for complex datahandling procedures in the 40 CFR part50, Appendix K test for attainment. Asa result, an area’s attainment statuswould be based directly on monitoringdata rather than on a calculated valueadjusted for missing data or less-than-every-day monitoring.

(3) Further, a concentration-basedform, averaged over 3 years, would alsohave greater stability than the expectedexceedance form and, thus, wouldfacilitate the development of morestable implementation programs by theStates.

The proposal discussed variousspecific percentile values for such a

form (e.g., 90th to 99th percentiles),taking into account two factors. First,the 24-hour PM2.5 standard is intendedto supplement the annual PM2.5

standard by providing additionalprotection against extremely high peakdays, localized ‘‘hot spots,’’ and risksarising from seasonal emissions.Second, given an appropriate level ofhealth protection, the form of the 24-hour PM2.5 standard should provide anappropriate degree of increased stabilityrelative to the current form. TheAdministrator noted in the proposal thata more stable statistic would reduce theimpact of a single high exposure eventthat may be due to unusualmeteorological conditions alone, andthus would provide a more stable basisupon which to design effective controlprograms.

With these purposes in mind, theAdministrator observed in the proposalthat while a percentile value such as the90th or 95th would provide substantiallyincreased stability when compared to amore extreme air quality statistic (e.g.,the current 1-expected-exceedanceform), it would likely not serve as aneffective supplement to the annualstandard, because it would allow a largenumber of days with peak PM2.5

concentrations above the standard level.For example, in a 365-day data base, the90th and 95th percentiles would equalthe 37th and 19th highest 24-hourconcentrations, respectively. On theother hand, a percentile value selectedmuch closer to the tail of the air qualitydistribution (e.g. a 99th or greaterpercentile) would not likely providesignificantly more health protection orsignificantly increased stability ascompared to a 1-exceedance form. Inbalancing these issues in the proposal,the Administrator ultimately proposed a98th percentile value form of thestandard.

Some commenters maintained thatEPA should retain the current 1-expected-exceedance form for the 24-hour PM2.5 standard to limit the numberof days per year that the standard isexceeded. These commenters apparentlygave little weight to EPA’s rationale thata concentration-based form is moredirectly related to ambient PMconcentrations that are associated withhealth effects because it takes intoaccount the magnitude of PMconcentrations, not just whether theconcentrations are above a specificlevel. These commenters alsodiscounted the other advantages of aconcentration-based percentile formoutlined above in this unit. A numberof other commenters supported theconcentration-based percentile form forthe reasons outlined in the proposal but,

as discussed below in this unit, arguedfor alternative percentile values thatwere higher or lower than the proposed98th percentile value.

EPA continues to believe that aconcentration-based percentile form ismore reflective of the health risk posedby elevated PM concentrations, becauseit gives proportionally greater weight todays when concentrations are wellabove the level of the standard than todays when the concentrations are justabove the standard. This factor, coupledwith the other advantages outlinedabove in this unit, leads EPA toconclude that a concentration-basedpercentile form will provide for moreeffective health protection than a 1-expected-exceedance form.

Some commenters supporting a singleexceedance form or a more restrictiveconcentration-based percentile form(e.g. a 99th percentile) expressed concernthat the proposed 98th percentile formcould allow too many highconcentration excursions, and thus failto provide adequate protection againstseasonal emissions problems orlocalized peaks. In particular, somecommenters expressed concerns that inareas with strongly seasonal emissions,such as western areas with winterinversions, over a three year period anarea could experience severalexcursions in which levels could reachas high as 250 µg/m3 and still complywith both the annual and dailystandards if the remainder of the dayshad low levels (e.g., 10 µg/m3).Although this combination of events istheoretically possible, EPA believes it isunlikely. Moreover, if such episodicevents did occur, the Act provides foremergency State or Federal action toaddress them.35 In view of the limits ontruely episodic peak concentrations,EPA believes that an appropriatelyselected 24-hour standard with aconcentration-based 98th percentile formcan provide a stable and adequatelyprotective supplement to the annualstandard in areas with periodic peakconcentrations.

Other commenters who were alsoconcerned with monitoringrequirements associated with spatialaveraging in the annual standard,argued that a 98th percentile form,coupled with the proposed monitoringrequirements that would limit


36 The 40 CFR part 58 monitoring rule proposedto limit sites that would be eligible for comparisonsto the 24-hour standard to population-orientedmonitoring sites.

compliance monitors for the 24-hourstandard to population-oriented sites,would not protect people residing in ornear localized ‘‘hot spots’’ in someareas.36 The Administrator believes thatthe siting requirements as proposed andfinalized in 40 CFR part 58 forpopulation-oriented sites will provideadequate safeguards for such residentialareas.

Other commenters, who otherwiseopposed setting PM2.5 standards,recommended that alternative lowerpercentiles (e.g., 95th percentiles) beused, if EPA proceeds to set suchstandards. As discussed above in thisunit, however, EPA continues to holdthe view that a 90th to 95th percentileform would not provide an adequatelimit against periodic peak values inareas with low annual values andperiodic high seasonal or source-oriented peaks.

After carefully assessing thecomments received, the Administratoris persuaded that the adoption of a 98th

percentile form for the 24-hour PM2.5

standard measured at each population-oriented monitoring site in an areawould provide an effective supplementto the annual PM2.5 standard. This formwill provide adequate protection against24-hour peak PM2.5 levels in locationsdominated by single point sources, aswell as in areas dominated by seasonalemissions. The Administrator alsobelieves that a 98th percentile form, withmore frequent sampling and averagedover 3 years, will provide increasedstability and robustness asrecommended by several members ofthe CASAC panel. For these reasons, theAdministrator has decided to adopt the98th percentile form for the final PM2.5

24-hour standard. The 24-hour PM2.5

standard would be attained when the 3-year average of the 98th percentile of 24-hour concentrations at each populatedoriented monitor within an area is lessthan or equal to the level of thestandard. Further details regarding theinterpretation of the form, as well asassociated calculations and other datahandling conventions are specified inthe new 40 CFR part 50, Appendix N.

F. Levels for the Annual and 24-HourPM2.5 Standards

As discussed in Unit II.D. of thispreamble, the Administrator believesthat an annual PM2.5 standard canprovide the requisite reduction in riskassociated with both annual and 24-hour averaging times in most areas of

the United States. Under this approach,the 24-hour standard would be intendedto provide supplemental protectionagainst extreme peak fine particle levelsthat may occur in some localizedsituations or in areas with distinctvariations in seasonal fine particlelevels. In reaching judgments as toappropriate levels to propose for boththe annual and 24-hour PM2.5 standards,the Administrator has considered thecombined protection afforded by boththe annual and 24-hour standards,taking into account the forms discussedin Unit II.E. of this preamble.

With this approach in mind, theAdministrator has considered theavailable health effects evidence andrelated air quality information presentedin the Criteria Document andsummarized in chapters IV--VII of theStaff Paper, which provides the basis fordecisions on standard levels that wouldreduce risk sufficiently to protect publichealth with an adequate margin ofsafety, recognizing that such standardswill not be risk-free. In so doing, theAdministrator has considered both thestrengths and the limitations of theavailable evidence and information, aswell as alternative interpretations of thescientific evidence advanced by variousCASAC panel members (Wolff, 1996b;Lippmann et al., 1996) and publiccommenters, arising primarily from theinherent uncertainties and limitations inthe health effects studies.

Beyond those factors, but clearlyrelated to them, a range of views havebeen expressed by CASAC panelmembers and the public as to theappropriate policy response to theavailable health effects evidence andrelated air quality information. Towardone end of the spectrum, the view hasbeen expressed that only a very limitedpolicy response is appropriate in light ofthe many key uncertainties andunanswered questions that, takentogether, call into question thefundamental issue of causality in thereported associations between ambientlevels of PM2.5 and mortality and otherserious health effects. Toward the otherend, the view has been expressed thatthe consistency and coherence of theepidemiological evidence should beinterpreted as demonstrating causalityin the relationships between PM2.5 andhealth endpoints that are clearlyadverse, and that uncertainties in theunderlying health effects informationshould be treated, regardless of theirnature, as warranting a maximallyprecautionary policy response. A thirdview would suggest an alternativepolicy response, taking into account notonly the consistency and coherence ofthe health effects evidence, but also the

recognition of key uncertainties andunanswered questions that increasinglycall into question the likelihood of PM-related effects as PM2.5 concentrationsdecrease below the mean values in areaswhere effects have been observed and/or as such concentrations approachbackground levels.

Reflecting these divergent views, bothof the science itself and of how thescience should be used in making policydecisions on proposed standards, theAdministrator considered threealternative approaches to selectingappropriate standard levels, asdescribed in the proposal, ultimatelydeciding to propose standards based ona balanced view of the strengths anduncertainties of the scientificinformation that reflects theintermediate approach.

Judging by the public commentsreceived, EPA accurately reflected thebases for divergent views. A substantialbody of public comments supportedrevising the PM standards by addingPM2.5 standards with levels at least asstringent as those proposed by theAdministrator. In general, however,comments on levels for PM2.5 standardsrevealed a strong dichotomy betweenthose who recommended even strongerstandards than proposed, and those whocounseled against revising the standardsat all. As noted above in this unit, manyin this latter group made contingentrecommendations with respect to thelevels and other aspects of PM2.5

standards, if the Administratorconcluded that any revisions wereappropriate.

This latter group of ‘‘contingent’’commenters recommended levels wellabove those proposed by theAdministrator. These commentersplaced great weight on factors outlinedin Units II.B. and II.C. of this preamblethat led them to oppose any revisions tothe PM standards, including theuncertainties and limitations in theavailable health effects studiesconsidered individually, such as thepossible existence of effects thresholdsand unanswered questions regarding thecausal agent(s) responsible for thereported health effects. Further, theyemphasized the limited amount ofresearch currently available that hasmeasured PM2.5 directly. A substantialgroup recommended that PM2.5

standards be selected so as to beequivalent or close in stringency to thecurrent PM10 standards, and cited theopinions of some CASAC PM panelmembers as support. Some of thesecommenters provided supplementalanalyses of air quality data, arguing thatthey demonstrate that ‘‘equivalent’’standards would be at PM2.5 levels as


37 Nationwide PM2.5 estimates have been derivedfrom the current PM air quality data base, butreflect a significant degree of uncertainty due to thehighly variable relationship between PM2.5 andPM10 air quality values across locations and seasons(Fitz-Simons et al., 1996). The American Iron andSteel Institute (AISI) submitted a useful data base(Cooper Associates, 1997) on PM2.5/PM10

relationships that examines both these predictionsand the issue of equivalence. An EPA examinationof this material, which found some problems withthe analysis and with commenters’ conclusions thatappear inconsistent with the Cooper report, isincluded in the Response to Comments.

38 Some commenters suggest that CASAC andEPA support for PM2.5 standards is based on theneed to stimulate additional monitoring andresearch. While the Administrator agrees that theadditional monitoring and research that wouldaccompany establishment of equivalent ormarginally tighter PM2.5 standards are veryimportant goals, they do not form an adequaterationale for establishing air quality standards.

39 As stated previously, section 109(d) of the Actrequires that, after reviewing the existing criteriaand standards for PM, the Administrator make suchrevisions in the standards and promulgate such new

standards as are appropriate under section 109(b) ofthe Act.

40 This range of levels for a 24-hour PM2.5

standard is close to the lower bound levelsrecommended by four CASAC panel members (20µg/m3); no member supported an annual PM2.5

standard as low as 10 to 12 µg/m3.

41 Some confusion is apparent in commentsregarding the basis on which the Administratorselected levels for the proposed PM2.5 standards,with some commenters suggesting two or at mostthree studies were used, and others suggesting thatEPA relied extensively on uncertain conversionfactors to estimate levels for the standards. Thesecomments are in error. To clarify, as stated in theproposal, the Administrator is basing her decisionto revise the standards on the full range of PMhealth effects studies summarized in the CriteriaDocument and Staff Paper, but in selecting specificlevels for PM2.5 standards, is relying chiefly on U.S.and Canadian studies, listed in Tables V-12 and V-13 of the Staff Paper, that measured fine PM levels.To ease identification and use of these key studies,the short-term exposure studies and key PM airquality statistics are cited in Koman (1996) and alllong-term exposure studies are cited in thispreamble. The referenced memorandum (Koman,1996) has been updated (Koman, 1997) to clarifykey aspects of the studies cited and relevant airquality statistics. In accordance with EPA andCASAC views on the relative strength of thesestudies, greater weight is placed on short-termexposure studies than on long-term exposurestudies. Where studies found statisticallysignificant associations with PM2.5 components(e.g., sulfates and/or acids, in Thurston et al., 1994;Dockery et al., 1996), the corresponding PM2.5 orPM2.1 values from the study are cited. Noconversions were made from the originalmeasurements used in these studies.

high as approximately 95 µg/m3 24-houraverage and 27 µg/m3 annual average.

Having evaluated these comments, theAdministrator rejects both theirunderlying rationale and the specificrecommendations for PM2.5 standardlevels that result in similar or onlymarginally more protection than thatafforded by the current PM10 standards.Aside from technical problems in thecommenters’ supporting analyses on theissue of defining ‘‘equivalent’’standards,37 the Administrator finds thisapproach inconsistent with herconclusions regarding the adequacy ofthe current standards and the need toprovide additional protection asarticulated in Unit II.B. of this preamble.The Administrator believes that, despitewell recognized uncertainties, theconsistency and coherence of theepidemiological evidence and theseriousness of the health effects requirea more protective response thanprovided by ‘‘equivalence’’ or amarginal strengthening of the standards.Moreover, EPA believes that thestandard levels should be based on themost recent assessment of the scientificcriteria for PM, not on applyinguncertain ratios to standard decisionsbased on much more limited evidencein 1987. The Administrator also rejectsthe premise of some38 who suggest thatadopting a standard that prompts littleor no additional control would cause nodelay in risk reduction as compared toconducting monitoring and researchnow and setting a more stringentstandard after the next review. Thesecomments do not consider the realitiesof implementing air quality standards,which ensure that such an approachwould add several years to the riskreduction process. Thus, aside from herobligations under the statute,39 the

Administrator believes that the mostprudent and appropriate course is toestablish appropriately protectivestandards now that put into motionmonitoring and strategy developmentprograms, while at the same timepursuing an expanded research programto improve implementation and toinform the next periodic review of thecriteria and standards.

In sharp contrast to the commentersdiscussed immediately above, a numberof other commenters strongly supportedstandard levels more stringent thanthose proposed by EPA. Thesecommenters supported EPA’sconclusions regarding theepidemiological studies, but wouldplace much less weight on uncertaintiesrelated to the concentration-responserelationships for PM2.5 as a surrogate forPM and the relative importance ofvarious PM components. Based on theirevaluation of the information, and citingthe support of some CASAC panelmembers, these commenters variouslyrecommended 24-hour PM2.5 standardsas low as 18 to 20 µg/m3 and annualstandards of 10 to 12 µg/m3.40

EPA notes that setting such standardswould result in commensuratereductions in health risks only if, infact, there is a continuum of health risksdown to the lower end of the ranges ofair quality observed in the keyepidemiological studies, and only if thereported associations are, in fact,causally related to PM2.5 at the lowestconcentrations measured. Settingstandards at low levels where thepossibility of effects thresholds isgreater, and where there is greaterpotential that other elements in the airpollution mix (or some subset ofparticles within the fine fraction)become more responsible for (or modify)the effects being causally attributed toPM2.5, might result in regulatoryprograms that go beyond those that areneeded to effectively reduce risks topublic health. While placing substantialweight on the results of the key healthstudies in the higher range ofconcentrations observed, EPA ispersuaded that the inherent scientificuncertainties are too great to supportstandards based on the lowestconcentrations measured in suchstudies, which approach the maximumrange of PM2.5 values estimated forshort-term background conditions.

Having considered the commentsreflecting the two contrasting viewssummarized above in this unit, theAdministrator concludes that theapproach she set forth in the proposalis the most appropriate for selectinglevels for annual and 24-hour PM2.5

standards. This approach focusesprimarily on standard levels designed tolimit annual PM2.5 concentrations tosomewhat below those where the bodyof epidemiological evidence is mostconsistent and coherent, in recognitionof both the strengths and the limitationsof the full range of scientific andtechnical information on the healtheffects of PM, as well as associateduncertainties, as interpreted by theCriteria Document, Staff Paper, andCASAC. The Administrator believes thatthis approach appropriately reflects theweight of the evidence as a whole.

In identifying PM2.5 standard levelsconsistent with this overall approach,the Administrator has placed greatestweight on those epidemiological studiesreporting associations between healtheffects and direct measures of fineparticles, most notably those recentstudies conducted in North America(summarized in Tables V-12 and V-13 ofthe Staff Paper).41 Key considerationsand study results upon which thisapproach is based are presented asfollows.

As previously discussed, theAdministrator has concluded that it isappropriate to select the level of theannual standard so as to protect againstthe range of effects associated with bothshort- and long-term exposures to PM,


42 As discussed in the proposal and Appendix Eof the Staff Paper (U.S. EPA, 1996b, p. E-4), thereis generally greatest statistical confidence inobserved associations for levels at and above themean concentration.

43 Based on a public comment, EPA found thatthe mean of 18 µg/m3 in Pope et al. (1995) reportedin the Criteria Document and elsewhere wasactually the mean of median values. Based ontypical air quality relationships, the conventionalarithmetic mean would be approximately 21 to 22µg/m3 (Freas, 1997). The lowest medianconcentration measured in this study (9 µg/m3),which was relied upon by some commenters as abasis for annual standards of 10 µg/m3, is about 11to 12 µg/m3 as an arithmetic mean.

44 Based on public comments and a furtherevaluation of the underlying study, EPA concludesthat the comparable assessment of theconcentration-response function summarized inTable E-3 for Pope et al. (1995) is not appropriate,because it was based on a supplemental ‘‘ecologic’’comparison for these cities and not on the far morereliable prospective-cohort analysis that was themain focus of the paper.

with the 24-hour standard level selectedto provide supplemental protectionagainst peak concentrations that mightoccur over limited areas and/or forlimited time periods. In selecting thelevel for the annual standard, therefore,the Administrator has considered bothshort- and long-term exposure studies.

In accordance with EPA staff andCASAC views on the relative strengthsof the epidemiological studies, theAdministrator has placed greateremphasis on the short-term exposurestudies in selecting the level of theannual standard. The approach she tookto this issue consisted of determining aprovisional level based on the short-term exposure studies, and thendetermining whether the long-termexposure studies are consistent withthat level or, instead, suggest the needfor a lower level. The effects estimatesfrom the short-term exposure studies (inTable V-12 of the Staff Paper) are basedon analyses of daily PM2.5

concentrations that occurred over thecourse of the study period. While effectsmay occur over the full range ofconcentrations observed in the studies,consistent with the discussion of thisissue in Unit II.D. of this preamble, thestrongest evidence for short-term PM2.5

effects occurs at concentrations near thelong-term (e.g., annual) average. Morespecifically, the strength of the evidenceof effects increases for concentrationsthat are at or above the long-term (e.g.,annual) mean levels reported for thesestudies.42 Given the serious nature ofthe potential effects, the Administratorbelieves it is both prudent andappropriate to select a level for anannual standard at or below suchconcentrations. An examination of thelong-term means from the combined sixcity analyses of daily mortality(Schwartz et al., 1996a) and morbidity(Schwartz et al., 1994), together withthose from studies in individual citiesfor which statistically significant PM-effects associations are reported (fromTable V-12 in the Staff Paper), findsmean concentrations ranging from about16 to about 21 µg/m3 (Koman, 1996;1997). In addition, the meanconcentrations in cities where short-term exposure associations arecharacterized in the Criteria Documentas nearly statistically significant (U.S.EPA, 1996a, p. 13-40) range from about11 µg/m3 to 30 µg/m3. Taken together,and placing greatest weight on thosestudies that were clearly statistically

significant, this evidence suggests thatan annual standard level of 15 µg/m3 isappropriate to reduce the risk of effectsfrom short-term exposure to fineparticles.

Before reaching a final conclusion, theAdministrator also examined this levelin light of the effects reported inepidemiological studies of long-termexposures to fine particles (Table V-13in the Staff Paper), which may reflectthe accumulation of daily effects overtime as well as potential effectsuniquely associated with long-termexposures. Even though subject toadditional uncertainties, the long-termexposure studies provide importantinsights with respect to the overallprotection afforded by an annualstandard. These studies were examinedfor general consistency and support forthe levels derived from the short-termexposure studies, and to determinewhether they provide evidence that amore stringent level is needed.

The most direct comparison with thedaily fine particle mortality studies isprovided by two long-term prospectivecohort studies (Dockery et al., 1993;Pope et al., 1995). The annual meanPM2.5 concentration for the multiplecities included in these studies (6 and50 cities, respectively) was 18 µg/m3

(Dockery et al., 1993), and about 21-22µg/m3 for the larger Pope et al. (1995)study.43 The Staff Paper assessment ofthe concentration-response results fromDockery et al. (1993) concluded that theevidence for increased risk was moreapparent at annual concentrations at orabove 15 µg/m3 (Table E-3; U.S. EPA;1996b).44 EPA notes that the estimatedmean values for most of the cities inPope et al. (1995) are above 15 µg/m3.As noted in the Staff Paper and theCriteria Document, the estimatedmagnitude of effects in both long-termexposure mortality studies may berelated to higher historicalconcentrations than the affectedcommunities experienced during the

time period of the studies; thisconsideration suggests that a level of 15µg/m3 would incorporate a margin ofsafety. An examination of morbidityeffects and long-term exposures isprovided by the recent ‘‘24 city’’studies, which found that reduced lungfunction and increased respiratorysymptoms in children followed thegradient in annual mean concentrationsof fine particles and/or acid-sulfatecomponents of fine particles (Raizenneet al., 1996; Dockery et al., 1996). Theresults indicate a greater likelihood ofeffects at annual mean PM2.1 levelsabove about 15 µg/m3 (U.S. EPA, 1996b;Figure V-7). In the judgment of theAdministrator, these studies areconsistent with a standard level of 15µg/m3. While they provide somesuggestion of risks extending to lowerconcentrations, they do not provide asufficient basis for establishing a lowerannual standard level.

Taking the epidemiological studies ofboth short- and long-term exposurestogether, the Administrator believes theconcordance of evidence for PM effectsand associated levels provides clearsupport for an annual PM2.5 standardlevel of 15 µg/m3. This level is belowthe range of annual data most stronglyassociated with both short- and long-term exposure effects, and because evensmall changes in annual means in thisconcentration range can makesignificant differences in overall riskreduction and total populationexposures, the Administrator believes itwill provide an adequate margin ofsafety against the effects observed inthese epidemiological studies.Moreover, the means in areas wherePM2.5 concentrations were statisticallysignificantly associated with dailymortality (about 16 to 21 µg/m3) reflecta 7 to 9-year average; thus, the use of a3-year mean will provide additionalprotection. Although the possibility ofeffects at lower annual concentrationscannot be excluded, the evidence forthat possibility is highly uncertain and,as previously discussed, the likelihoodof significant health risk, if any,becomes smaller as concentrationsapproach the lower end of the range ofair quality observed in the keyepidemiological studies and/orbackground levels.

The final annual standard willprovide substantial protection againstshort-term as well as long-termexposures to particles. Nevertheless, forthe reasons specified above, a spatiallyaveraged annual standard cannot beexpected to offer an adequate margin ofsafety against the effects of all potentialshort-term exposures in areas withstrong local or seasonal sources. The


broad-based community studiesconsidered in this review generallycould not evaluate such peak exposureconditions directly. Given the publichealth purposes of the 24-hour standard,the Administrator believes it should beset at a level that generally supplementsthe control afforded by an annualstandard and proposed an approachbased on providing a reasonable degreeof protection against the peak levelsobserved or expected in communitieswhere health effects have beenassociated with daily levels of fineparticles.

For the reasons specified in theprevious unit, the Administrator hasdecided to use a 98th percentileconcentration-based form of thestandard. As noted in the proposal, the98th percentile 24-hour PM2.5

concentrations in cities with statisticallysignificant or nearly significant short-term fine particle exposure-effectsassociations ranged from 34 µg/m3 to ashigh as 90 µg/m3 (Koman, 1996, 1997).Based on an examination of theseresults, EPA originally proposed a levelfor the 24-hour standard of 50 µg/m3,and solicited comments on higher andlower alternative levels.

In considering comments onalternative levels for the purpose ofmaking a final decision on the 24-hourstandard, the Administrator recognizesthe significant uncertainties inidentifying the extent of the incrementalrisk associated with single peakexposures to PM2.5 in areas where theannual standard is met. Clearly, therisks associated with the 98th percentileair quality data used in the selecting theproposed level are from the same studycities that experienced long-term levelsat varying amounts above that selectedfor the annual standard. It is unclearwhat risks might have been associatedwith such peak levels had the long-termaverages in these areas been below thatselected for the annual standard.Regardless of this uncertainty, it is clearthat reducing the annual concentrationsin such areas to that of the annualstandard would reduce the riskassociated with peak days, whatever themagnitude, as well as that associatedwith the far more numerous days withconcentrations near the annual average.Given these uncertainties and thesignificant degree of protection affordedby the annual standard, theAdministrator is persuaded that it isappropriate to adopt a differentapproach for selecting the levels of the24-hour standard than the oneproposed.

In making a final decision on anappropriate level for the 24-hourstandard, the Administrator considered

several key factors: the significantprotection afforded against short-termexposures by the annual PM2.5 standard;the role of the 24-hour standard inproviding supplemental protectionagainst peak exposures not addressed bythe annual standard; the air quality andeffects information in the studies citedabove; the uncertainties in the risksassociated with infrequent and isolatedpeak exposures in areas that meet theannual standard; the range of levelsrecommended by EPA staff and CASACpanel members; and the extensivepublic comment on the alternativelevels proposed, which ranged between20 and 65 µg/m3. Because of theapproach of establishing the annualstandard as the controlling standard,and, in particular, the decision to set thelevel at the lower end of the annualrange, there is no need to considerlevels in the lower portion of the 24-hour range below the level proposed.Therefore, the Administrator focused onevaluating the margin of safetyassociated with levels between 50 and65 µg/m3.

As has been discussed in previousunits, the extent of total risk over thecourse of a year associated solely witha limited number of peak exposures isuncertain, but it is considerably smallerthan that associated with the entire airquality distribution. Further, the riskassociated with infrequent peak 24-hourexposures in otherwise clean areas isnot well enough understood at this timeto provide a basis for selecting the morerestrictive levels in the range of 50 to 65µg/m3. On the other hand, it is clear thatany standard level within this rangewould provide some margin of safety.Taking into account the factors outlinedabove, the Administrator has concludedthat a 24-hour standard at the level of65 µg/m3 would provide an effectivelimit in the role as a supplement to theannual standard. This level is at theupper end of the range recommended bystaff and most CASAC panel members,and below the levels suggested by someCASAC panel members and by anumber of public commenters.Although this level is not risk free, theAdministrator believes that it wouldprovide an appropriate degree ofadditional protection over that providedby the annual PM2.5 standard.Accordingly, after weighing thesefactors in light of the scientificuncertainties, the Administratorbelieves that a 98th percentile 24-hourPM2.5 standard of 65 µg/m3 wouldprovide an adequate margin of safetyagainst infrequent or isolated peakconcentrations that could occur in areas

that attain the annual standard of 15 µg/m3.

In the Administrator’s judgment, thefactors discussed above provide amplereason to believe that both annual and24-hour PM2.5 standards are appropriateto protect public health from adversehealth effects associated with short- andlong-term exposures to ambient fineparticles. Further, she believes thesefactors provide a clear basis for judgingthat an annual PM2.5 standard set at 15µg/m3, in combination with a 24-hourstandard set at 65 µg/m3, will protectpublic health with an adequate marginof safety.

G. Conclusions Regarding the CurrentPM10 Standards

1. Averaging time and form. Inconjunction with PM2.5 standards, thenew function of PM10 standard(s) is toprotect against potential effectsassociated with coarse fraction particlesin the size range of 2.5 to 10 µm. Coarsefraction particles are plausiblyassociated with certain effects from bothlong- and short-term exposures (EPA1996a,b). Based on qualitativeconsiderations, deposition of coarsefraction particles in the respiratorysystem could be expected to aggravateeffects in individuals with asthma. TheCriteria Document and Staff Paperfound support for this expectation inlimited epidemiological evidence on theeffects of coarse fraction particles,suggesting that aggravation of asthmaand respiratory infections andsymptoms may be associated with dailyor episodic increases in PM10 that aredominated by coarse fraction particles.The potential build-up of insolublecoarse fraction particles in the lung afterlong-term exposures to high levelsshould also be considered.

Based on assessments of the availableinformation in the Criteria Documentand Staff Paper, both the staff andCASAC recommended retention of anannual PM10 standard. The staff, withCASAC concurrence, recommendedretention of the current annualarithmetic mean form of the standard,which is the same form being adoptedfor the annual PM2.5 standard. As notedin the staff assessment, the currentannual PM10 standard offers substantialprotection against the effects of bothlong- and short-term exposure to coarsefraction particles. Public comment wasnearly unanimous in recommendingretention of this standard. TheAdministrator therefore has decided tocontinue a long-term PM10 standard asan annual arithmetic mean, averagedover 3 years.

The staff and CASAC alsorecommended that consideration be


45 Some commenters, including someenvironmental groups and the State of California(Cal EPA, 1997), suggested that the large number ofrecent studies showing effects at PM10 levels belowthe current standards provides a basis forestablishing stricter annual and 24-hour PM10

standards, in conjunction with PM2.5 standards. Asdiscussed in Units II.B. and C. of this preamble,while these studies could be used either to tightenthe PM10 standards or to add standards that tightencontrol of the fine fraction of PM10, the weight ofevidence from all of the relevant information morereadily supports the development of additionalprotection for the PM2.5 fraction.

given to retention of a 24-hour standardto provide additional protection againstpotential effects of short-term exposuresto coarse fraction particles. The staff,with CASAC concurrence, alsorecommended that if a 24-hour standardis retained, the form of the standardshould be revised to provide a morerobust target for coarse fraction particlecontrols. The Administrator originallyproposed a 98th percentile form for the24-hour PM10 standard based primarilyon the reasons outlined above in thisunit regarding the proposed form of the24-hour PM2.5 standard.

The EPA received few commentssupporting elimination of the 24-hourPM10 standard. The main exceptionswere some industries, most notably themining industry, which as noted abovein this unit, argued that the availabledata provide little evidence for coarseparticle effects at current ambient levels.These groups, who generally opposedPM2.5 standards, also argued that thedaily PM10 standard could be eliminatedif PM2.5 standards were set. Based on thepotential aggravation of respiratorysymptoms from short-term exposure tocoarse fraction particles discussed in theCriteria Document and by numerouscommenters, as well as therecommendations of a majority ofCASAC panelists who also supportedPM2.5 standards, the Administratorconcludes it is appropriate to retain a24-hour PM10 standard.

In general, comments received on theform of the 24-hour PM10 standardparalleled those on the form of the PM2.5

standard. Substantial concerns wereexpressed by environmental groups,some States, and others that the 98th

percentile would not provide anadequate limit on the number andmagnitude of 24-hour peak PM10

excursions. While a number of thesecommenters suggested keeping thecurrent 1-expected-exceedance form,EPA believes that a concentration- basedpercentile form offers significantadvantages, as outlined above in thisunit, for both PM indicators. Some airpollution control officials, who wereconcerned about the extent to which the24-hour PM10 standard would berelaxed under the proposed form,suggested consideration of a 99th

percentile form with increasedmonitoring as an appropriatelyprotective form. Other commenters,particularly some industry groups andsome States, strongly supportedconcentration-based percentile forms,with some recommending considerationof the 95th percentile form.

The proposal noted that a percentilevalue selected closer to the ‘‘tail’’ of theair quality distribution (e.g., a 99th or

greater percentile) would notsignificantly increase stability ascompared to the current form. However,an association of 8 State air pollutionagencies commented that a 99th

percentile form could provide increasedstability if combined with a daily or 1-in-3-day sampling frequency and withgreater data capture. In addition, EPAnotes that this concentration-based formis inherently more stable than thecurrent exceedance-based form.

Many of these and other commenterswere concerned that the uncertainties inthe available scientific information onthe effects of coarse particles were areason to be concerned that, assumingthe current standard level was kept, a98th percentile form would represent asignificant relaxation in protectionrelative to the current standards. Unlikethe situation for the new PM2.5

standards, in the case of the PM10

standards, the 24-hour standard hasgenerally been the ‘‘controlling’’standard, making changes to the form ofthe 24-hour standard potentially moresignificant to the overall national levelof protection afforded. Given theuncertainties in the available scientificevidence with respect to the potentialhealth effects of short-term exposures tocoarse fraction particles, theAdministrator is persuaded that thesomewhat more cautious approach withrespect to revising the 24-hour PM10

standard recommended by manycommenters is appropriate. The onlyapproaches available for increasing theextent of protection for this standard ascompared to that of the proposedstandard involve modifying the form orreducing the level. For reasonsdiscussed in the following section, theAdministrator believes it is notappropriate to revise the level of thestandard. In order to provide adequateprotection against the potential riskassociated with multiple short-termpeak exposures to coarse fractionparticles, the Administator acceptscommenters’ recommendations todecrease the frequency of peak values,while still providing for a more stablecontrol target than afforded by thecurrent 1-expected-exceedance form.Therefore, the Administrator concludesthat the 99th percentile concentration-based form, averaged over 3 years, andcombined with more frequent sampling,would be an appropriate form for a 24-hour PM10 standard.

2. Levels for the annual and 24-hourPM10 standards—a. Annual PM10

standard. As a result of the more limitedinformation for coarse fraction particles,the Administrator’s approach forselecting a level of the standard isdirectly related to the approach taken in

the last review of the PM NAAQS. Inthat review, evidence from limitedquantitative studies was used inconjunction with support from thequalitative literature in selecting thelevel of the current annual PM10

standard. In the current review, the staffassessment of the major quantitativebasis for the level of that standard (Wareet al., 1986), together with a more recentrelated study (Dockery et al., 1989),recommended the same range of levelsof concern (40 to 50 µg/m3) as in the1986 staff paper. The staff concludesthat it is possible, but not certain, thatcoarse fraction particles, in combinationwith fine particles, may have influencedthe observed effects at these levels.Based on particle depositionconsiderations, it is possible thatcumulative deposition of coarse fractionparticles could be of concern inchildren, who are more prone to beactive outdoors than sensitive adultpopulations.

Qualitative evidence of other long-term coarse particle effects, mostnotably from long-term build-up ofsilica-containing materials, supports theneed for a long-term standard, but doesnot provide evidence of effects belowthe range of 40 to 50 µg/m3 (U.S. EPA,1996a, p. 13-79). The staff concludesthat the qualitative evidence withrespect to biological aerosols alsosupports the need to limit coarsematerials, but should not form the majorbasis for a national standard (U.S. EPA,1996a, p. 13-79). In addition, staff notesthat the nature and distribution of suchmaterials, which vary from endemicfungi (e.g., valley fever) to pollens largerthan 10 µm, are not appropriatelyaddressed by traditional air pollutioncontrol programs.

Based on its review of the availableinformation, CASAC found ‘‘aconsensus that retaining an annual PM10

NAAQS at the current level isreasonable at this time’’ (Wolff, 1996b).With few exceptions, public commentssupported levels at least as stringent asthe current annual PM10 standard.45

Taking into account these commentsand the above considerations, as morefully detailed in the Staff Paper and the


46 Congress adopted section 169A of the Actbecause of concern that the NAAQS and Preventionof Significant Deterioration programs might notprovide adequate visibility protection nationally,particularly for ‘‘areas of great scenic importance.’’See H.R. Rep. No. 95–294,at 203–205 (1977).

CASAC recommendations, theAdministrator has decided to retain thecurrent annual PM10 standard of 50 µg/m3 to protect against the known andpotential effects of long-term exposureto coarse fraction particles.

b. 24-hour PM10 standard. Asdiscussed above in this unit, EPA staffand CASAC also recommended thatconsideration be given to a 24-hourstandard for coarse fraction particles asmeasured by PM10. Unlike the case forthe annual standard, however, the stafffound that the original quantitative basisfor the level of the current 24-hour PM10

standard (150 µg/m3) is no longerappropriate. Instead, the staff found thatthe main quantitative basis for a short-term standard is provided by the tworecent community studies of exposureto fugitive dust (Gordian et al., 1996;Hefflin et al., 1994). Because thesestudies reported multiple largeexceedances of the current 24-hourstandard, and because of limitations inthe studies themselves, the staffconcluded that they provide no basis tolower the level of the standard below150 µg/m3. Moreover, staff concludedthat none of the qualitative literatureregarding the potential effects of short-term exposure to coarse particlesprovides a basis for a lower standardlevel. Both EPA staff and CASACrecommended that if a 24-hour PM10

standard is retained, the level of thestandard should be maintained at 150µg/m3, although with a revised form.Beyond the comments summarizedabove recommending elimination of the24-hour standard, no commentersrecommended a less stringent level,while some others, as summarizedabove in this unit, recommended morestringent levels. Most comments favoredthe current level.

Having considered these factors andthe public comments, the Administratorjudges that, retention of a 24-hour PM10

standard at the level of 150 µ/m3 witha 99th percentile form is appropriate andwill provide adequate protection againstthe known and potential effects of short-term coarse fraction particle exposuresthat have been identified to date in thescientific literature.

H. Final Decisions on Primary PMStandards

For the reasons discussed above inthis unit, and taking into account theinformation and assessments presentedin the Criteria Document and the StaffPaper, the advice and recommendationsof CASAC, and public commentsreceived on the proposal, theAdministrator is revising the current PMNAAQS by adding new PM2.5 standardsand by revising the form of the current

24-hour PM10 standard. Specifically, theAdministrator is making the followingrevisions:

(1) The suite of PM standards isrevised to include an annual primaryPM2.5 standard and a 24-hour PM2.5

standard.(2) The annual PM2.5 standard is met

when the 3-year average of the annualarithmetic mean PM2.5 concentrations,from single or multiple community-oriented monitors (in accordance withEPA’s final rule on monitoring sitingguidance, 40 CFR part 58, published ina separate document elsewhere in thisissue of the Federal Register) is lessthan or equal to 15 µg/m3, withfractional parts of 0.05 or greaterrounding up.

(3) The 24-hour PM2.5 standard is metwhen the 3-year average of the 98th

percentile of 24-hour PM2.5

concentrations at each population-oriented monitor within an area is lessthan or equal to 65 µg/m3, withfractional parts of 0.5 or greaterrounding up.

(4) The form of the current 24-hourPM10 standard is revised to be based onthe 3-year average of the 99th percentileof 24-hour PM10 concentrations at eachmonitor within an area.In addition, the Administrator isretaining the current annual PM10

standard at the level of 50 µg/m3, whichis met when the 3-year average of theannual arithmetic mean PM10

concentrations at each monitor withinan area is less than or equal to 50 µg/m3, with fractional parts of 0.5 or greaterrounding up.

As discussed below in Units V. andVI. of this preamble, data handlingconventions and completeness criteriafor the revised standards are beingestablished (40 CFR part 50, AppendixN). The reference method for monitoringPM as PM10 for the revised standardshas been established (40 CFR part 50,Appendix M). A new reference methodis being established for monitoring PMas PM2.5 (40 CFR part 50, Appendix L).In a separate document publishedelsewhere in this issue of the FederalRegister, EPA is providing opportunityfor public comment on supplementalinformation relating to the newreference method for monitoring PM asPM2.5 (40 CFR part 50, Appendix L).

As indicated previously, EPA plans topropose related revisions to thePollutant Standards Index for PM (40CFR 58.50) and the significant harmlevel program (40 CFR 51.66) at a laterdate.

III. Rationale for the SecondaryStandards

The Criteria Document and StaffPaper examined the effects of PM onsuch aspects of public welfare asvisibility, materials damage, and soiling.The following discussion of therationale for revising the secondarystandards for PM focuses on thoseconsiderations most influential in theAdministrator’s decision.

A. Need for Revision of the CurrentSecondary Standards

1. Visibility impairment. This unit ofthe document presents theAdministrator’s decision to address thewelfare effects of PM on visibility bysetting secondary standards identical tothe suite of PM2.5 primary standards, inconjunction with the establishment of aregional haze program under section169A of the Act.46 In theAdministrator’s judgment, this approachis the most effective way to addressvisibility impairment given the regionalvariations in concentrations of non-anthropogenic PM as well as otherregional factors that affect visibility,such as humidity. By augmenting theprotection provided by secondarystandards set identical to the suite ofPM2.5 primary standards with a regionalhaze program, the Administratorbelieves that an appropriate degree ofvisibility protection can be achieved inthe various regions of the country.

In coming to this decision, theAdministrator took into account severalfactors, including: The pertinentscientific and technical information inthe Criteria Document and Staff Paper,difficulties inherent in attempting toestablish national secondary standardsto address visibility impairment, thedegree of visibility improvementexpected through attainment ofsecondary standards equivalent to thesuite of PM2.5 primary standards, theeffectiveness of addressing the welfareeffects of PM on visibility through thecombination of a regional haze programand secondary standards for PM2.5

equivalent to the suite of primarystandards, and comments receivedduring the public comment period. TheAdministrator’s consideration of each ofthese factors is discussed below in thisunit.

The Administrator first concluded,based on information presented andreferenced in the Criteria Document and


47 There are 156 mandatory Class I Federal areasprotected by the visibility provisions in sections169A and 169B of the Act. These areas are definedin section 162 of the Act as those national parksexceeding 6,000 acres, wilderness areas andmemorial parks exceeding 5,000 acres, and allinternational parks which were in existence onAugust 7, 1977.

48 Visual range can be defined as the maximumdistance at which one can identify a black objectagainst the horizon sky. It is typically described inmiles or kilometers. Light extinction is the sum oflight scattering and absorption by particles andgases in the atmosphere. It is typically expressed interms of inverse megameters (Mm-1), with largervalues representing poorer visibility. The deciviewmetric describes perceived visual changes in alinear fashion over its entire range, analogous to thedecibel scale for sound. A deciview of 0 representspristine conditions. Under many scenic conditions,a change of 1 deciview is considered perceptible bythe average person.

49 Congress adopted a visibility protectionprogram in section 169A of the Act because itrecognized the impracticability of revising theNAAQS to protect visibility in all areas of thecountry: ‘‘It would be impracticable to require amajor city such as New York or Los Angeles to meetthe same visibility standards as the Grand Canyonand Yellowstone Park.’’ See H.R. Rep. No. 95–294at 205. (1977)

50 Estimates of annual average visibilityimprovements assume that, on a percentage basis,the reduction for each fine particle component isequal to the % reduction in the mass of fineparticles, and that the overall light extinctionefficiency of the fine particle pollutant mix does notchange. Further, for the estimates presented here,the reductions in fine mass at monitored locationsare assumed to reflect the spatial averageconcentrations through the viewing distance.(Damberg and Polkowsky, 1996.)

Staff Paper, that particulate matter canand does produce adverse effects onvisibility in various locations,depending on the PM concentrationsinvolved and other factors discussedbelow. It has been demonstrated thatimpairment of visibility is an importanteffect of PM on public welfare, and thatit is experienced throughout the UnitedStates, in multi-state regions, urbanareas, and remote mandatory Class IFederal areas47 alike. Visibility is animportant welfare effect because it hasdirect significance to people’senjoyment of daily activities in all partsof the country. Individuals value goodvisibility for the well-being it providesthem directly, both where they live andwork, and in places where they enjoyrecreational opportunities. Visibility ishighly valued in significant naturalareas, such as national parks andwilderness areas, because of the specialemphasis given to protecting these landsnow and for future generations. TheCriteria Document cites many studiesdesigned to quantify the benefitsassociated with improvements invisibility.

The Administrator consideredinformation from the Staff Paper andCriteria Document regarding the effectof the composition of particulate matteron visibility. Visibility conditions aredetermined by the scattering andabsorption of light by particles andgases, from both natural andanthropogenic sources. Visibility can bedescribed in terms of visual range, lightextinction, or deciview48. The classes offine particles principally responsible forvisibility impairment are sulfates,nitrates, organic matter, elementalcarbon (soot), and soil dust. Fineparticles are more efficient per unitmass at scattering light than coarseparticles. The scattering efficiency ofcertain classes of fine particles, such assulfates, nitrates, and some organics,increases as relative humidity rises

because these particles can absorb waterand grow to sizes comparable to thewavelength of visible light. In additionto limiting the distance that one can see,the scattering and absorption of lightcaused by air pollution can also degradethe color, clarity, and contrast of scenes.

The Administrator next consideredwhat would be an appropriate level fora secondary standard to address adverseeffects of particulate matter on visibility.The determination of a single nationallevel is complicated by regionaldifferences in visibility impairment dueto several factors, including backgroundand current levels of PM, compositionof particulate matter, and averagerelative humidity.

The Criteria Document and StaffPaper describe estimated backgroundlevels of PM and natural lightextinction. In the United States,estimated annual mean backgroundlevels of PM2.5 are significantly lower inthe West than in the East. Based onestimated background fine particle andlight extinction levels summarized inTable VIII-2 of the Staff Paper, naturallyoccurring visual range in the East isapproximately 105 to 195 kilometers,whereas in the West it is approximately190 to 270 kilometers. This significantregional difference in estimatedbackground conditions results from twomain factors. First, in the westernUnited States, visibility is moresensitive to an additional 1–2 µg/m3 ofPM2.5 in the atmosphere than in theeastern United States. Secondly, lightscattering is increased for certainparticles (e.g., sulfates, nitrates, andsome organics) due to higher averagerelative humidity in the East.

The combination of naturallyoccurring and manmade emissions alsoleads to significant differences incurrent visibility conditions betweenthe eastern United States, 23–39kilometers average visual range, andwestern United States, 55–150kilometers average visual range. TableVIII-4 of the Staff Paper indicates thatthe current level of annual average lightextinction in several western locations,such as the Colorado Plateau, is aboutequal to the level of background lightextinction, i.e., the level generallyregarded as representing the absence ofanthropogenic emissions in NorthAmerica, in the East. This regionaldifference is due to higher backgroundparticle concentrations in the East, acomposition of fine particles in the Eastthat, in association with higher easternhumidity levels, is more efficient atlight scattering, and significantly lowerconcentrations of anthropogenic PM inremote western locations as comparedwith remote eastern sites.

Because of these regional differences,it is the Administrator’s judgment thata national secondary standard intendedto maintain or improve visibilityconditions on the Colorado Plateau orother parts of the West would have tobe set at or even below naturalbackground levels in the East, whichwould effectively require elimination ofall eastern anthropogenic emissions.Conversely, a national secondarystandard that would achieve anappropriate degree of visibilityimprovement in the East would permitfurther degradation in the West. Due tothis regional variability in visibilityconditions created by differingbackground fine particle levels, fineparticle composition, and humidityeffects, the Administrator finds thataddressing visibility solely throughsetting more stringent nationalsecondary standards would not be anappropriate means to protect the publicwelfare from adverse impacts of PM onvisibility in all parts of the country.49

Aside from the problem of regionalvariability, the Administrator has alsodetermined that the Agency currentlylacks sufficient information to establisha level for a national secondary standardthat would represent a threshold abovewhich visibility conditions wouldalways be adverse and below whichvisibility conditions would always beacceptable. Because visibility varies notonly with PM concentration, but alsowith PM composition and humiditylevels, attaining even a lowconcentration of fine particles might ormight not provide adequate protection,depending on these factors.

The Administrator next assessedpotential visibility improvements50 thatwould result from attainment of the newprimary standards for PM2.5. Thespatially averaged form of the annualstandard is well suited to the protectionof visibility, which involves effects ofPM throughout an extended viewingdistance across an urban area. Indeed, as


51 IMPROVE (Interagency Monitoring ofPROtected Visual Environments) is a visibilitymonitoring network managed cooperatively by EPA,Federal land management agencies, and Staterepresentatives. An analysis of IMPROVE data for1992–1995 is found in Sisler et al. (1996).

the generally controlling standardfocused on reducing urban and regionalscale fine particle levels, most of thevisibility protection provided by thePM2.5 primary standards would bederived from the annual standard. Inmany cities having annual mean PM2.5

concentrations exceeding 17 µg/m3,improvements in annual averagevisibility resulting from attainment ofthe new annual PM2.5 primary standardare expected to be perceptible (i.e., toexceed 1 deciview). Based on annualmean PM2.5 data reported in Table 12-2 of the Criteria Document and Table V-12 in the Staff Paper, many cities in theNortheast, Midwest, and Southeast, aswell as Los Angeles, would be expectedto see perceptible improvement invisibility if the annual PM2.5 primarystandard is attained.

In Washington, DC, for example,where the IMPROVE network51 showsannual mean PM2.5 concentrations atabout 19 µg/m3 during 1992–1995,approximate annual average visibilitywould be expected to improve from 21km visual range (29 deciview) to 27 km(27 deciview). Annual average visibilityin Philadelphia, where annual PM2.5

levels have been recently measured at17 µg/m3, would be expected to changefrom about 24 to 27 km, animprovement of about 1 deciview. InLos Angeles, where recent data showsannual mean PM2.5 concentrations atapproximately 30 µg/m3, visibilitywould be expected to improve fromabout 19 to 34 km (30 to 24 deciview)if the new annual primary PM2.5

standard is attained.It is important to note that some urban

areas, many in the eastern United States,would be expected to have annual meanPM2.5 concentrations reduced below theprimary standard level of 15 µg/m3

when implementation of regionalcontrol strategies for PM and other airquality programs, such as thoseaddressing acid rain and mobilesources, are taken into account together.On the other hand, some urban areaswith annual PM2.5 levels at or below the15 µg/m3 level would be expected to seelittle, if any, improvement in annualaverage visibility. This may beparticularly true of certain westernurban areas that are dominated bycoarse rather than fine particles.

The Administrator also consideredthe potential effect on urban visibility ifthe 24-hour 98th percentile PM2.5

standard of 65 m3 is attained. In areas

with violations caused by localized hotspots, the 24-hour standard might havelittle effect other than on visible sourceemissions. In other areas, for example,with seasonally high woodsmoke, amore areawide improvement is possible.In such urban areas, attainment of the24-hour standard would be expected toreduce, to some degree, the number andintensity of ‘‘bad visibility’’ days, i.e.,the 20% of days having the greatestimpairment over the course of a year.For example, maximum 24-hour PM2.5

concentrations have been recorded inrecent years at over 140 µg/m3 at severalCalifornia locations. If the level andfrequency of peak PM concentrationsare reduced, improvements would beexpected in those days where visibilityis worst, even in urban areas havingannual averages below the annual PM2.5

primary standard.Having concluded that attainment of

the annual and 24-hour PM2.5 primarystandards would lead to visibilityimprovements in many eastern andsome western urban areas, theAdministrator also considered potentialimprovements to visibility on a regionalscale. In the rural East, attainment of thePM2.5 primary standards could result inregional visibility improvement, e.g., incertain mandatory Class I Federal areassuch as Shenandoah and Great SmokyMountains National Park, if regionalcontrol strategies are adopted andcarried out in order to reduce the impactof long-range transport of fine particlessuch as sulfates. Fine particle emissionreductions achieved by other air qualityprograms, such as those to reduce acidrain or mobile source emissions, are alsoexpected to improve Eastern regionalvisibility conditions (U.S. EPA, 1993).In the West, strategies to attain theprimary PM2.5 standards are less likelyto significantly improve visibility on aregional basis. However, areasdownwind from large urban areas, suchas Southern California, would likely seesome improvement in annual averagevisibility.

Based on the foregoing, theAdministrator concludes thatattainment of PM2.5 secondary standardsset at the level of the primary standardsfor PM2.5 would be expected to result invisibility improvements in the easternUnited States at both urban and regionalscales, but little or no change in thewestern United States except in andnear certain urban areas. Additionally,the Administrator determined thatattainment of secondary standardsequivalent to the suite of PM2.5 primarystandards for particulate matter wouldaddress some but not all of the effectsof particulate matter on visibility. The

extent to which these effects would beaddressed is expected to vary regionally.

The Administrator then consideredthe potential effectiveness of a regionalhaze program to address the remainingeffects of particulate matter on visibility(i.e., those that would not be addressedthrough attainment of secondarystandards identical to the suite of PM2.5

primary standards). A program toaddress the widespread, regionallyuniform type of haze caused by amultitude of sources is required bysections 169A and 169B of the Act. In1977, Congress established as a nationalgoal ‘‘the prevention of any future, andthe remedying of any existing,impairment of visibility in mandatoryClass I Federal areas which impairmentresults from manmade air pollution’’,section 169A(a)(1) of the Act. The EPAis required by section 169A(a)(4) of theAct to promulgate regulations to ensurethat ‘‘reasonable progress’’ is achievedtoward meeting the national goal. EPAoriginally deferred establishment of aprogram to address regional haze in1980 due to the need for greaterscientific and technical knowledge, butthe current Criteria Document and StaffPaper cite information supporting theAdministrator’s conclusion that thescientific state of understanding andanalytical tools are now adequate todevelop such a program. Becauseregional emission reductions are neededto make visibility improvements inmandatory Class I Federal areas, thestructure and requirements of sections169A and 169B of the Act, provide forvisibility protection programs that canbe more responsive to the factorscontributing to regional differences invisibility than can programs addressinga nationally applicable secondaryNAAQS. The visibility goal is moreprotective than a secondary NAAQSsince the goal addresses any man-madeimpairment rather than just impairmentat levels determined to be adverse.

Thus, an important factor consideredin this review is whether a regional hazeprogram, in conjunction with secondarystandards set identical to the suite ofPM2.5 primary standards, would provideappropriate protection for visibility innon-Class I areas. The Administratorcontinues to believe that the twoprograms and associated controlstrategies should provide suchprotection due to the regionalapproaches needed to manage emissionsof pollutants that impair visibility inmany of these areas. Regional strategiesimplemented to attain the NAAQS, meetother air program goals, and makereasonable progress toward the nationalvisibility goal in mandatory Class IFederal areas are expected to improve


52 EPA established the Grand Canyon VisibilityTransport Commission (GCVTC) in 1991 undersection 169B of the Act. Section 169B(d) requiresvisibility transport commissions to assess the‘‘adverse impacts on visibility from potential orprojected growth in emissions’’ and to recommendto EPA measures to remedy such adverse impacts.The Commission issued its final report in June1996.

53 The Southern Appalachian Mountain Initiativeis a voluntary effort begun in 1993. Participantsinclude eight southeastern States, Federal landmanagers, EPA, and representatives from industryand environmental groups. A final report has notbeen issued to date.

54 Indeed, Congress recognized when it adoptedsection 169A that the ‘‘visibility problem is causedprimarily by emission into the atmosphere of sulfurdioxide, oxides of nitrogen and particulate matter,especially fine particulate matter, frominadequately controlled sources.’’ H.R. Rep. No. 95–294 at 204 (1977).

visibility in many urban and non-ClassI areas as well. The followingrecommendation from the 1993 report ofthe National Research Council,Protecting Visibility in National Parksand Wilderness Areas, addresses thispoint:

Efforts to improve visibility in Class I areasalso would benefit visibility outside theseareas. Because most visibility impairment isregional in scale, the same haze that degradesvisibility within or looking out from anational park also degrades visibility outsideit. Class I areas cannot be regarded aspotential islands of clean air in a pollutedsea.

Before making a final decisions on thesecondary standards, the Administratoralso considered a number of publiccomments that addressed this aspect ofthe proposal. Some commenterssuggested setting secondary standardsfor PM2.5 more stringent than theproposed primary standards for thepurpose of addressing visibilityimpairment and other environmentaleffects. For the reasons discussed abovein this unit, however, the Administratorhas concluded that this may not be aneffective and would not be anappropriate means of protecting againstvisibility impairment in all parts of thecountry. Other commenters raised thepossibility of establishing a nationallyapplicable secondary standard definedas a ‘‘floor,’’ or increment, aboveregionally specific background levels ofPM2.5 or associated visibility. Althoughthis idea is of interest and may warrantfurther study, the Administratordetermined that it was not appropriateto pursue such an approach at this timefor two principal reasons. First, theAgency does not currently haveadequate scientific information toestablish a specific floor or incrementlevel that would protect against adverseeffects nationally, nor is it clear as aconceptual matter whether furtherinformation would support selection ofa single, uniform increment asproviding an appropriate degree ofprotection in all areas of the country.Second, there are serious, unresolvedquestions about whether such anapproach is consistent with thestatutory language and purposes ofsection 109 of the Act.

Other commenters argued thatnational secondary standards equivalentto the proposed PM2.5 primary standardsare not necessary or not supported bythe Administrator’s findings. As notedearlier, however, it is clear that coarseand fine particles can cause adverseeffects on visibility and significantquantitative data exist to demonstratethat visibility impairment occurs atsmall concentrations of PM2.5.

Substantial efforts have been put forthto assess the effects of PM on visibility.For example, the Grand CanyonVisibility Transport Commission52 spentseveral years and significant effortstudying the effects of pollution on 16mandatory Class I Federal areas on theColorado plateau and has maderecommendations to the Administratorfor actions to improve visibility in theseareas (GCVTC, 1996). All of themandatory Class I Federal areas studiedby the GCVTC with monitoring datahave annual mean PM2.5 concentrationsbelow 5 µg/m3 (Sisler, 1996) while alsodocumenting anthropogenic visibilityimpairment. The Southern AppalachianMountain Initiative53 is currentlyassessing air pollution impacts onvisibility, terrestrial resources, andaquatic resources in the southeasternU.S. in order to recommend measures toremedy existing and prevent futureadverse effects on these air qualityrelated values. The IMPROVE networkshows that all of the mandatory Class IFederal areas in the SAMI region haveannual mean PM2.5 concentrations for1992–95 between 11.0–13.5 µg/m3

(Sisler, 1996). The inclusion in section169A of the Act of a national visibilitygoal of no manmade impairment alsoplaces significant value on reducing PMconcentrations and resulting visibilityimpairment to low levels.54 Thedifferences between the fine particlelevels associated with visibilityimpairment in eastern and westernmandatory Class I Federal areas providefurther impetus to act under theprovisions of sections 169A and 169Benabling the Administrator to establisha regionally-tailored visibility programto address impairment of visibility inmandatory Class I Federal areas. Forthese reasons, the Administrator hasconcluded that a national regional hazeprogram allowing for regionalapproaches to addressing fine particlepollution, combined with a nationally

applicable level of protection achievedthrough secondary PM2.5 standards setequal to the suite of primary standards,would be more effective in addressingregional variations in the adverse effectsof PM2.5 on visibility than establishingnational secondary standards forparticulate matter that are lower thanthe suite of PM2.5 primary standards.The Administrator emphasizes that inorder to appropriately address theregional differences in adverse effects ofparticulate matter on visibility, it isessential to establish secondarystandards for PM2.5 equivalent to theprimary standards and an effective newregional haze program. A regional hazeprogram will be particularly importantin those areas of the country that do notexceed any of the primary standards forPM2.5, yet still experience significantvisibility impairment due to particulatematter. The EPA will propose a regionalhaze regulation in the near future.

In addition to providing a moreregionally tailored approach thanestablishing a more stringent nationalsecondary standard, an effectiveregional haze program will also fulfillthe Administrator’s regulatoryresponsibility under sections 169A and169B of the Act to address bothreasonably attributable impairment andregional haze impairment in mandatoryClass I Federal areas. Indeed, regionalhaze has been shown to be the principalcause of visibility impairment inmandatory Class I Federal areas today.Thus, the promulgation of a regionalhaze program in conjunction withsecondary standards for PM2.5

equivalent to the suite of primarystandards will serve as an appropriateapproach for addressing adverse effectsof visibility that vary regionally, and itwill also establish a comprehensiveprogram for making reasonable progresstoward the national visibility goal inmandatory Class I Federal areas byaddressing visibility impairment in theform of both source-specific impactsand regional haze. Further, the regionalhaze rulemaking will fulfill theAdministrator’s responsibilities toaddress the visibility protectionrecommendations of the Grand CanyonVisibility Transport Commission,pursuant to section 169B(e) of the Act.

The Administrator recognizes thatpeople living in certain urban areas mayplace a high value on unique scenicresources in or near these areas, and asa result might experience visibilityproblems attributable to sources thatwould not necessarily be addressed bythe combined effects of a regional hazeprogram and secondary standardsidentical to the suite of primarystandards for PM2.5. Commenters from


certain western cities and States raisedthis issue. In the Administrator’sjudgment, State or local regulatoryapproaches, such as past action inColorado to establish a local visibilitystandard for the City of Denver, wouldbe more appropriate and effective inaddressing these special situationsbecause of the localized and uniquecharacteristics of the problems involved.Visibility in an urban area located neara mandatory Class I Federal area canalso be improved through Stateimplementation of the current visibilityregulations, by which emissionlimitations can be imposed on a sourceor group of sources found to becontributing to ‘‘reasonablyattributable’’ impairment in themandatory Class I Federal area. EPAalso intends to pursue opportunities toobtain information on urban and non-Class I area visibility throughexamination of available fine particlemonitoring data. Current or plannedmonitoring networks and initiatives,such as monitoring and chemicalanalysis of PM2.5 in urban andbackground sites, efforts to bettercharacterize real-time environmentalconditions in major populations centers,and new automated airport visibilitymonitoring networks should providedata needed to evaluate trends in theseareas. This information should help tobetter characterize the nature andspatial extent of urban and non-Class Ivisibility problems and thus serve toinform future decisions on NAAQSrevisions or other appropriate measures.

Based on all of the considerationsdiscussed, the Administrator hasdecided to establish secondarystandards identical to the suite of PM2.5

primary standards, in conjunction witha regional haze program under sections169A and 169B of the Act, as the mostappropriate and effective means ofaddressing the welfare effects associatedwith visibility impairment. Together,the two programs and associated controlstrategies should provide appropriateprotection against the effects of PM onvisibility and enable all regions of thecountry to make reasonable progresstoward the national visibility goal.

2. Materials damage and soilingeffects. Annual and 24-hour secondarystandards for materials damage andsoiling effects of PM were established in1987 at levels equal in all respects to theprimary standards. As discussed in theCriteria Document and Staff Paper,particles affect materials by promotingand accelerating the corrosion of metals,by degrading paints, and bydeteriorating building materials such asconcrete and limestone. Soiling is foundto reduce the aesthetic quality of

buildings and objects of historical orsocial interest. Past studies have foundthat residential properties in highlypolluted areas typically have lowervalues than those in less polluted areas.Thus, at high enough concentrations,particles become a nuisance and resultin increased cost and decreasedenjoyment of the environment.

In the proposal, EPA proposed toestablish secondary standards for PM10

and PM2.5 identical to the suite ofproposed primary standards. Severalcomments recommended settingsecondary standards at levels morestringent than the proposed primarystandards in order to address variouswelfare effects of PM, including soilingand materials damage, acid deposition,and visibility. Some commentersspecifically suggested changing the formor level of the proposed 24-hour, 98thpercentile PM standards to betterprotect against elevated PM episodesand associated soiling, materialsdamage, and visibility effects.

After reviewing the extent of relevantstudies and other information providedsince the 1987 review of the PMstandards, the Administrator concurswith staff and CASAC conclusions thatthe available data do not provide asufficient basis for establishing aseparate secondary standard based onsoiling or materials damage alone. In theAdministrator’s judgment, however,setting secondary standards identical tothe suite of PM2.5 and PM10 primarystandards would provide increasedprotection against the effects of fineparticles and retain an appropriatedegree of control on coarse particles.Accordingly, the Administratorestablishes the secondary standards forPM2.5 identical to the suite of primarystandards to protect against materialsdamage and soiling effects of PM.

B. Decision on the Secondary StandardsThe Administrator establishes

secondary standards identical to thesuite of primary standards. In theAdministrator’s judgment, theestablishment of these standards, inconjunction with implementation of aregional haze program, will provideappropriate protection against thewelfare effects associated with particlepollution.

IV. Other IssuesCommenters have raised a number of

legal and procedural issues that arediscussed in this unit. These include:

(1) Whether EPA must giveconsideration to costs and similarfactors in setting NAAQS.

(2) Whether EPA erred in its selectionof a methodology for determining the

level of a NAAQS that protects publichealth with an adequate margin ofsafety.

(3) Whether EPA committed aprocedural error by not entering into therulemaking docket underlying data fromcertain epidemiological studies.

(4) Whether the 1990 amendments tothe Act preclude EPA from revising thePM NAAQS to establish a new PM2.5

indicator.Responses to other legal and proceduralissues are included in the Response-to-Comments Document.

A. Consideration of CostsFor more than a quarter of a century,

EPA has interpreted section 109 of theAct as precluding consideration of theeconomic costs or technical feasibilityof implementing NAAQS in settingthem. As indicated in the proposal, anumber of judicial decisions haveconfirmed this interpretation. NaturalResources Defense Council v.Administrator, 902 F.2d 962, 972–973(D.C. Cir. 1990)(PM NAAQS)(‘‘PM10’’),vacated, in part, dismissed, 921 F.2d326 (D.C. Cir.), certs. dismissed, 498U.S. 1075, and cert. denied, 498 U.S.1082 (1991); Natural Resources DefenseCouncil v. EPA, 824 F.2d 1146, 1157–1159 (D.C. Cir. 1987)(en banc)(CAAsection 112 standards for vinylchloride)(‘‘Vinyl Chloride’’); AmericanPetroleum Institute v. Costle, 665 F.2d1176, 1185–1186 (D.C. Cir. 1981)(ozoneNAAQS)(‘‘Ozone’’), cert. denied, 455U.S. 1034 (1982); Lead Industries Ass’nv. EPA, 647 F.2d 1130, 1148–1151 (D.C.Cir.)(lead NAAQS)(Lead Industries),cert. denied, 449 U.S. 1042 (1980).

Some commenters have argued thatcosts and similar factors should,nonetheless, be considered, both in thisrulemaking and in the rulemaking onproposed revisions to the NAAQS forozone. Although most of thecommenters’ arguments are inconsistentwith the judicial decisions cited in thisunit, several commenters have arguedthat those decisions are not dispositive.For reasons discussed in this unit andin the Response-to-CommentsDocument, EPA disagrees with thesecomments and maintains itslongstanding interpretation of the Act asprecluding consideration of costs andsimilar factors in setting NAAQS.

1. Background. Given the nature ofthe points raised, a brief review of theissue seems useful before addressing thecomments. The requirement that EPAestablish national ambient air qualitystandards for certain pollutants, to beimplemented by the States, was enactedin 1970 as part of a set ofcomprehensive amendments thatestablished the basic framework for


55 36 FR 8186, April 30, 1971. EPA hasmaintained this interpretation consistently sincethen.

56 That consideration of such factors was notintended in NAAQS decisions is also supported bysection 109(a)(1) of the Act. For pollutants forwhich air quality criteria had been issued prior tothe 1970 amendments, that provision required EPAto propose NAAQS within 30 days after enactmentand to take final action 90 days later. The criteriaissued previously did not include information oncosts and similar factors, and it would have beendifficult if not impossible for EPA to supplementthem in time to include meaningful considerationof such factors in NAAQS proposed 30 days afterenactment.

57 See, e.g., sections 110(e)(1), 111(a)(1), 231(b) ofthe 1970 Act; see also, e.g., sections 113(d)(4)(C)(ii),125(a)(3), 202(a)(3)(C), 317 of the 1977 Act.

58 Union Electric Co. v. EPA, 427 U.S. 246, 257–58 (1976).

59 The Senate report on the 1970 amendmentsstated: ‘‘In the Committee discussions, considerableconcern was expressed regarding the use of theconcept of technical feasibility as the basis ofambient air standards. The Committee determinedthat (1) the health of people is more important thanthe question of whether the early achievement ofambient air quality standards protective of health istechnically feasible; and, (2) the growth of pollutionload in many areas, even with application ofavailable technology, would still be deleterious topublic health.’’

‘‘Therefore, the Committee determined thatexisting sources of pollutants either should meetthe standard of the law or be closed down * * *.’’

S. Rep. No. 91–1196, at 2–3 (1970).60 These limitations would, of course, make little

sense if such factors could be considered in settingthe NAAQS themselves.

61 Such requirements ‘‘are expressly designed toforce regulated sources to develop pollution controldevices that might at the time appear to beeconomically or technologically infeasible.’’’ Id.(quoting Union Electric Co. v. EPA, 427 U.S. at 257).

62 In the PM10 case, for example, the Courtconsidered an argument that EPA should haveconsidered potential health consequences ofunemployment that might result from revision ofthe primary NAAQS for PM:

‘‘This claim is entirely without merit. In threeprevious cases, this court has emphatically statedthat § 109 does not permit EPA to consider suchcosts in promulgating national ambient air qualitystandards * * * . It is only health effects relatingto pollutants in the air that EPA may consider * ** . Consideration of costs associated with allegedhealth risks from unemployment would be flatlyinconsistent with the statute, legislative history andcase law on this point.’’

902 F.2d at 973 (emphasis in original; citationsomitted).

Federal, State, and local air pollutioncontrol. When EPA promulgated theoriginal NAAQS in 1971, its firstAdministrator, William D. Ruckelshaus,concluded that costs and similar factorscould not be considered in thatdecision.55 This conclusion was notchallenged in litigation on the originalNAAQS. It has been confirmed sincethen, however, by every judicialdecision that has considered the issue.

As discussed in this unit, EPA’sinterpretation rests primarily on thelanguage, structure, and legislativehistory of the statutory scheme adoptedin 1970. It is also supported by thejudicial decisions cited in this unit, aswell as by legislative developmentssince 1970 that reaffirm Congress’original approach to the issue.

Without cataloguing all relevantaspects of the 1970 amendments andtheir legislative history, several basicpoints should be noted. Under section109(b) of the Act, NAAQS are to be‘‘based on’’ the air quality criteria issuedunder section 108 of the Act. Undersection 108(a)(2) of the Act, the kind ofinformation EPA is required to includein criteria documents is limited toinformation about health and welfareeffects ‘‘which may be expected fromthe presence of [a] pollutant in theambient air * * * .’’ There is no mentionof the costs or difficulty ofimplementing the NAAQS, nor of‘‘effects’’ that might result fromimplementing the NAAQS (as opposedto effects of pollution in the air).56 Bycontrast, Congress explicitly providedfor consideration of costs and similarfactors in decisions under other sectionsof the Act.57 Moreover, States werepermitted to consider economic andtechnological feasibility in developingplans to implement the NAAQS to theextent such consideration did notinterfere with meeting statutorydeadlines for attainment of thestandards.58 Finally, the legislativehistory indicated that Congress had

considered the issue and haddeliberately chosen to mandate NAAQSthat would protect health regardless ofconcerns about feasibility.59

The first judicial decision on the issuecame in the Lead Industries case. Anindustry petitioner argued that EPAshould have considered economic andtechnological feasibility in allowing a‘‘margin of safety’’ in setting primarystandards for lead. Based on a detailedreview of the language, structure, andlegislative history of the statutoryscheme, the U.S. Court of Appeals forthe District of Columbia Circuitconcluded that:

This argument is totally without merit.[The petitioner] is unable to point toanything in either the language of the Act orits legislative history that offers any supportfor its claim * * * . To the contrary, thestatute and its legislative history make clearthat economic considerations play no part inthe promulgation of ambient air qualitystandards under section 109.

647 F.2d at 1148.The Court cited a number of reasons

for this conclusion. Id. at 1148–1150.Among other things, it noted thecontrast between section 109(b) of theAct and other provisions in whichCongress had explicitly provided forconsideration of economic andtechnological feasibility, as well as therequirement that NAAQS be based onair quality criteria defined withoutreference to such factors. Id. at 1148–1149 and n.37. The Court also notedthat, in developing plans to implementNAAQS, States may consider economicand technological feasibility only to theextent that this does not interfere withmeeting the statutory deadlines forattainment of the standards; and thatEPA may not consider such factors at allin deciding whether to approve Stateimplementation plans. Id. at 1149 n.37(citing Union Electric Co. v. EPA, 427U.S. 246, 257–258, 266 (1976)).60

As to the legislative history of the1970 amendments, the Court observedthat:

[T]he absence of any provision requiringconsideration of these factors was noaccident; it was the result of a deliberatedecision by Congress to subordinate suchconcerns to the achievement of health goals.

Id. at 1149. Citing several leadingSupreme Court decisions, as well as theSenate report quoted in this unit, theCourt noted that Congress had intendeda drastic change in approach toward thecontrol of air pollution in the 1970amendments and was well aware thatsections 108–110 of the Act imposedrequirements of a ‘‘technology-forcing’’character. Id.61

The Court also noted that Congresshad already acted, in furtheramendments adopted in 1977, to relievesome of the burdens imposed by the1970 amendments. Id. at 1150 n.38.Observing that Congress had, however,declined to amend section 109(b) of theAct to provide for consideration of costsand similar factors as requested byindustrial interests, Id. n.39, the Courtconcluded:

A policy choice such as this is one whichonly Congress, not the courts and not EPA,can make. Indeed, the debates on the [1970amendments] indicate that Congress wasquite conscious of this fact * * * .

* * * [I]f there is a problem with theeconomic or technological feasibility of thelead standards, [the petitioner], or any otherparty affected by the standards, must take itscase to Congress, the only institution withthe authority to remedy the problem.

Id. at 1150.After the decision in Lead Industries,

Supreme Court review was sought onthe question whether costs and similarfactors could be considered in settingNAAQS, among other issues. TheSupreme Court declined to review thedecision. Lead Industries Ass’n v. EPA,449 U.S. 1042 (1980). The subsequentdecisions in Ozone, Vinyl Chloride, andPM10, cited in this unit, stronglyreaffirmed the interpretation adopted inLead Industries.62 Supreme Court


63 See, e.g., H.R. Rep. No. 95-294, at 207–217(l977).

64 See, e.g., Id. at 110–112; Id. at 43-51.65 Section 109(d)(2)(C)(iv) of the Act. Some

commenters have argued that this provisionrequires EPA to consider such effects in settingNAAQS. From the language and structure of section109(d) of the Act, however, it is clear that CASAC’sresponsibility to advise on these factors is separatefrom its responsibility to review and recommendrevision of air quality criteria and NAAQS, and thatthe advice pertains to the implementation ofNAAQS rather than to setting them. The legislativehistory confirms this view, indicating that theadvice was intended for the benefit of the Statesand Congress. See H.R. Rep. No. 95-294, at 183(1977).

66 The 1977 amendments also required EPA toprepare economic impact assessments for specifiedactions but limited the requirement to non-health-based standards, excluding decisions undersections 109 and 112 of the Act. Section 317; H.R.Rep. No. 95-294, at 51–52 (1977). In this and other

respects, Congress continued the approach it tookin the l970 amendments, making careful choices asto when consideration of costs and similar factorswould be required and giving paramount priority toprotection of health. See 123 Cong. Rec. H8993(daily ed. Aug. 4, 1977) (Clean Air ConferenceReport (1977); Statement of Intent; Clarification ofSelect Provisions), reprinted in 3 Senate Committeeon Environment and Public Works, 95th Cong., ALegislative History of the Clean Air ActAmendments of 1977, at 319 (1978).

67 In the interim, the National Commission on AirQuality had also submitted its report to Congress asrequired by a provision of the 1977 amendments.Among other things, the Commission recommendedthat the statutory approach of requiring NAAQS tobe set at levels necessary to protect public health,without consideration of economic factors, becontinued without change. National Commission onAir Quality, To Breathe Clean Air 55 (1981).

68 As the Administrator indicated in EPA’sproposal to revise the PM standards:

‘‘[T]hat review has revealed a highly limited database—particularly where quantitative studies areconcerned—and a wide range of views amongqualified professionals about the exact pollutionlevels at which health effects are likely to occur.The setting of an ‘adequate margin of safety’ belowthese levels calls for a further judgment—in an areafor which the scientific data base is even moresparse and uncertain * * * .’’

‘‘* * * [L]ong and expert review of public healthissues has to date revealed no scientific method ofassessing exactly what level of standards publichealth requires. The scientific review indicatessubstantial uncertainties concerning the health risksassociated with lower levels of particulate matter.’’(49 FR 10408, 10409, March 20, l984)

69 Congress was clearly aware of the 1987decision to revise the PM NAAQS, which amongother things involved changing the indicator forparticulate matter from ‘‘total suspendedparticulate’’ to PM10, because it enacted specialnonattainment provisions, as well as provisions forPSD increments, applicable to PM10. Sections 188–190 of the Act; section 166(f) of the Act. It wasclearly aware of the Vinyl Chloride decisionbecause it amended section 112 of the Act inresponse to that decision, essentially creating a newscheme for setting emission standards for hazardouspollutants.

70 H.R. Rep. No. 101–490, pt. 1, at 145 (1990). Seealso S. Rep. No. 101–228, at 5 (1989).

71 Additional responses to points raised by thiscommenter and others are included, as appropriate,in the Response-to-Comments Document.

72 Several other commenters argue that the citeddecisions are not dispositive because they held onlythat EPA is not required to consider costs andsimilar factors in setting NAAQS. As discussed inthis unit in connection with Chevron, however, thedecisions clearly concluded that Congress intendedto preclude consideration of such factors, and thatEPA is not free to alter that congressional choice.Although these conclusions are technically dicta,nothing in the Court’s opinions suggests that itwould have interpreted section 109 of the Actdifferently had EPA claimed authority to considercosts and similar factors in NAAQS decisions.Indeed, the tone of the opinions argues to thecontrary. See, e.g., PM10, 902 F.2d at 973. Cf. EthylCorp. v. EPA, 51 F.3d 1053 (D.C. Cir. 1995).

review of the Ozone and PM10 decisionswas sought but denied. AmericanPetroleum Institute v. Gorsuch, 455 U.S.1034 (1984); American Iron and SteelInstitute v. EPA, 498 U.S. 1082 (1991).

The Lead Industries opinion focusedlargely, though not exclusively, on the1970 amendments and their legislativehistory. Perhaps as a result, it did notcanvass all the factors that, in fact,supported its conclusions at the time.For example, when Congress enactedmajor amendments to the Act in 1977,it was clearly aware that some areas ofthe country had experienced difficultyin attempting to attain some of theNAAQS.63 It was also aware that theremight be no health-effects thresholds forthe pollutants involved, and thatsignificant uncertainties are inherent insetting health-based standards under theAct.64 In response, Congress madesignificant changes in the provisions forimplementation of the NAAQS,including changes intended to ease theburdens of attainment. It also amendedsections 108 and 109 of the Act inseveral ways; for example, by requiringperiodic review and, if appropriate,revision of air quality criteria andNAAQS and by establishing a specialscientific advisory committee (CASAC)to advise EPA on such reviews. Notably,Congress recognized thatimplementation of NAAQS could cause‘‘adverse public health, welfare, social,economic, or energy effects’’ andcharged CASAC with advising EPA onsuch matters.65 Yet it made no changesin section 109(b) or section 108(a)(2) ofthe Act; that is, in the substantivecriteria for setting or revising NAAQS.In other words, Congress chose toaddress economic and other difficultiesassociated with attainment of theNAAQS by adjusting the scheme fortheir implementation, rather than bychanging the instructions for settingthem.66

Congress enacted major amendmentsto the Act again in 1990, well after theLead Industries and Ozone decisionsthat interpreted section 109 of the Actas precluding consideration of costs inNAAQS decisions.67 In doing so,Congress was clearly aware ofintervening developments such as EPA’sdecision to revise the PM NAAQS in1987—the result of an elaborate reviewin which the Administrator stronglyunderscored the scientific uncertaintiesinvolved68—and the Vinyl Chloride casedrawing a sharp distinction betweensections 109 and 112 of the Act withregard to consideration of costs andsimilar factors.69 Indeed, the legislativehistory of the 1990 amendments reflectsCongress’ understanding that primaryNAAQS were to be based on protectionof health ‘‘without regard to theeconomic or technical feasibility ofattainment.’’70 Again, however,Congress chose to respond to severe,

widespread, and persistent problemswith attaining the NAAQS by adjustingthe scheme for their implementationrather than by changing the basis forsetting them. See, e.g., sections 181–192of the Act.

2. Public comments. As notedpreviously, a number of commentershave argued that costs and similarfactors should be considered in EPA’sfinal decisions on revision of both theparticulate and ozone NAAQS. Asidefrom arguments that are simplyinconsistent with the judicial decisionscited in this unit, some of thecommenters argue that those decisionsare not dispositive for a variety ofreasons. One commenter submitted aparticularly comprehensive version ofthis argument; the following discussionfocuses primarily on points raised bythat commenter, among others.71

As a general matter, the commenteracknowledges that Congress intended topreclude consideration of economiccosts and similar factors in settingNAAQS. The commenter argues,however, that this is so only when thescientific basis for NAAQS is ‘‘clear andcompelling’’ or ‘‘unambiguous.’’ Fromthat premise, the commenter advancesthree key assertions:

a. Where non-threshold pollutants areinvolved and the health evidence isambiguous, section 109 of the Act mustbe interpreted to allow consideration ofall relevant factors, including thepractical consequences of EPA’sdecisions.

b. To the extent the judicial decisionscited in this unit are read as precludingthis, they rest on a faulty analysis thatpre-dates and cannot survive scrutinyunder Chevron, U.S.A. v. NaturalResources Defense Council, 467 U.S.837 (1984).72

c. Because EPA has discretion toconsider costs and similar factors wherethe health evidence is ambiguous, itmust do so in light of Executive Order12866 (58 FR 51735, October 4, 1993),and two recent statutes, the Unfunded


73 See, e.g., Lead Industries, 647 F.2d at 1146–1147, 1153–1156, 1160–1161, 1167 n.106. Inenacting the 1970 amendments, Congress was awarethat there were gaps in the scientific informationavailable then as a basis for establishing the originalNAAQS. See, e.g., S. Rep. No. 91-1196, at 9–11(1970). If anything, Congress had an even greaterunderstanding of the point when it enacted the1977 amendments without changing the substantivecriteria for setting NAAQS. See H.R. Rep. No. 95-294, at 43–51, 181–182 (1977).

74 Lead Industries, 647 F.2d at 1147 (quotingEthyl Corp. v. EPA, 541 F.2d 1, 24–27 (D.C. Cir.)(en banc), cert. denied, 426 U.S. 941 (1976)).

75 They may have methodological flaws, forexample, but nonetheless report effects that are ofserious medical significance; or they may be ofimpeccable quality but involve effects of uncertainsignificance. Others may involve results that arestriking but hard to explain in terms of previousknowledge, or results that seem plausible andimportant but are not yet replicated by otherstudies.

76 See, e.g., Lead Industries, 647 F.2d at 1155–1156; H.R. Rep. No. 94-295, at 43–51 (1977).

77 As previously discussed, the Administratorstrongly emphasized the uncertainties involved inthat review. As a result of the uncertainties, heproposed ‘‘relatively broad’’ ranges for comment,though he focused on lower levels within the rangesas providing greater margins of safety against thehealth risks involved. See 49 FR 10408, 10409,March 20, l984.

78 See, e.g., Lead Industries, 647 F.2d at 1152–53and n. 43, 1159–60; Ozone, 665 F.2d at 1185, 1187;PM10, 902 F.2d at 969–71, 972.

79 Indeed, the present decisions on the NAAQSfor PM and ozone are based on some of the bestscientific information the Agency has ever beenable to rely on in NAAQS decision-making. Inparticular, the science underlying these decisions ismuch more extensive and of much better qualitythan the science underlying the existing NAAQS forPM and ozone.

80 In practice, analysis of this question issometimes referred to as a ‘‘Chevron step one’’analysis.

81 See, e.g., 647 F.2d at 1148–51, 1152–53 andn.43, 1160–61.

Mandate Reform Act of 1995, 2 U.S.C.1501–1571 (UMRA), and the SmallBusiness Regulatory EnforcementFairness Act of 1996, Pub. L. 104–121,110 Stat. 857 (SBREFA), which in partamended the Regulatory Flexibility Act,5 U.S.C. 601–808.

EPA believes all three assertions areclearly incorrect. Regarding the firstpoint, it should be evident, both fromprevious NAAQS decisions and fromthe court opinions upholding them, thatthe scientific basis for NAAQS decisionshas never pointed clearly andunambiguously to a single ‘‘rightanswer.’’73 This is inherent in thestatutory scheme for the establishmentand revision of NAAQS, which in effectrequires them to be based on the ‘‘latestscientific knowledge’’ on potentialhealth and welfare effects of thepollutant in question. See sections109(b) and 108(a)(2) of the Act.Although advances in science increaseour understanding of such effects, theyalso raise new questions. For thisreason, the key studies for any givendecision on revision of a NAAQS are,almost by definition, ‘‘at the very‘frontiers of scientific knowledge.’’’73

That is, studies that call into questionthe adequacy of a standard are alwaysthose that go beyond previous studies—by reporting new kinds of effects, forexample, or effects at lowerconcentrations than those at whicheffects have been reported previously.

As with pioneering work in otherfields, such studies may have a varietyof strengths and limitations.875 As aresult, the validity and implications ofsuch studies may be both uncertain andhighly controversial. Given theprecautionary nature of section 109 ofthe Act,76 however, it is precisely thesekinds of studies that the Administratormust grapple with when advances in

science suggest that revision of aNAAQS is appropriate.

As a result, the EPA staff typicallyrecommends for consideration, and theAdministrator may propose forcomment, a range of alternatives basedon what the commenter would call‘‘ambiguous’’ science. In this respect,the current reviews of the NAAQS forozone and particulate matter are notunusual and do not differ, for example,from the review that led to adoption ofthe PM10 NAAQS in 1987.77 Indeed, theNAAQS that were upheld in the LeadIndustries, Ozone, and PM10 decisionswere all based on highly controversialhealth evidence; the Lead Industriesdecision took note of congressionalstatements recognizing that there maybe no thresholds for criteria pollutants;and the Ozone and PM10 decisionsnoted the Administrator’s findings thatclear thresholds could not be identifiedfor ozone and particulate matter,respectively.78 Thus, the presentdecisions on revision of the NAAQS forozone and particulate matter cannot bedistinguished from those past decisionsin terms of the nature of the healthevidence or pollutants involved.78

Regarding the second of thecommenter’s key assertions, EPAdetermines it is clear that the judicialdecisions cited in this unit werecorrectly decided and continue to begood law under Chevron. In Chevron,the Supreme Court essentiallyreaffirmed the principle that courtsmust defer to reasonable agencyinterpretations of the statutes theyadminister where Congress hasdelegated authority to them to elucidateparticular statutory provisions. Wherethe intent of Congress on an issue isclear, however, it must be given effectby the agency and the courts. See 467U.S. at 842–45. Thus, the first questionon review of an agency’s interpretationunder Chevron is ‘‘whether Congresshas directly spoken to the precisequestion at issue.’’ If the courtdetermines that it has not, the remaining

question for the court is ‘‘whether theagency’s answer is based on apermissible construction of the statute.’’467 U.S. at 842–843 (footnote omitted).In determining whether Congress ‘‘hadan intention on the precise question atissue,’’ a court employs ‘‘traditionaltools of statutory construction.’’ Id. at843 n.9.80

In essence, the commenter’s argumenthere is that the Lead Industries decisiondid not address whether Congress had‘‘spoken directly’’ to the precise issueposed by the commenter; that is,whether section 109 of the Act must beinterpreted differently for NAAQSdecisions involving non-thresholdpollutants and ‘‘ambiguous’’ healthevidence. The Lead Industries opinion,which pre-dated Chevron, did not posethe question in those terms. Its focus,however, was clearly on what Congressintended to be the basis for NAAQSdecisions, in a context the Courtunderstood to involve considerableuncertainty and debate about the healthevidence, as well as the possibility thatthere was no threshold for health effectsof the pollutant.81 In short, the healthevidence was hardly ‘‘unambiguous,’’yet the Court interpreted section 109 ofthis Act as precluding consideration ofcosts and similar factors even inallowing a margin of safety. Nothing inthe Lead Industries decision or in thesubsequent cases suggests in any waythat section 109 of the Act should beinterpreted differently based on thenature of the pollutants or healthevidence involved, and the Court’sfindings on congressional intent admitof no exceptions:

* * * [T]he statute and its legislativehistory make clear that economicconsiderations play no part in thepromulgation of ambient air qualitystandards under Section 109.

647 F.2d at 1148.Alternatively, the commenter argues

that the Lead Industries case decidedthe issue incorrectly in light of theprinciples announced subsequently inChevron. In this context, the commenteressentially argues that the LeadIndustries decision rested on two factorsthat are no longer probative:

(1) That there was no indication thatCongress meant to allow considerationof costs in NAAQS decisions, and

(2) That Congress specificallyprovided for such consideration in othersections of the Act but not in section109.


82 See 647 F.2d at 1148–51. By contrast, thecommenter’s argument that Congress actuallyintended EPA to consider such factors relies heavilyon statements made in subsequent legislativehistory, most of which were made in floor debate,that sought to justify controversial amendments toestablish a different program than the NAAQS anddid not involve any proposed changes in section109 of the Act or related provisions; and statementsin early judicial decisions involving programsunder other statutory provisions. In context, EPAdetermines these and other statements cited by thecommenter are consistent with and do not alter theconclusion that Congress intended to precludeconsideration of costs and similar factors undersection 109 of the Act.

83 The commenter argues that the post-Chevroncases accepted the Lead Industries analysisuncritically rather than re-examining it underChevron. Clearly, this elevates form over substance.It is true that neither case referred to Chevron indiscussing the point at issue. In Vinyl Chloride,however, the Court retraced the steps in the LeadIndustries analysis in some detail, characterizedsome of the key evidence reviewed in that analysisin terms going beyond mere rote repetition (e.g., ‘‘afar clearer statement than anything in the presentcase that Congress considered the alternatives’’),and used Chevron-like language in discussing thesignificance of that evidence; that is, that itdemonstrated congressional intention on the pointat issue. E.g., 824 F.2d at 1159. Given that the VinylChloride case was decided three years afterChevron, that it was an en banc decision of the D.C.Circuit involving interpretation of statutorylanguage very similar to that in Lead Industries, andthat the Court cited Chevron twice in analyzing thelanguage and history of section 112 of the Act, itseems highly unlikely that the Court was unmindfulof Chevron principles in concluding that Congressintended to preclude consideration of costs undersection 109 of the Act but not under section 112 ofthe Act.

In the PM10 decision, the Court confirmed thesharp distinction it had drawn, based on suchevidence of congressional intent, between sections109 and 112 of the Act in Vinyl Chloride. 902 F.2dat 972–973. Although discussion of the point wasbrief and did not mention Chevron, the industrypetitioner raising the point had cited Chevron inarguing that the Lead Industries interpretation wasnot binding, and that EPA’s decision on the PM10

standards should be reversed on the ground that itrested on a legal position that EPA unjustifiablybelieved was mandated by Congress. Reply Brief ofthe American Iron and Steel Institute at 11 andn.10, Natural Resources Defense Council v.Administrator, 902 F.2d 962 (D.C. Cir. 1990) (Nos.87-1438 et al.). Thus, Chevron issues were properlybefore the Court and were brought squarely to itsattention.

84 See also 52 FR 24854, July 1, 1987.

On the first point, the commenterargues that EPA is free under Chevronto consider costs and similar factors (byreinterpreting section 109 of the Act)unless there is evidence that Congressintended to restrict its discretion. As tothe second point, the commenter arguesthat similar reasoning was rejected inVinyl Chloride.

In Vinyl Chloride, however, an enbanc decision that post-dated Chevron,the Court essentially underscored thepoint that such issues cannot be decidedmechanically but must turn, instead, onmore analytical attention to relevantindicia of congressional intent. See, e.g.,824 F.2d at 1157 n.4; Id. at 1157–1163.With reference to NAAQS decisions inparticular, the Court concluded thatthere were concrete indications ofcongressional intent to precludeconsideration of costs and similarfactors; for example, the fact that section108 of the Act ‘‘enumerate[s] specificfactors to consider and pointedlyexclude[s] feasibility.’’ 824 F.2d at 1159.In a later case, moreover, the same Courtheld that EPA could not considercertain factors, in decisions undersection 211(f)(4) of the Act, for reasonsexactly parallel to those that thecommenter criticizes in Lead Industries.See Ethyl Corp. v. EPA, 51 F.3d 1053,1057–1063 (D.C. Cir. 1995).

Beyond this, the commenter’scharacterization of the Lead Industriesdecision ignores or discounts much ofthe key evidence cited by the Court,including the language, structure, andlegislative history of the statutoryscheme established in 1970, for itsconclusion that Congress intended topreclude consideration of costs andsimilar factors in NAAQS decisions.82

As indicated in this unit, the VinylChloride and PM10 cases, both of whichpost-dated Chevron, reached the sameconclusion.

Moreover, this series of decisionswent far beyond mere deference to anagency interpretation. As indicated inthe Vinyl Chloride case, the LeadIndustries court found ‘‘clear evidence’’of Congressional intent, which was tolimit the factors EPA may consider

under section 109 of the Act. 824 F.2d1159. Consistent with Chevron, thesefindings were based on traditional toolsof statutory construction. See Id. at1157–1159; Lead Industries, 647 F.2d at1148–1151. In terms of the analyticalframework later established by Chevron,these were Chevron step one findings,meaning that the statute spoke directlyto the issue and that the courts, as wellas the agency, must give effect toCongress’ intent as so ascertained. See467 U.S. at 842–843.83 Thus, absent amore recent legislative enactmentoverriding that intent, EPA has nodiscretion to alter its longstandinginterpretation that consideration of costsand similar factors is precluded inNAAQS decisions under section 109 ofthe Act.84

As to the commenter’s third keyassertion, Executive Order 12866,UMRA sections 202 and 205, and theRegulatory Flexibility Act (RFA), asamended by SBREFA, do not conflictwith this interpretation or require adifferent result. Basically, thecommenter argues that the ExecutiveOrder, UMRA, and the RFA (asamended by SBREFA) require agenciesto use cost (or similar factors) as a

decisional criterion in makingregulatory decisions, and that thismodifies the Clean Air Act’s directivethat EPA is precluded from consideringcosts when setting a NAAQS. Thecommenter’s argument is flawed on anumber of grounds. First, UMRA andthe RFA (as amended by SBREFA) donot conflict with section 109 of the Actbecause they do not apply to thisdecision, as discussed in Unit VIII. ofthis preamble. Second, the ExecutiveOrder and both statutes are quite clearthat they do not override the substantiveprovisions in an authorizing statute.Third, the commenter’s premise thatUMRA and the RFA (as amended bySBREFA) establish substantivedecisional criteria that agencies arerequired to follow is wrong.

As a matter of law, the ExecutiveOrder cannot (and does not purport to)override the Clean Air Act. TheExecutive Order does not conflict withsection 109 of the Act because therequirement that agencies ‘‘selectapproaches that maximize net benefits’’does not apply if a ‘‘statute requiresanother regulatory approach.’’ ExecutiveOrder 12866, section (1)(a), (58 FR51735, October 4, 1993). More generally,the Executive Order provides thatagencies are to adhere to its regulatoryprinciples only ‘‘to the extent permittedby law.’’ Id., section (1)(b).

UMRA sections 202 and 205 do notapply to this decision, as discussed inUnit VIII. of this preamble. Even whenthey do apply to a regulatory action,they do not establish decisional criteriathat an agency must follow, much lessoverride decisional criteria establishedin the statute authorizing the regulatoryaction. UMRA does not require anagency to select any particularalternative. Rather, an agency can selectan alternative that is not the least costly,most cost-effective or least burdensomeif the agency explains why. Section205(b)(1) of UMRA. Such anexplanation is not required if the leastcostly, most cost-effective or leastburdensome alternative would havebeen ‘‘inconsistent with law,’’ section205(b)(2) of UMRA, and the onlyalternatives that an agency shouldconsider are ones that ‘‘achieve[] theobjectives of the rule,’’ section 205(a) ofUMRA. The UMRA Conference Reportconfirms that UMRA does not overridethe authorizing statute. ‘‘This section[202] does not require the preparation ofany estimate or analysis if the agency isprohibited by law from considering theestimate or analysis in adopting therule.’’ 141 Cong. Rec. H3063 (daily ed.March 13, 1995).

The RFA (as amended by SBREFA)also does not apply to this decision, as


85 126 Cong. Rec. 21452, 21455 (1980)(Description of Major Issues and Section-By-SectionAnalysis of Substitute for S. 299).

86 Contrary to one of the comments received,EPA’s use of risk assessment in this rulemaking isby no means a departure from past practice. TheEPA first considered and began applying riskassessment methods in the late 1970’s (44 FR 8210,8211, February 8, 1979).

discussed in Unit VIII. of this preamble.As is the case with UMRA, even whenthe RFA (as amended by SBREFA) doesapply to a regulatory action, it does notestablish decisional criteria that anagency must follow, much less overridethe underlying substantive statute.When the RFA was adopted in 1980,Congress made clear that it did not alterthe substantive standards contained inauthorizing statutes: ‘‘The requirementsof section 603 and 604 of this title [toprepare initial and final regulatoryflexibility analyses] do not alter in anymanner standards otherwise applicableby law to agency action.’’ Section 606 ofthe RFA. The legislative history furtherexplains that section 606 ‘‘succinctlystates that this bill does not alter thesubstantive standard contained inunderlying statutes which defines theagency’s mandate.’’85 When Congresspassed SBREFA in 1996 and amendedparts of the RFA, it did not amendsection 606.

Even when a regulatory decision issubject to sections 603 and 604 of theRFA and an agency is therefore requiredto analyze alternatives that minimizesignificant economic impacts on smallentities, the RFA (as amended bySBREFA) does not establish decisionalcriteria that an agency is required tofollow. Both section 603 and 604 of theRFA provide that the alternatives anagency should consider are to be‘‘consistent with the stated objectives ofapplicable statutes.’’ Section 603(c) and604(a)(5) of the RFA. Furthermore,although the RFA (as amended bySBREFA) requires agencies to consideralternatives that minimize impacts onsmall entities subject to the rules’requirements and to explain their choiceof regulatory alternatives, it does notrequire agencies to select suchalternatives. For these reasons, the RFA(as amended by SBREFA) does notconflict with or override the Clean AirAct’s preclusion of considering costsand similar factors in setting NAAQS.

3. Conclusion. In summary, EPAdetermines that the judicial decisionscited in this unit are both correct anddispositive on the question ofconsidering costs in setting NAAQS,and that the Agency is not free toreinterpret the Act on that question.

B. Margin of SafetySeveral commenters questioned the

approach used by the Administrator inspecifying PM standards that protectpublic health with an adequate marginof safety. Rather than the integrative

approach applied by the Administrator,these commenters maintained that EPAmust employ a two-step process. Oneline of argument was that theAdministrator must first determine a‘‘safe level’’ and then apply a margin ofsafety taking into account costs andsocietal impacts. It was argued that thiswas the only approach that wouldenable the Administrator to reach areasoned decision on a standard levelthat protects public health againstunacceptable risk of harm, such that anyremaining risk was ‘‘acceptable.’’ Ineffect, these commenters adopted thetwo-step methodology endorsed byVinyl Chloride, 824 F.2d 1146, forsetting hazardous air pollutantstandards under section 112 of the Act.Another commenter also maintainedthat the Administrator must apply atwo-step process but from a differentperspective. It was argued that EPAshould first identify the lowest observedeffect level and then apply a margin ofsafety to address uncertainties and toprotect the most sensitive individualswithin the at-risk population(s). Thiscommenter also maintained that the useof risk assessment in establishing aNAAQS was a departure from pastpractice, and that this departure was notadequately explained.

In recognition of the complexitiesfacing the Administrator in determininga standard that protects public healthwith an adequate margin of safety, thecourts have declined to impose anyspecific requirements on theAdministrator’s methodologicalapproach. Thus, in Lead Industries thecourt held that the selection of anyparticular approach to providing anadequate margin of safety ‘‘is a policychoice of the type Congress specificallyleft to the Administrator’s judgment.This court must allow him thediscretion to determine which approachwill best fulfill the goals of the Act.’’647 F.2d at 1161–1162. As a result, theAdministrator is not limited to anysingle approach to determining anadequate margin of safety and, in theexercise of her judgment, may choose anintegrative approach, a two-stepapproach, or perhaps some otherapproach, depending on the particularcircumstances confronting her in agiven NAAQS review.

With respect to the approachesadvanced in comment, the PM10 casemade clear that the two-step processendorsed in Vinyl Chloride wasnecessary because of the need undersection 112 of the Act to ‘‘severdeterminations that must be basedsolely on health considerations fromthose that may include economic andtechnical considerations.’’ 902 F.2d at

973. Because the Administrator may notconsider cost and technologicalfeasibility under section 109 of the Act,however, the Court concluded that ‘‘therationale for parsing the Administrator’sdetermination into two steps isinapposite.’’ Id.

The claim that EPA must follow atwo-step process of first identifying thelowest observed effects level and thenapplying a margin of safety has alsobeen rejected by the courts. In LeadIndustries, the Court specifically heldthat the Administrator need not apply amargin of safety at the end of theanalytical process but may take intoaccount margin of safety considerationsthroughout the process as long as suchconsiderations are fully explained andsupported by the record. 647 F.2d 1161–1162. Accord, PM10, 902 F.2d at 973–974.

Because such factors as the nature andseverity of the health effects involved,the size of the sensitive population(s) atrisk, the types of health informationavailable, and the kind and degree ofuncertainties that must be addressedwill vary from one pollutant to another,the most appropriate approach toestablishing a NAAQS with an adequatemargin of safety may be different foreach standard under review. Thus, nogeneralized paradigm such as thatimbedded in EPA’s cancer risk policycan substitute for the Administrator’scareful and reasoned assessment of allrelevant health factors in reaching sucha judgment. As noted in this unit, bothCongress and the courts have left to theAdministrator’s discretion the choice ofanalytical approaches and tools,including risk assessments, rather thanprescribing a particular formula forreaching such determinations.86 Becauseof the inherent uncertainties that theAdministrator must address in marginof safety determinations, they are largelyjudgmental in nature, particularly withrespect to non-threshold pollutants, andmay not be amenable to quantificationin terms of what risk is ‘‘acceptable’’ orany other metric. In view of theseconsiderations, the task of theAdministrator is to select an approachthat best takes into account the natureof the health effects and otherinformation assessed in the air qualitycriteria for the pollutant in question andto apply appropriate and reasonedanalysis to ensure that scientific


87 Contrary to this commenter’s assertion, boththe health and air quality data used in the 1996Schwartz study are available to interested parties.EPA’s Office of Research and Developmentmaintains a copy of the air pollution database usedin the Schwartz study and it has previously beenmade available in response to Freedom ofInformation Act requests from interested parties,such as the American Iron and Steel Institute (AISI).The Harvard School of Public Health has also madethis data available to several collaborators and tothe Health Effects Institute. With regard to thehealth data underlying the Schwartz study, thatmortality data was compiled by the National Centerfor Health Statistics (NCHS) and can be purchasedfrom the NCHS by interested parties. Thus, there isno real data availability concern with regard to the1996 Schwartz study. However, even were this notthe case, for the reasons discussed more fully in thisunit and elsewhere in the preamble, EPA believesit would be entitled to rely upon this study andother studies, including the Dockery and Popestudies, regardless of the availability of theunderlying health data.

88 API’s letter stated that ‘‘API petitions EPA toidentify all studies that rely, in any way, on data

not available for public review as part of therulemaking process and remove those studies fromthe record.’’ To the extent this letter constitutes a‘‘petition’’ for EPA action, EPA hereby denies the‘‘petition’’ for the reasons stated in this unit andelsewhere in this preamble.

89 One commenter argued that the failure toobtain and disclose the underlying data was aviolation of the Administrative Procedure Act(APA). The NAAQS rulemaking is promulgatedunder section 307(d) of the Act; the APA generallydoes not apply to such rulemakings. See section307(d)(1) of the Act.

90 It is important to note that while EPA did usethe Dockery and Pope studies to confirm itsconclusions regarding the health effects of fineparticulate air pollution and thus as support for itsdecision to revise the PM standard, these studies donot provide the sole (or even primary) basis forEPA’s decision regarding PM2.5, despite theassertions of numerous commenters. The proposedstandards are based on a consideration of a largebody of epidemiological studies, a clear majority ofwhich suggest PM is strongly linked to mortalityand other serious health effects at concentrationspermitted under the current standards. Althoughthe specific levels of the PM2.5 standards are basedon a more limited number of studies that actuallymeasured fine particles and/or components of fineparticles, the Dockery and Pope studies were notused in initially selecting the annual fine particlestandard level, which was principally based onexamination of other daily mortality and respiratoryeffects studies (Koman, 1996, 1997) that foundsignificant associations between fine PM and effectsin cities with annual average PM2.5 concentrationsof about 16 to 21 µg/m3. Only then were the long-term Dockery and Pope studies examined and usedto help corroborate this result; in the opinion of theAdministrator, neither study alone (or together)

provided sufficient evidence to support morestringent levels below those identified from thedaily studies. Thus, removal of the Dockery andPope studies would not affect the conclusions aboutthe significance of the risks and therefore, whilethese long-term studies tend to strengthen the needfor fine particle control and provide importantinsights into the nature of PM effects, removal ofthese two studies from consideration would nothave changed the selected standard level.

91 Some commenters noted that with regard to thehealth data underlying the 1993 Dockery and 1995Pope studies, since EPA provided partial fundingfor these studies, EPA has access to this data andcannot shield itself from the duty to obtain this databy claiming that it is not in its possession. Althougha legal argument potentially exists that EPA mayobtain access to such data, this legal argument hasnot been tested in the courts. More importantly,EPA’s ability to rely on studies without reviewingthe raw data should not depend on whether someAgency of the Federal government funded thescience.

uncertainties are taken into account inan appropriate manner.

In this instance, the Administrator hasclearly articulated the factors she hasconsidered, the judgments she has hadto make in the face of uncertain andincomplete information, and alternativeviews as to how such informationshould be interpreted, in reaching herdecision on standard specifications thatwill protect public health with anadequate margin of safety. See Unit II.of this preamble. Her conclusions onthese matters are fully supported by therecord.

C. Data AvailabilitySeveral commenters questioned EPA’s

ability to rely on studies demonstratingan association between PM and excessmortality without obtaining anddisclosing the raw ‘‘data’’ underlyingthese studies for public review andcomment. In particular, a number ofcommenters cited Dockery, D.W., et al.1993 and Pope, C.A. III, et al., 1995, asstudies upon which EPA relied withoutobtaining and disclosing the underlyingraw data. One commenter also cited J.Schwartz et al., 1996 in the samecontext.87 According to the commenters,without the underlying data used inthese studies, the reliability of thesestudies cannot be assessed accurately.These commenters requested that EPAobtain the relevant data and make itavailable for public review. In light ofthe court-ordered requirement that EPApublish its rule by July 19, 1997, thecommenters argued that EPA mustretain the current PM10 NAAQS pendingadditional review of the raw data andthe studies at issue. One commenter, theAmerican Petroleum Institute (API)requested that EPA remove the studiesfrom the docket, unless the underlyingdata was also included in the docket.88

A few commenters argued that section307(d) of the Act requires that EPAobtain the raw data underlying thesestudies and that a failure to do socontradicts the plain language of section307(d)(3) of the Act, which requires EPAto place in the docket any ‘‘factual dataon which the proposed rule is based.’’Other commenters argued that undersection 307(d)(8) of the Act, a failure toobtain and disclose the underlying rawdata used in the studies wouldconstitute an error ‘‘so serious andrelated to matters of such centralrelevance to the rule that there is asubstantial likelihood that the rulewould have been significantly changedif such errors had not been made.’’ Id.According to one commenter, withoutthe raw data and an opportunity for ananalysis of it, ‘‘EPA has no legalalternative other than to conclude thatno new air quality standard would beappropriate within the meaning of CAAsection 109(a)(1)(B).’’ Finally, a numberof commenters have argued that recentcaselaw under the Clean Air Act andother statutes makes clear that EPA hasa legal obligation to obtain and disclosethe data used in these studies.89

In developing the proposed revisionsto the PM NAAQS, the Administratorrelied on the scientific studies cited inthe rulemaking record, rather than onthe raw data underlying them.90 In this

case, the raw data consists of responsesto health questionnaires based oninformation supplied by individualcitizens, or computer tabulations of thisinformation, which remainsconfidential, and air quality andmonitoring data, most of which is nowpublicly available. EPA does notgenerally undertake evaluations of raw,unanalyzed scientific data as part of itspublic health standard setting process.Only in extreme cases—for examplewhere there are credible allegations offraud, abuse or misconduct—would areview of raw data be warranted. Itwould be impractical and unnecessaryfor EPA to review underlying data forevery study upon which it relies assupport for every proposed rule orstandard. If EPA and othergovernmental agencies could not rely onpublished studies without conductingan independent analysis of theenormous volume of raw dataunderlying them, then much plainlyrelevant scientific information wouldbecome unavailable to EPA for use insetting standards to protect publichealth and the environment. Inaddition, such data are often theproperty of scientific investigators andare often not readily available becauseof the proprietary interests of theinvestigators or because of arrangementsmade to maintain confidentialityregarding personal health status andlifestyle information of individualsincluded in such data. Withoutprovisions of confidentiality, thepossibility of conducting such studiescould be severely compromised.91

In this case, the merits of the studiesconsidered and used in developing thePM2.5 standard have been discussed anddebated extensively over the pastseveral years, both as part of the EPAreview of the pertinent science and ina number of other public forums. Thestudies at issue were critically evaluated


by the Agency’s Office of Research andDevelopment (ORD) and by the EPA’sindependent Clean Air ScientificAdvisory Committee (CASAC), in amulti-year process for assessment of thescience at issue. As with other studieson which EPA relied, particularattention was given to the strengths andlimitations of the Dockery, Schwartzand Pope studies during this process,which involved numerous opportunitiesfor public participation and extensiveinput from interested parties. Theresults of these studies are not onlyconsistent with each other, but they arealso consistent with the results of otherstudies demonstrating significantassociations between long-termexposure to fine particle indicators andmortality. See U.S. EPA, 1996b, p. V–62.The CASAC concluded that EPA’sassessments of the pertinent scienceproperly characterized both the currentstate of knowledge and the range ofpolicy options for revising thestandards.

In fact, many peer reviewed studieshave reported associations between PMand premature death; the Dockery,Schwartz and Pope studies are amongthe most recent studies to corroboratethis association. In the early 1990s,several studies were published showingassociations at levels below the currentPM standards. Some critics beganraising questions about the extent towhich the results could be reproducedand the unavailability of underlyingdata. In response, an independent groupof investigators under the auspices ofthe Health Effects Institute (HEI), ahighly respected research organizationjointly funded by EPA and severalmotor vehicle manufacturers, undertooka reanalysis of several such studies. Theoriginal investigators of several studies,including studies conducted at HarvardUniversity, Brigham Young University,and the San Francisco Bay Area AirQuality Management District providedtheir raw air quality data sets to the HEIinvestigation team for reanalysis. HEI’sreanalysis produced numerical resultsfrom the data sets for all six cities thatclosely agree with and, in general,confirm the results of the originalinvestigators. Thus, as noted in Unit II.of this preamble, these reanalyses byrespected independent scientistsconfirmed the reliability andreproduceability of prior work of theoriginal investigators, including work byDockery et al. (1992), Pope et al. (1992),Schwartz and Dockery (1992a), andSchwartz (1993).

Thus, the 1993 Dockery and 1995Pope studies build upon previousstudies done by a number of differentresearchers and have been subject to an

extensive peer review process by EPA’sORD, CASAC and HEI. They alsounderwent a peer review process at thetime of their publication in reputablescientific journals. Given theconsistency and coherence of thescientific evidence and the scrutiny thestudies have received in peer reviewand in the extensive scientific reviewprocess described in this unit, EPA doesnot agree that review of the underlyingdata for these studies is also necessary.Considering the various reviewsdescribed in this unit and the fact thatEPA has received no specific andsubstantiated reason, such as plausibleallegations of fraud or scientific abuse,to doubt the overall validity of theirconclusions, EPA agrees with CASACthat revision of the standard isappropriate, based on these and otherstudies.

In spite of EPA and CASAC’sconclusion that it is appropriate to relyon the Pope, Dockery and other studiesto establish a PM2.5 NAAQS, EPA alsobelieves in public disclosure andsupports efforts to seek appropriaterelease of data underlying the studies inquestion. On January 31, 1997, EPAwrote to the principal scientificinvestigators at the Harvard School ofPublic Health and at Brigham YoungUniversity and urged them to make thedata associated with their studiesavailable to interested parties. Studiesconducted by these investigators reliedon data compiled as part of the HarvardSix-Cities Study and data compiled bythe American Cancer Society (ACS) aspart of the Cancer Prevention Study II.

The studies in question combinedhealth data on individuals with airpollution data. The air pollution dataare publicly available. The health dataconsist of personal and confidentialinformation, e.g. age, sex, weight,eduction level, smoking history,occupational exposures, medicalhistory. These data are not publiclyavailable. In compiling these data,researchers have promised studyparticipants that private, personalinformation would be kept confidentialunder signed assurances ofconfidentiality. Data-sharingarrangements with outside parties must,therefore, accommodate interests bothin making data accessible and inprotecting the confidentiality of theinformation contained within them.

Both the Harvard School of PublicHealth and the American Cancer Societyhave made such arrangements. Bothhave processes which allow ousidescientists, in collaboration with Harvardand ACS researchers, to access theirdatabases for the conduct of legitimatescientific research. Scientists from all

over the world have applied for andhave been granted such access andnumerous studies have been conductedand published using the databases.

Because of increased interest resultingfrom EPA’s rulemaking on PM standardsand at the request of the Harvard Schoolof Public Health, HEI is takingadditional steps to provide a forum foroutside researchers to access health dataassociated with the Harvard-Six CitiesStudy and perhaps others. HEI hasconvened an expert panel of esteemedscientists to access underlying data andto conduct additional reanalyses. Thisarrangement appears to provide aconstructive venue for testing legitimatescientific hypotheses while protectingthe confidentiality of the underlyingdata.

Nevertheless, as noted previously,EPA has full confidence in the scientificintegrity of the Dockery, Schwartz, andPope studies and their suitability for usein the Agency’s rulemaking on PM,without undertaking a separate oradditional review and analysis of theunderlying raw data. The decision topropose revisions of the current PMstandards was based on carefulassessment of the scientific andtechnical information presented in thePM Criteria Document and Staff Paper.The decision was also consistent withthe consensus of CASAC that ‘‘althoughan understanding of health effects of PMis far from complete, the Staff Paper,when revised, will provide an adequatesummary of our present understandingof the scientific basis for makingregulatory decisions concerning PMstandards.’’ The extensive PMepidemiological data base providesevidence that serious adverse healtheffects, e.g., mortality, exacerbation ofchronic disease, increased hospitaladmissions, respiratory symptoms, andpulmonary function decrements, insensitive subpopulations, e.g., theelderly, individuals withcardiopulmonary disease and children,are attributable to PM at levels belowthe current standards. The increase inrisk is significant from an overall publichealth perspective because of the largenumber of individuals in sensitivesubpopulations that are exposed toambient PM and the significance of thehealth effects. These considerations, aswell as others discussed in the proposaland Staff Paper, such as the need toconsider fine and coarse particles asdistinct classes, led both theAdministrator and CASAC to concludethat revision of the current standards isclearly appropriate. This conclusionremains unchanged despite the fact thatEPA is without the actual raw and


92 EPA also does not agree that because thelanguage of section 307(d) of the Act mentions‘‘factual data’’ as well as ‘‘the methodology used inobtaining and analyzing the data,’’ EPA cannot relyon a study alone. In this case, the study is the‘‘factual data’’ and EPA’s methodology used inobtaining and analyzing the ‘‘factual data’’ is themethod that EPA used to review and rely upon thestudies. This methodology is discussed extensivelyin the staff paper and summarized in some detailelsewhere in this preamble. In fact, as is clear fromthe overall structure of section 307(d) of the Act, aswell as the legislative history cited in this unit,section 307(d) of the Act merely requires that EPAsummarize and disclose the information andmethodology that it relied upon in developing itsrule. It leaves unchanged the ‘‘level’’ of support thatan agency must bring to bear in drafting a proposedrule.

unanalyzed health data underlying thestudies.

A number of commenters citedsection 307(d) of the Act in support oftheir position that EPA is required toobtain and disclose the underlying rawdata. Under section 307(d)(3) of the Act,EPA is required to issue a notice ofproposed rulemaking in the FederalRegister that is accompanied by a‘‘statement of basis and purpose’’ thatincludes ‘‘a summary’’ of:

(A) The factual data on which theproposed rule is based.

(B) The methodology used inobtaining the data and in analyzing thedata.

Thus, it is clear from the language ofsection 307(d) of the Act that whereEPA relies on any ‘‘data’’ as support inits rulemakings under the Clean Air Act,it has an obligation to include such dataor information in the rulemaking docketthat is open to the public. Where EPAfails to do so and the error is ‘‘so seriousand related to matters of such centralrelevance to the rule that there is asubstantial likelihood that the rulewould have been significantly changedif such errors had not been made,’’ areviewing court may overturn the rule.

In this case, as noted previously, EPAdid not rely upon the raw health datasupporting the Dockery and Popestudies; it relied instead upon thestudies themselves. These studies mayproperly be considered ‘‘data.’’ The EPAhas never had the raw health data in itspossession; thus EPA has neitherreviewed it nor had an opportunity toplace it in the docket. The EPA did relyon the studies and these studies areincluded in the docket and are availablefor public review. Because EPA neitherreviewed nor relied upon the raw data,there is no obligation to obtain it or tomake it available.

Some commenters argued that thelanguage of section 307(d) of the Act,which refers to the ‘‘factual data’’ andwhich also discusses the ‘‘methodologyused in obtaining and analyzing thedata’’ distinguishes between raw dataand studies. In the view of these andother commenters, the plain language ofsection 307(d) of the Act requires thatEPA obtain and disclose the raw dataused in the Dockery and Pope studies.According to these commenters, withoutsuch raw ‘‘data,’’ EPA cannot legallypromulgate its rule.

The EPA disagrees with this narrowinterpretation of the word ‘‘data’’ and ofsection 307(d) of the Act. Data can takemany forms, including studies, reports,tabulations, graphs and summaries, aswell as more raw forms, such asquestionnaire responses, test results andeven actual physical specimens. The

‘‘factual data’’ called for by section307(d) of the Act may clearly includepeer-reviewed scientific studies. Nordoes section 307(d) of the Act prohibitEPA from relying on a study forstandard setting without obtaining theraw, underlying data supporting astudy. Indeed, as noted in the legislativehistory to section 307(d) of the Act,

* * * [t]he [House Commerce] Committeerecognizes that the factual support needed fora rule may vary greatly according to thesubject being addressed and that rules onsome subjects, such as procedures, may notrequire any factual basis at all. There is nointention to increase the amount of ‘factual’support now required to support ‘policyjudgments where no factual certainties existor where facts alone do not provide theanswer,’ Industrial Union Department, AFL-CIO v. Hodgson, 499 F.2d 467, 476 (D.C. Cir.1974). Nor is there any intent to diminish theAdministrator’s authority to adoptprecautionary regulations based on ashowing of risk * * * .

H.R. Rep. No. 95-294, at 323 (1977)(footnote omitted). As this legislativehistory makes clear, the language insection 307(d) of the Act is not intendedto require EPA to change the amount of‘‘factual support’’ that EPA mustassemble in order to promulgate a ruleand EPA may adopt ‘‘precautionary’’regulations ‘‘where no factual certaintiesexist.’’ Given this clarification in thelegislative history, it is evident that EPAis entitled under section 307(d) of theAct to rely on studies rather than rawdata in developing its Clean Air Actrules, despite the arguably ambiguoususe of the term ‘‘data.’’92

Moreover, EPA has relied on studiesin the past (including studies using theundisclosed Six Cities data) withoutobtaining or disclosing the underlyingraw data, and EPA’s reliance on suchstudies to set Clean Air Act standardshas been upheld in court. In NRDC v.EPA, 902 F. 2d 962 (D.C. Cir. 1990), theD.C. Circuit declined to delay its reviewof the PM10 NAAQS rulemaking due toconcerns raised by the American Ironand Steel Institute about the integrity of

the Six Cities data base. 902 F.2d at 974.In that case, EPA had relied upon anearlier Dockery study based on the SixCities data base. Although the NationalInstitutes of Health (NIH) undertook areview of the allegations regarding theSix Cities database, the courtnevertheless upheld EPA’s reliance onthat Dockery study without waiting forthe results of the NIH review. NIHeventually concluded that theallegations were without merit.According to the court in the NRDCcase:

AISI claims that the EPA relied too muchon the Six Cities Study, which is comprisedof the Dockery study and the Ware study ** * . We do not agree that the Administrator’sselection of the twenty-four hour standardlacks the necessary reasoned analysis andsupportive evidence * * * . After carefullyreviewing the record, we find EPA’s selectionof the twenty four hour standard reasonablein light of the divergent results in the studiesand the agency’s mandate to provide anadequate margin of safety. Studies containedin the record provided evidence of adversehealth effects at levels below 250 µg/m3.

902 F.2d at 969 (footnotes omitted;emphasis in original). The court alsostated that:

In setting a standard under section 109 ofthe Act, the Administrator must ‘‘take intoaccount all the relevant studies revealed inthe record‘‘ and ‘‘make an informed judgmentbased on available evidence.’’ AmericanPetroleum Institute v. Costle, 665 F.2d at1187. The record shows that theAdministrator did so. The Administratorrelied on studies which showed adverseeffects at and below the 250 µg/m3 level.AISI essentially asks this court to givedifferent weight to the studies than did theAdministrator. We must decline. It is simplynot the court’s role to ‘‘second-guess thescientific judgments of the EPA. * * * [T]heAdministrator did not act arbitrarily indrawing conclusions from the uncertain andconflicting data. The Administrator mayreasonably apply his expertise to drawconclusions from ‘‘imperfect data,’’ EthylCorp., 541 F.2d at 28, as he did here.

Id. at 971.As this language makes plain, the

term ‘‘data’’ may include a study reliedupon by EPA. It should be equally plainthat EPA may properly rely on such astudy in setting a standard despite thefact that such ‘‘data’’ may be‘‘imperfect,’’ ‘‘conflicting,’’ and‘‘uncertain.’’ There are numerous othercases in which EPA has relied onstudies in setting standards under theClean Air Act. See, e.g., EngineManufacturers Association v. EPA, 88 F.3d 1075, 1099 (D.C. Cir.1996)(upholding EPA’s use of the 1993Dockery study for setting mobile sourcestandards); API v. Costle, 665 F.2d 1176,1185 (D.C. Cir. 1981)(Administrator’sconclusion that normal body functions


are disrupted by ozone is ‘‘supported bythe studies’’).

A number of commenters citedEndangered Species Committee v.Babbitt, 852 F. Supp. 32 (D.D.C. 1994)(hereafter ‘‘Gnatcatcher’’) in support ofthe proposition that EPA must obtainand disclose the raw data underlyingthe Dockery and Pope studies. Relyingon cases such as Connecticut Light andPower Co. v. NRC, 673 F.2d 525 (D.C.Cir. 1982), Portland Cement v.Ruckelshaus, 486 F.2d 375 (D.C. Cir.1973), and United States v. Nova ScotiaFood Processing Corp, 568 F.2d 240(2nd Cir. 1977), these commenterssuggest that ‘‘a body of legal decisionsis emerging whereby federal courts areincreasingly dubious of final regulatorydecisions that are being made absentpublic scrutiny of the data underlyingand purportedly supporting suchdecisions.’’ According to thesecommenters, based on Gnatcatcher andother cases, failure by EPA to obtain andplace in the docket the raw unanalyzeddata used in the Dockery and Popestudies constitutes serious proceduralerror under the Clean Air Act.

Under Connecticut Light and Power,agencies must make available technicalstudies and data that have been reliedupon during the rulemaking process inorder for the public to have an adequateopportunity for notice and comment.There is no question that EPA has donethis with regard to the Dockery andPope studies, which are included in therulemaking docket. The PortlandCement case makes clear that where anagency actually relies on factual data itcannot ‘‘promulgate rules on the basis ofinadequate data, or on data that, [to a]critical degree, is known only to theagency.’’ 486 F.2d at 393. See also, NovaScotia, 568 F.2d 240, at 251 (where allof the research was collected by theagency, and none of it was disclosed ‘‘asthe material upon which the proposedrule would be fashioned,’’ errorresulted); CMA v. EPA, 870 F.2d 177,200 (5th Cir. 1989) (‘‘fairness requiresthat the agency afford interested partiesan opportunity to challenge theunderlying factual data relied on by theagency’’).

However, in this case, EPA did notrely on, nor did it ever have or review,the underlying data used in the Dockeryand Pope studies. Instead, it relied uponthe studies themselves. Thus, the casescited in this unit are inapposite. Theystand only for the proposition thatwhere an agency actually reviews andrelies on ‘‘data,’’ which may be rawdata, a study or a variety of other formsof information, it must make these dataavailable. They do not and cannot standfor the proposition that an agency may

not rely on a study alone and mustalways obtain the raw and unanalyzeddata underlying a study. Indeed, as oneD.C. Circuit case noted: ‘‘PortlandCement and Nova Scotia simply cannotbe twisted so as to require notices ofproposed or interim rules to containelaborate reproductions of underlyingstudies.’’ Petry v. Block, 737 F.2d 1193,1198 (D.C. Cir. 1984). Requiring EPA toobtain, analyze and disclose the dataunderlying the Pope and Dockerystudies, which EPA neither reviewednor relied upon, would be to requireEPA to attempt such an ‘‘elaboratereproduction.’’ Such a step is notrequired under the law and would makeit extremely difficult, if not impossible,for EPA to regulate in complex,technical areas ‘‘at the frontiers ofscience.’’ Baltimore Gas and Electric Co.v. NRC, 462 U.S. 87 (1983).

The district court’s decision in theGnatcatcher case is similarly inapposite.That case concerned a scientific studyregarding the range of the CaliforniaGnatcatcher, a small insectivoroussongbird. As the Gnatcatcher opinionitself notes, ‘‘courts have generallyallowed agencies to rely on scientificreports.’’ Gnatcatcher, 852 F.Supp. at37. Thus, the question at issue inGnatcatcher was whether specificcircumstances exist in which an agencymay not be entitled to rely on studiesalone. In the Gnatcatcher case, a singleauthor had published two directlycontradictory studies on the same issue,while relying on the same data. In lightof this clear contradiction, commentersin that rulemaking argued that withoutthe underlying data it was impossible todetermine whether the conclusions ineither study were correct. The districtcourt noted that:

The Secretary had before him a report byan author who, two years before hadanalyzed the same data and come to anopposite conclusion. It is the disputed natureof this report that distinguishes this fromother cases where a scientific report alonehas been considered sufficient for ESApurposes.

Id. Thus, according to the court: ‘‘Whilecourts have generally allowed agenciesto rely on scientific reports * * * this isnot sufficient in this case because thereport itself is under serious question.’’Id.

The EPA’s current reliance on theDockery and Pope studies bears noresemblance to the circumstancespresent in the Gnatcatcher decision. Asnoted previously, these studies havebeen subject to extensive peer reviewand scrutiny, and neither researcher haspublished a contradictory study on thesame issue, much less using the samedata base. The EPA is not aware of, nor

have any of the commenters raised anyparticular issues relating to either grosserror, fraud or scientific abuse arisingfrom the data. Indeed, as notedpreviously, the prior work of theseparticular researchers has been subjectto extensive independent scrutiny andreanalysis, which has confirmed, ratherthan called into question, theunderlying validity of their conclusionsand the integrity of their researchmethods. Reading Gnatcatcher tosuggest that EPA cannot rely on such astudy, where the study and its methodshave been subject to extensive peerreview, would place the district court’srationale in Gnatcatcher in conflict withapplicable D.C. Circuit precedent thatmakes evident the right of agencies torely on studies alone. See, e.g., EngineManufacturers Association v. EPA, 88F.3d 1075, 1099 (D.C. Cir 1996); API v.Costle, 665 F.2d 1176, 1185 (D.C. Cir.1981), ‘‘studies discussed in the CriteriaDocument constitute a rational basis forthe finding that adverse health effectsoccur at ozone levels of 0.15-0.25 ppmfor sensitive individuals’’; see also,NRDC v. Thomas, 805 F.2d 410, 418(D.C. Cir. 1986)(EPA use of a summaryof confidential data that was notdisclosed provides ‘‘a reasonedexplanation for moving from a 4.0 to 5.0long term NOx standard’’).

In addition, to require EPA to obtainand analyze the data prior to revisingthe standard would also contradict the‘‘common sense notion that Congress, inproviding for notice and commentunder the APA, could not have intendedto subject the agencies—and the publicon whose behalf they regulate—to [a]sort of interminable back and forth.’’International Fabricare Institute v. EPA,972 F.2d 384, 399 (D.C. Cir. 1992). Inthe view of some commenters, EPA hasno choice but to either postpone itsdecision for a year or more awaiting areview of data or choose to retain thecurrent standard. Yet were EPA to adoptsuch an approach, these commenterswould undoubtedly insist that EPA berequired to include an analysis of thedata in the docket; further questionswould likely be raised regarding the re-analysis and once again EPA might finditself unable to promulgate its rulepending review of further hypotheticalquestions. This type of unendinginquiry is not required under the law.As the D.C. Circuit has noted:

* * * [D]isagreement among the experts isinevitable when the issues involved are at the‘‘very frontiers of scientific knowledge,’’ andsuch disagreement does not prevent us fromfinding that the Administrator’s decisions areadequately supported by the evidence in therecord * * * . It is not our function to resolvedisagreement among the experts or to judge


93 One commenter argued that EPA’s failure toplace the ‘‘data’’ in the docket was not an ‘‘error’’but a ‘‘refusal to comply with the clear language ofthe law that should be reviewed by the courts undersection 307(d)(9)(C), rather than 307(d)(9)(D).’’ Asnoted previously, EPA does not agree with thisinterpretation of section 307(d)(3) of the Act. Underapplicable caselaw, the term ‘‘data’’ may includeinformation in many forms, including studies that

EPA has placed in the docket. See EndangeredSpecies Committee v. Babbitt, 852 F. Supp. 32, 37(D.D.C., 1994) (‘‘data can come in many forms: itcan be a scientific report, it can be graphs andtabulations * * * it can be raw numbers’’).

94 A number of commenters did argue thesestudies do not form a sufficient basis for EPA’sdecision to revise the NAAQS and that attempts toreplicate these studies have not been universallysuccessful. These same commenters also listed anumber of hypothetical questions and issues thatmight be resolved through a review of theunderlying data and suggested that before EPA mayproperly rely on these studies to revise the NAAQS,a variety of confounders (such as smoking) shouldalso be ruled out by reviewing the data. As set forthmore fully in Unit II. of this preamble, neither EPAnor CASAC agrees that any of these factorsprecludes reliance on the studies in question.

the merits of competing expert views * * *. Cf. Hercules, Inc. v. EPA, 598 F.2d 91,115(D.C. Cir. 1978) (‘‘[c]hoice among scientifictest data is precisely the type of judgmentthat must be made by EPA, not this court’’).

Lead Industries Association v. EPA, 647F.2d 1130, 1160 (D.C. Cir. 1980).

Neither Gnatcatcher, nor any othercase can fairly be read to suggest thatEPA has an obligation to respond to allpossible questions that might be raisedregarding its scientific conclusions orthat where EPA relies on a study, itmust engage in a multi-phased andpossibly unending re-examination of thedata supporting such a study until allcommenters are satisfied in full with thedetails of the underlying science. Evenassuming that EPA could obtain theconfidential Six Cities data throughlitigation, a substantial delay of manymonths, if not years, would likely result,in order for both EPA and industry toreanalyze the data. In the meantime,some tens of thousands of prematuredeaths could result. Neither the CleanAir Act nor relevant case law requiresor permits such a result.

Indeed, the suggestion that EPAcannot and should not rely upon thePope, Dockery, and Schwartz studies,unless and until interested parties havehad an opportunity to examine andreanalyze the underlying raw data, isextraordinary. Given the precautionarynature of section 109 of the Act and theneed to allow an adequate margin ofsafety, see Lead Industries, 647 F.2d at1154, 1155, there are limits on EPA’sdiscretion to disregard even studies thatare clearly flawed, if they arenonetheless ‘‘useful’’ in indicating thekind and extent of health effects thatmay result from the presence of apollutant in the ambient air. Seesections 109(b)(1) and 108(a)(2) of theAct.

A few commenters cited Kennecott v.EPA, 684 F.2d 1007 (D.C. Cir. 1982) andargued that under sections 307(d)(8) and307(d)(9)(D) of the Act, a failure by EPAto obtain and include in the docket thedata underlying the Pope and Dockerystudies would constitute an ‘‘error’’ thatis ‘‘so serious and related to matters ofsuch central relevance to the rule thatthere is a substantial likelihood that therule would have been significantlychanged if such error[] had not beenmade.’’93 EPA disagrees. Peer reviewed

studies conducted by outside partieswere not at issue in Kennecott.Kennecott involved a dispute overfinancial analyses that EPA itself hadpreviously conducted and used inearlier rulemakings. The courtdetermined that the financial analyses atissue must have provided at least partof the factual basis for EPA’s rule, andEPA referenced these analyses in thepreamble to the final rule withoutplacing them in the docket until oneweek before promulgation. The factualcircumstances in Kennecott aresubstantially different than the currentsituation and thus, Kennecott cannotfairly be read to establish the applicablelegal standard with regard to EPA’sreliance on peer reviewed studies foruse in setting the NAAQS.

In this case, EPA—well beforeproposal—has placed the informationthat it relied upon in the docket. Thisinformation is in the form of studies.These studies have been subject toextensive scrutiny and peer review. Todate no specific allegation has beenmade that the studies are clearly in erroror that the data underlying them are thesubject of fraud, scientific misconduct,or gross error going to the basic validityof the studies.94 Instead, variouscommenters have merely stated theirview that were the raw data behindthese studies available, they would beable to better verify and assess theresults reached in the studies.

As one commenter noted, ‘‘In theabsence of data on which EPA’sproposal is based, [key scientific] issuesremain shrouded in uncertainty andskepticism. The disclosure of the datawould allow for robust scientificanalysis and discussion of these issues.’’A similarly hypothetical concern israised by another commenter who statedthat ‘‘seeing the data would clarifysubstantial questions of methodology’’and ‘‘had the Harvard data beenavailable, a far broader evaluation of thedefects of the Harvard Studies wouldhave been possible with the same

expenditure of time and money.’’ Yet,despite having spent ‘‘in theneighborhood of a million dollars toduplicate and reanalyze the Harvarddata set’’ this commenter was unable toallege any particular defect in themethodology or results of these studiesand noted instead that ‘‘the track recordto date suggests that the claimedassociations to PM2.5 and health effectswould not have held up under such abroader evaluation.’’

EPA is not required to await theresults of such an inquiry beforeproceeding to regulate to protect humanhealth and the environment. Theconcerns raised by the commentersregarding these studies remainhypothetical; the comments themselvesraise no allegations of fraud, scientificmisconduct or gross error that calls intoquestion the fundamental validity of thestudies. Given this fact, EPA does notagree with the commenters that relianceon these studies and/or a failure to placethe underlying data in the docketconstitutes an error, much less an errorthat is ‘‘so serious and related to mattersof such central relevance that there is asubstantial likelihood that the rulewould have been significantlychanged.’’ EPA is entitled to rely uponthese studies and it has satisfied itsobligation to provide the ‘‘factual data’’upon which the proposed rule is basedby placing these studies in the docket.

In fact, the concerns raised by thecommenters ultimately boil down to adisagreement with EPA over the level ofscientific certainty necessary to adoptthe NAAQS revisions. In settingstandards under the Clean Air Act, EPAis not required to resolve all scientificissues to the complete satisfaction ofevery interested party. As noted by theD.C. Circuit in Lead IndustriesAssociation v. EPA, 647 F.2d 1130, 1160(D.C. Cir. 1980):

To be sure, the Administrator’sconclusions were not unchallenged; both LIAand the Administrator are able to point to animpressive array of experts supporting eachof their respective positions. However,disagreement among the experts is inevitablewhen the issues involved are at the ‘‘veryfrontiers of scientific knowledge,’’ and suchdisagreement does not preclude us fromfinding that the Administrator’s decisions areadequately supported by the evidence in therecord. It may be that LIA expects this courtto conclude that LIA’s experts are right, andthe experts whose testimony supports EPAare wrong. If so, LIA has seriouslymisconceived our role * * * . It is not ourfunction to resolve disagreement among theexperts or to judge the merits of competingexpert views * * * . Cf. Hercules, Inc., v. EPA,598 F.2d 91, 115 (D.C. Cir. 1978) (‘‘[c]hoiceamong scientific test data is precisely thetype of judgment that must be made by EPA,not this court’’).


647 F.2d at 1160 (footnotes omitted).The EPA’s rationale for proposing to

add a fine particle standard was detailedin the preamble to the proposed rule,most notably at 61 FR 65654–65662,December 13, 1996. This decision isbased on the extensive review of thescience and policy issues contained inthe PM Criteria Document and StaffPaper; the CASAC concluded, afterextensive review, that both of thesedocuments were appropriate for use indecision making on standards. Thesedocuments contain a full discussion ofboth what is known about PM and theinformation gaps and uncertainties.Considering the full weight of thescientific evidence, including theuncertainties, the CASAC recommendedthat the Administrator adopt fineparticle standards and a number ofpanel members based their support fora PM2.5 standard on the followingreasoning:

[T]here is strong consistency andcoherence of information indicating that highconcentrations of urban air pollutionadversely affect human health, there arealready NAAQS that deal with all of themajor components of that pollution exceptPM2.5, and there are strong reasons to believethat PM2.5 is at least as important as PM10-2.5

in producing adverse health effects.

Wolff, 1996.Given the consistency and coherence

of the evidence that premature mortalityand sickness occur in large numbers ofAmericans at concentrations permittedby the current standards, it would beirresponsible to delay action that wouldput more appropriate air quality goalsinto place based on the most recentscientific information. After a review ofthe comments submitted, the Agency’sconclusion that it is appropriate to relyon the existing studies remainsunchanged.

D. 1990 AmendmentsContrary to the view expressed in

some public comments, the provisionsof subpart 4 of Part D of Title I of theAct, enacted in 1990, do not precludeEPA from adopting PM2.5 as anadditional indicator for PM andestablishing standards for PM2.5. Theprovisions of subpart 4 of Part D of TitleI of the Act simply do not limit EPA’sclear authority under section 109 of theAct to revise the PM standards.

The basic contention is that becausethe provisions of subpart 4 of Part D ofTitle I of the Act refer to PM10, theyprohibit EPA from regulating any othertype of PM, for example, by revising theexisting NAAQS for PM by adopting anambient air quality standard for PM2.5.These provisions, however, do not leadto such a conclusion. Moreover, this

view ignores provisions indicating thatCongress believed that EPA could reviseany existing NAAQS or adopt a newNAAQS.

At the outset, it should be noted thatCongress expressly authorized EPA torevise any ambient air quality standardand to adopt a new NAAQS in section109 of the Act. That section, whichrequires EPA to review and revise, asappropriate, each NAAQS every fiveyears, contains no language expressly orimplicitly prohibiting EPA from revisinga NAAQS or adopting a new NAAQS. IfCongress had intended to preclude EPAfrom reviewing and revising a NAAQSor adopting a new NAAQS, which arepart of EPA’s fundamental functions,Congress would have specifically doneso. Clearly, Congress knew how topreclude EPA from exercising otherwiseexisting regulatory authority and did soin other instances. See section202(b)(1)(C) of the Act (expresslyprecluding EPA from modifying certainmotor vehicle standards prior to modelyear 2004); section 112(b)(2) of the Act(preventing EPA from adding to the listof hazardous air pollutants any airpollutants that are listed under section108(a) of the Act unless they meet thespecific exceptions of section 112(b)(2)of the Act); section 249(e)(3), (f) andsection 250(b) (limiting EPA’s authorityregarding certain clean-fuel vehicleprograms). No such language wasincluded either in section 109 of the Actor elsewhere in the Act and no suchimplication may properly be based onthe provisions of subpart 4 of Part D ofTitle I of the Act.

Second, other provisions of the Actexpressly contemplate EPA’s ability topromulgate a new or revised NAAQS,and provide no indication that suchability is limited to standards other thanthose whose implementation is thesubject of subparts 2, 3 and 4 of Part Dof Title I of the Act. For example,section 110(a)(2)(H)(i) of the Actprovides that SIPs are to provide forrevisions ‘‘from time to time as may benecessary to take account of revisions ofsuch national primary or secondaryambient air quality standard * * * .’’Section 107(d)(1)(A) of the Act providesa process for designating areas asattainment, nonattainment, orunclassifiable ‘‘after promulgation of anew or revised standard for anypollutant under section 109 * * * .’’Section 172(e) of the Act addressesmodifications of national primaryambient air quality standards. Finally,section 172(a)(1) of the Act expresslycontemplates that EPA may revise astandard in effect at the time ofenactment of the 1990 Clean Air ActAmendments. Section 172(a)(1)(A) of

the Act provides EPA with authority toclassify nonattainment areas on or afterthe designation of an area asnonattainment with respect to ‘‘anyrevised standard, including a revision ofany standard in effect on the date of theenactment of the Clean Air ActAmendments of 1990.’’ Plainly,Congress had no intention of prohibitingEPA from revising any of the ambientstandards in effect at the time of theenactment of the 1990 amendments.

Third, the provisions of subpart 4 ofPart D of Title I of the Act do notsupport the contention that theysomehow preclude EPA from exercisingits authority to adopt a revised PMNAAQS based on a metric other thanPM10. The fact that Congress laid out animplementation program for the PMstandard existing at the time of the 1990amendments in no way suggests thatCongress intended to preclude EPAfrom exercising the authority itprovided EPA to revise the NAAQSwhen the health data on which EPAbases such decisions warranted achange in the standard.

The fact that Congress drafted subpart4 of Part D of Title I of the Act in 1990to specify the implementation regimefor the PM standard then in effect, aPM10 standard, in terms that explicitlyrefer to PM10 in no way suggests thatCongress meant to preclude EPA fromadopting a PM standard based onanother metric if scientific informationsupported such a change. Obviously,PM10 was the standard in existence in1990 and Congress drafted subpart 4 ofPart D of Title I of the Act, the purposeof which was to delineate animplementation regime for thatstandard, in terms of that standard.There is simply no language in subpart4 of Part D of Title I of the Act thatlimits EPA’s ability to establish adifferent PM standard if such a standardwere warranted under section 109 of theAct or indicates any implicit intent onthe part of Congress to limit EPA’sauthority under section 109 of the Actin such a way. Subpart 4 of Part D ofTitle I of the Act simply does not speakto the question of whether EPA mayestablish a PM standard based on adifferent metric. In addition, section107(d)(4) of the Act, the only provisionoutside of subpart 4 of Part D of TitleI of the Act invoked as a basis for theview that the Act prohibits EPA fromadopting a PM2.5 standard, does notsupport that view. That provisionsimply preserved pre-existingdesignations for ‘‘total suspendedparticulates,’’ the PM metric utilizedprior to PM10, for certain purposes. Itprovides no suggestion that Congressintended to prohibit EPA from adopting


a metric other than PM10. Indeed, ifanything, it indicates that Congress wasfully aware that EPA had previouslychanged the PM metric used in the PMNAAQS and confirms the view thatCongress would have explicitly barredEPA from changing the metric had itintended to do so.

Finally, for the reasons stated in thisunit, EPA’s analysis of its ability toimplement a PM2.5 standard under theprovisions of subpart 1 of Part D of TitleI does not support the view thatCongress prohibited EPA frompromulgating such a standard. Congressclearly specified an approach to theimplementation of the PM10 standard inthe provisions of subpart 4 of Part D ofTitle I of the Act. The EPA believes thatthe clear and express linkage of thatapproach to the PM10 standard indicatesthat a different PM standard should beimplemented under the generalprinciples of subpart 1 of Part D of TitleI of the Act. That Congress directedspecifically how EPA and the Statesshould implement the PM10 standarddoes not carry with it the implicationthat Congress intended to prohibit EPAfrom exercising its otherwise clear andexpress authority to adopt a PMstandard based on a different metric inorder to carry out one of its fundamentalmissions, the establishment of ambientair quality standards to protect publichealth with an adequate margin ofsafety. It is entirely reasonable andlogical for Congress to, on the one hand,specify an implementation regime forthe PM standard in effect at the time ofenactment of the 1990 amendments, but,on the other hand, leave EPA free toexercise the authority provided it byCongress in section 109 of the Act toadopt a new or revised standard whenEPA determined that such a standardwas needed to protect public healthwith an adequate margin of safety.Congress explicitly required EPA toreview and revise as appropriate theNAAQS every five years. If Congress didnot intend for EPA to revise the NAAQSwhen warranted, it would not haverequired EPA to review and revise them.If Congress had intended to prohibitEPA from exercising such afundamental authority it would haveclearly specified, as it did in otherinstances, that EPA could not do so.

V. Revisions to 40 CFR Part 50,Appendix K—Intrepretation of the PMNAAQS

Because the revocation of the existingPM10 standards will become effective ata later date (as discussed in Unit VII. ofthis preamble), EPA is retaining 40 CFRpart 50, Appendix K, although it isbeing published today in revised format

to conform with the format of the otherappendices in this part. A newAppendix N to 40 CFR part 50 explainsthe computations necessary fordetermining when the primary andsecondary PM2.5 and PM10 standardsbeing adopted today are met. Thediscussion in this unit sometimes refersto the contents of the new Appendix Nas revisions to Appendix K, so as tohighlight how the new Appendix Ndiffers from the current Appendix K.

Key elements of the new 40 CFR part50, Appendix N, particularly as theydiffer from those of Appendix K, areoutlined in this unit.

A. PM2.5 Computations and DataHandling Conventions

As discussed in Unit II.E. of thispreamble, the form of the annual PM2.5

standard is a spatially averaged annualmean averaged over 3 years, and theform of the 24-hour PM2.5 standard is a98th percentile concentration averagedover 3 years.

With regard to the annual PM2.5

standard, the EPA proposed a formexpressed as the annual arithmeticmean, averaged over 3 years andspatially averaged over all designatedmonitoring sites to represent populationexposures. As discussed in Unit II.E.1.of this preamble, the form of the annualPM2.5 standard has been clarified tomake explicit that implementingagencies have the flexibility to basecomparison of the standard level withmeasured values from either a singlecommunity-oriented site or an averageof measured values from such monitorswithin the constraints enumerated in 40CFR part 58. The new Appendix N of 40CFR part 50 reflects this clarification.The spatial average, if used, is to becarried out using data from monitoringsites designated in a State PMMonitoring Network Description inaccordance with the provisions of 40CFR part 58.

Also, the EPA proposed that, forspatial averaging, the requirements for 3years of data for comparison with thestandard be fulfilled by the spatialaveraging network as a whole, not byindividual monitors within the network.The EPA received comments regardingthe application of the 75 percent datacompleteness requirement to spatialaveraging. The commenters stated thatthe inclusion or exclusion of a site notmeeting the data completenessrequirements from a spatial average,based on the level of the single siteaverage, would bias the spatial averagefor that year. The EPA has responded tothe comment by demonstrating inExample 1 in 40 CFR part 50, AppendixN the application of the data

completeness criterion that is consistentwith a spatially averaged network.Specifically, the application of the datacompleteness requirement has beenaltered in the example if a particular sitehas quarters in a year that do not meetthe minimum data completenessrequirement. Instead of comparing asite’s annual average to the level of thestandard to decide whether or not tokeep the site in the calculations, theannual average for all the sites (thespatial average) is compared to the levelof the standard. If the spatial average isabove the level of the standard, the siteis kept in the calculations. If it is below,the site is omitted from the calculations.

The EPA also proposed that averagingover calendar quarters be retained forthe annual average form of the standard.Although several commenters statedthat the step of calculating quarterlyaverages to obtain the annual averagewas unnecessary, the EPA maintainsthat quarterly averages are important toensure representative sampling in areaswith extreme seasonal variation.

Regarding the 75 percent datacompleteness requirement, the proposalstated that a given year meets datacompleteness requirements when atleast 75 percent of the scheduledsampling days for each quarter havevalid data, and high values measured inincomplete quarters shall not be ignoredbut shall be included if their valuecauses the annual calculation to beabove the level of the standard. Somecommenters felt that this treatment wasunfair in that measured data below thestandard in incomplete quarters are notretained. In addition, the commentersfelt that this could create a bias wherea single sample could inflate an annualaverage to a level above the standard.The EPA agrees and has incorporated in40 CFR part 50, Appendix N thefollowing provisions.

(1) A statement has been added thatless than complete data may be used incertain cases subject to the approval ofthe appropriate Regional Administratorin accordance with EPA guidance fordealing with less than complete data.This statement was considerednecessary for those situations wheremeasured data and air quality analyseswould indicate that the area met or didnot meet the standard although it didnot exactly meet the data completenessrequirements.

(2) A provision has been added thata minimal amount of data is neededbefore the requirement to retain highvalues in an incomplete quarter comesinto effect for the annual standards.Sites with at least 11 samples but lessthan 75 percent data completeness in aquarter will have to include high values


if they result in calculated values whichare above the level of the standard. Thisprovision is based upon the change insampling frequency set forth in therevisions to 40 CFR part 58 whicheffectively doubles the minimumsampling frequency from 1-in-6 daysampling to 1-in-3 day sampling. Thedata completeness requirement for theannual form of the standard under theoriginal 1-in-6 day sampling schedule isequivalent to a minimum of 37.5percent under the new samplingschedule of 1-in-3 days. This isequivalent to a minimum of 11 samplesin each quarter. Therefore, a minimumof 11 samples in a quarter should besufficient for an annual average abovethe level of the standard to be usedunder the new sampling schedule.

(3) In sharp contrast, this minimumrequirement was consideredunnecessary for the 24-hour form of thestandard when the 98th percentile isabove the level of the standard. That is,for a site with a 98th percentile above thelevel of the standard that does not meetthe 75 percent data completenessrequirement, the 98th percentile wouldbe equivalent to the maximum orsecond maximum daily concentration inthat year. While adding more datasamples up to the minimum datacompleteness requirement of 75 percentcould help to ensure that the secondmaximum value (rather than themaximum value) corresponds to the 98th

percentile, this difference is notconsidered significant enough to requiresome minimal number of data sampleswhen dealing with the form of the 24-hour standard.

With regard to the 24-hour PM2.5

standard, the proposed revision to 40CFR part 50, Appendix K defined the98th percentile as the daily value out ofa year of monitoring data below which98 percent of all values in the group fall.The calculation of the percentile formhas been revised to reflect generalcomments that the form of the standardand its calculation should be simplified.The EPA maintains that the revisedcalculation is consistent with thedefinition of the percentile being thatnumber below which a certain percentof the data fall.

Regarding the expression of theannual standard to the nearest 0.1 µg/m3

and the 24-hour standard to the nearest1 µg/m3, virtually no commentersdisagreed with the EPA’s proposedapproach. The few that did, however,took issue with the overall stringency ofthe standards, not the rationalediscussed in the proposal. The EPAmaintains its position that instrumentsensitivity and the number of measuredvalues used in calculating the values to

be compared to the standard, asdiscussed at length in the proposal,point to keeping the expressions of thestandards stated in this unit.

B. PM10 Computations and DataHandling Conventions

As discussed in Unit II.G. of thispreamble, the EPA proposed retainingthe current annual arithmetic mean,averaged over 3 years, as the form of theannual PM10 standard, and changing theform of the 24-hour PM10 standard to a98th percentile value form, averagedover 3 years. As discussed in Unit II.G.of this preamble, the form of the dailyPM10 standard has been revised to a 99th

percentile instead of the 98th percentile,and the related calculations have beenrevised accordingly. The same revisiondescribed above in Unit V.A. of thispreamble to simplify the formula usedto calculate the percentile form of the24-hour PM2.5 standard also applies tothe PM10 99th percentile calculation.

The revisions made to the annual and24-hour PM2.5 standards regarding the75 percent data completenessrequirement also apply to the annualand 24-hour PM10 standards. AppendixN of 40 CFR part 50 reflects this change.

As with the PM2.5 standards, the EPAmaintains its position that instrumentsensitivity and the number of measuredvalues used in calculating the values tobe compared to the standard, asdiscussed in detail in the proposal,point to keeping the expressions of thestandards to the nearest 1 µg/m3 for theannual standard and to the nearest 10µg/m3 for the 24-hour standard.

C. Changes That Apply to Both PM2.5

and PM10 ComputationsIn the proposal, the EPA stated that

revisions to 40 CFR part 50, AppendixK would not address the treatment ofexceptional events data, which areconsidered part of the standardsimplementation process. Since severalcommenters mentioned the handling ofthese events in conjunction with theproposed revisions to Appendix K, theEPA has addressed this concern inAppendix N of 40 CFR part 50, whichstates that whether to exclude, retain, ormake adjustments to data affected byuncontrollable or natural events issubject to the approval of theappropriate Regional Administrator.

Comments were also receivedexpressing the desire of some areas toconduct seasonal sampling, reducingthe frequency of monitoring during aperiod of expected low concentrationsto save resources. The proposed revisionto 40 CFR part 50, Appendix K did notprohibit this course of action, andreferred matters of sampling frequency

to 40 CFR 58.13. For clarification, 40CFR part 50, Appendix N adds thatexceptions to specified samplingfrequencies, such as a reducedfrequency during a season of expectedlow concentrations, shall be subject tothe approval of the appropriate RegionalAdministrator.

VI. Reference Methods for theDetermination of Particulate Matter asPM10 and PM2.5 in the Atmosphere

A. Revisions to 40 CFR Part 50,Appendix J—Reference Method for PM10

Because the revocation of the existingPM10 standards will become effective ata later date (as discussed in Unit VII. ofthis preamble), EPA is retainingAppendix J in its current form. A newAppendix M to 40 CFR part 50establishes the reference method formeasuring PM10 in the ambient air forthe revised PM10 standards. Thediscussion in this unit sometimes refersto the contents of the new Appendix Mas revisions to Appendix J, so as tohighlight how the new Appendix Mdiffers from the current Appendix J. Asdiscussed below, the only revision tothe Reference Method for PM10 relates tothe calculation of the volume of airsampled.

During the course of this standardsreview, EPA has received a number ofcomments regarding the appropriatenessof the current practice of adjustingmeasured PM10 concentrations to reflectstandard conditions of temperature andpressure (25° C and 760 mm Hg,respectively), as required by 40 CFR part50, Appendix J. The practice wasoriginally adopted to provide a standardbasis for comparing all pollutantsmeasured in terms of mass per unitvolume (e.g., µg/m3). As EPA hasreviewed the ambient standards forgaseous pollutants, however, technicalchanges have been made to expressthem on a pollutant volume/air volumebasis (i.e., ppm) that is insensitive todifferences in altitude and temperature.Such an approach is not applicable toparticulate pollutants. The questionarises whether continuing the pastpractice of making temperature andpressure adjustments for PM isappropriate or necessary.

Information in the Criteria Documenton the health and welfare effects of PMprovides no clear basis for making suchadjustments. Recent health effectsstudies have been conducted in cooland warm climates, and in cities at highaltitude, e.g., Denver, as well as near sealevel, e.g., Philadelphia (U.S. EPA,1996a). These studies provide noevidence that risk associated with PMexposures is affected by variations in


altitude. Accordingly, any effect thatwould be accounted for by temperatureand pressure adjustments would bebelow the detection limits ofepidemiological studies. While extremesof altitude might be expected to increasethe delivered dose of PM in those notacclimatized to such locations, thedosimetric studies summarized in theCriteria Document provide no clearsupport for any quantitative adjustmentto standard conditions. With respect towelfare effects, visibility is directlyrelated to the actual mass of fineparticles in the atmosphere. Adjustmentof PM concentrations collected at higheraltitudes to standard conditions wouldtherefore lead to an overstatement of theeffect of PM on visibility in suchlocations. Similarly, there is noevidence in the Criteria Documentsuggesting that effects on materialsdamage and soiling are dependent onaltitude.

Based on this assessment, EPAproposed to delete the requirement toadjust PM10 concentrations to standardconditions of temperature and pressurefrom 40 CFR part 50, Appendix J for therevised standards and to makecorresponding revisions in 40 CFR 50.3.Comments received on this issue weredivided. A number of commentorssupported EPA’s proposal for thereasons set forth above. A few Statesopposed the change because the lack ofadjustment for very cold temperature inareas near sea level could make thestandard more stringent. Somecommentors were concerned that theproposed change would relax protectionafforded for areas at high altitude. A fewcommentors expressed concern that‘‘sojourners’’ who visit high altitudearea would have higher ventilation ratesand receive reduced protection ascompared to local residents whoseventilation patterns were more adaptedto these conditions.

The EPA does not believe that thelocalized comparisons regardingincreased or decreased stringency ofstandards relative to the proposedchange are an appropriate rationale forkeeping the current adjustment fortemperature and pressure. The issue iswhether the available scientificevidence on the health and welfareeffects of PM provides a basis forcontinuing with the traditionaladjustments. The comments withrespect to sojourners at altitude arerelevant, but this issue was consideredin reaching the proposed decision.Furthermore, commentors providedneither laboratory nor epidemiologicevidence that would support theirtheoretical concerns regarding increasedannual or 24-hour PM effects at

altitudes typical of mountainous urbanareas in the United States.

Based on its assessment of theavailable evidence and publiccomments, EPA concludes that acontinuation of the practice of adjustingPM10 concentrations to standardconditions of temperature and pressureis not warranted or appropriate.Accordingly, this requirement is notincluded in 40 CFR part 50, AppendixM and corresponding revisions aremade in 40 CFR 50.3. In addition, EPAis also incorporating the proposedminor modifications to 40 CFR part 50,Appendix J in Appendix M.

B. 40 CFR Part 50, Appendix L - NewReference Method for PM2.5

1. Introduction. A new referencemethod for the measurement of fineparticles (as PM2.5) in the ambient airhas been developed for the primarypurpose of determining attainment ofthe new PM2.5 standards. The method isdescribed in the new 40 CFR part 50,Appendix L, and joins the otherreference methods (or measurementprinciples) specified for other criteriapollutants in other appendices to 40CFR part 50.

In developing the proposed newreference method for PM2.5, EPA staffconsulted with a number of individualsand groups in the monitoringcommunity, including instrumentmanufacturers, academics, consultants,and experts in State and local agencies.The approach and key specificationswere submitted to the CASAC TechnicalSubcommittee for Fine ParticleMonitoring, which held a publicmeeting to discuss the proposed newreference method for PM2.5 and relatedmonitoring issues on March 1, 1996.Comments on the proposed methodwere provided orally and in writing byinterested parties. The TechnicalSubcommittee indicated their overallsatisfaction with the method in a letter(Price, 1996) forwarded by CASAC tothe Administrator.

On December 13, 1996, EPA proposedthe new 40 CFR part 50, Appendix L at61 FR 65676 for public comment. Theproposal described in detail theapproach taken and the designspecifications and performancerequirements for the new PM2.5 sampler.On January 14, 1997, EPA held a publichearing on the proposed new 40 CFRpart 50, Appendix L and associated 40CFR parts 53 and 58 requirements.

2. Basic reference method approach.In addition to the primary purpose ofthe new PM2.5 reference method(determining attainment of thestandards), EPA considered a variety ofpossible secondary goals and objectives

that the PM2.5 reference method mightalso fulfill. Subsequently, variousalternative PM2.5 measurementtechniques were evaluated. From thisanalysis, EPA proposed to base its PM2.5

reference method on a conventionaltype sampler that collects 24-hourintegrated PM2.5 samples on a 47 mmTeflon filter that is subsequentlymoisture and temperature conditionedand analyzed gravimetrically. Thesampler is a low volume sampler thatoperates at a flow rate of 1 cubic meterper hour, for a total sample volume of24 m3 for the specified 24-hour samplecollection period. The sampler is easy tooperate, operates over a wide range ofambient conditions, produces ameasurement that is comparable to largesets of previously collected PM data inexisting databases, and provides aphysical sample that can be furtheranalyzed for chemical composition.

3. Public comments and responses—a. Sampler design. The EPA receivedmany general comments concerning theproposed sampler design. Commenterssuggested the use of a differentindicator, use of a different size cut,inclusion of additional constituents(e.g., acid aerosols, carbon, metals, andsemi-volatiles), and/or use of a multi-filter method. Early in the developmentprocess, design decisions were based onpublic input and the advice of CASACon these and other basic design issues.Other factors affecting the basic designof the method were the need forhistorical continuity, high measurementprecision, and simplicity of operation,all in response to current nationalmonitoring objectives and availableresources. In selecting the basicmeasurement approach, substantialweight was given to maintainingcomparability to PM2.5 samplers, such asthe ‘‘dichotomous sampler,’’ that werewidely used to obtain the data uponwhich the new standards are based.Given this objective, EPA concludes thatthe conventional PM measurementapproach is appropriate and willprovide PM2.5 measurements that arecomparable to the air quality data usedin the health studies that provide thebasis for the PM2.5 standards.

Although the sampler is conventionalin configuration, its design is muchmore sophisticated than that of previousPM samplers. This more sophisticatedsampler, together with improvedmanufacturing and operational qualityassurance, is necessary to achieve themore stringent data quality objectivesestablished for PM2.5 monitoring data.To meet precision requirements, thecritical mechanical components of theinlet, particle size separator, downtube,and upper portion of the filter holder


are specified by design. All otheraspects of the sampler are specified byperformance-based specifications.

Several commenters felt that theportions of the sampler that werespecified by design would stifle furtherimprovements and innovations.Although the EPA specifies methods byperformance whenever possible, for thePM2.5 reference method, development ofadequate performance specifications forinlet aspiration and particle sizediscrimination would have been a verydifficult, costly, lengthy, andproblematic process. Moreover,manufacturer testing of proposed inletand particle size discrimination devicesagainst such performance specificationswould require elaborate specializedfacilities and would be extremely costly.For these reasons, the EPA believes thatspecification of these criticalcomponents by design is a prudent andvery cost-effective way to ensure goodinter-manufacturer and intra-manufacturer precision of the PM2.5

measurements. Therefore, thesecomponents are specified by design, andother aspects of the sampler arespecified by performance, as proposed.Innovations and improved samplers ormeasurement methods are encouragedand provided for as Class II and IIIequivalent methods (see 40 CFR part53).

b. Inlet and impactor design. Severalcommenters addressed the inlet design,noting that the inlet could allowentrance of precipitation and possiblyinsects. In fact, the inlet selected for thesampler has been used effectively formany years to obtain many of the PM2.5

measurements that formed the basis ofthe epidemiological studies. While EPAacknowledges that there have been somereports of intrusion of precipitation, theAgency believes the problem isrelatively minor. Nevertheless, amodification of the inlet has beendeveloped to further reduce thepossibility of precipitation (and possiblysmall insects) reaching the sample filterto damage the PM2.5 sample. Extensivewind tunnel tests have shown nosignificant compromise in the PM2.5

aspiration performance of the modifiedinlet.

In addition, a new provision has beenadded, in 40 CFR part 50, Appendix L,section 7.3.8, to require that thesampling air entrance of the inlet be ata height of 2 ± 0.2 meters above thesupporting surface to help ensurehomogeneous air samples whencollocated samplers of different typesare operated simultaneously.

Other commenters addressed thesharpness of the size cut and how it isobtained, e.g., whether more than two

stages should be used and what size cutshould be used for each stage. Theseaspects were carefully considered inselecting the sampler configuration. Theselection by EPA of the previously usedPM10 inlet established the size cut forthe first stage, and the second stage wasdesigned to be simple, reliable, and lowin cost for user agencies. In EPA’sestimation, the advantages of thisconfiguration outweigh any modestadvantage that might have been gainedby designing a new inlet/separationconfiguration that would further refinethe cut points at each of two (or more)stages.

A few commenters questionedwhether the inlet was wind speeddependent at high wind speeds. Theselected inlet has been shown toperform well up to 24 km/hr with 10 µmaerosols and is expected to perform wellat higher speeds with 2.5 µm aerosols.The EPA again determined that theadvantages of using the selected inletoutweighed the possible minorimprovement in wind-speedcharacteristics that might have beenobtained by a newly-designed inlet.

Some commenters felt that other typesof particle discrimination techniquessuch as cyclones and virtual impactors,should be allowed. Again, thesealternatives were evaluated previouslyand the specified inlet and impactorwere determined to best meet thevarious objectives of the sampler.However, EPA has provided forconsiderations of other particle sizeselection techniques or devices forapproval if incorporated into candidateequivalent methods for PM2.5.

Several commenters addressed theimpactor design, noting that theimpactor should be changed to sharpenthe size-cut characteristic, to addressconcerns regarding possiblecontamination and/or performance lossdue to impactor oil, and to improve easeof access to service. To address the firstconcern, the initial prototype impactorhas been modified slightly to sharpen itssize-cut. The current impactor isdesigned to lower cost and to optimizecut sharpness, loading capacity,manufacturing simplicity,manufacturing quality control,serviceability, and reliability. A reportcontaining the penetration efficiency ofthe impactor is available in Docket No.A-95-54. With regard to impactor oilconcerns, the impactor oil selected hasa very low vapor pressure, and testinghas indicated no contamination of thesample filters with impactor oil. TheEPA believes that the impactor design isas accessible as possible, given thedesign objectives. Some flexibility maybe allowed for manufacturers to develop

improved closure devices or otherexternal modifications. Propermaintenance will, of course, be veryimportant and will be stressed in theassociated operator instruction manualsand in other training and guidancematerials. The EPA has been performingfield and laboratory tests that willprovide detailed guidance for allnecessary preventive maintenance.Proper installation procedures for theoil and the impactor filter, as well as allother maintenance requirements, will beavailable in the quality assuranceprocedures and guidance contained inthe new section 2.12 of Appendix L tobe added to EPA’s Quality AssuranceHandbook for Air PollutionMeasurement Systems (EPA/600/R-94/038b).

c. Anodized aluminum surface. Allinternal surfaces exposed to sample airprior to the filter are required to beanodized aluminum as stated in 40 CFRpart 50, Appendix L, section 7.3.7. Afew commenters expressed concern thatthe anodized aluminum surfaces in highvolume PM10 samplers have shownsubstantial pitting, particularly in theventuri flow control device. Theanodized aluminum surfaces arerequired in the PM2.5 sampler tomaintain comparability to previouslyused samplers. The EPA believes thatthe much lower flow rate in the PM2.5

sampler will greatly reduce the pittingtendency, and the active flow control inthe PM2.5 sampler is not dependent onthe physical dimensions of a criticalorifice as it is in a venturi flow controldevice.

d. Filter for PM2.5 sample collection.The proposed reference method calledfor the sample to be collected on a 47mm Teflon filter. Many of the commentsreceived on the measurement methodconcerned the proposed filter mediumand its performance. Commentersexpressed concerns with the use ofTeflon filters and with the selection ofa single-filter method. Severalcommenters recommended thatalternative filter media be allowed, inmost cases to support speciation and/orto allow the capture of all PMcomponents. Other comments notedpotential advantages of other media inoperating characteristics or chemistryrequirements. Operational concernsexpressed about Teflon filters includedtearing, possible loss of integrity, andhigh cost. Other concerns were thatTeflon is generally not conducive tocarbon analysis, and that Teflon filtersmay not hold deposited PM. Manycommenters recommended use of amulti-filter sampler to support chemicalspeciation in addition to compliancedetermination.


To address some of these generalconcerns about the performance of thespecified filter material, some minorrefinements to the filter specificationsconcerning the filter diameter and thefilter support ring have been made toensure proper performance of the filterin the specified filter holder. Additionalclarifications have been made to themaximum moisture pickup and thefilter weight stability requirements.Although Teflon may preclude certainchemical analyses (e.g., elemental andorganic carbon), the EPA believes thatTeflon filter material is the best overallchoice to meet the objectives ofcompliance monitoring and to providegood measurement precision. Otherfilter media are likely to providereduced gravimetric precision andpreclude more types of subsequentchemical analysis. Additional oralternative samplers or filter types canbe considered as candidate equivalentmethods under 40 CFR part 53 and canbe used for non-compliance monitoring,where necessary.

Compliance monitoring based onmass concentration of PM2.5 is theprimary objective of the referencemethod. Multi-filter capability wouldhave substantially increased the costand complexity of the sampler.However, multi-filter samplers can beconsidered as candidate equivalentmethods. In addition, multi-filtersamplers can be used as special purposemonitors (SPMs) to performcharacterization studies, developcontrol strategies, and conduct otherspecial studies as has been donepreviously for PM10.

In response to numerous commentsreceived on 40 CFR part 50, AppendixL and on the provisions of 40 CFR part58 regarding the need for chemicalspeciation, the EPA is assigning a highpriority to a chemical speciation trendsnetwork through section 105 of the Actgrant allocation program and will issueguidance describing the monitoringmethods and scenarios under whichspeciation should be performed. Theprogram will incorporate additionalPM2.5 samplers that allow for thesimultaneous collection of aerosols onmultiple filter media.

The associated requirement forarchiving filters has been removed from40 CFR part 50, Appendix L, section10.17 and relocated to 40 CFR part 58,Appendix A. This change has beenmade because this is a supplementalmonitoring requirement and not anintegral part of the reference method fordetermining compliance with the PM2.5

NAAQS.Provisions of 40 CFR part 50,

Appendix L have been clarified to apply

not only to a single-sample sampler, butalso to a sequential-sample sampler,provided that all specifications are metand no deviations, modifications, orexceptions are made to the inlet,downtube, impactor, or the upperportion of the filter holder. Samplersthat have minor changes ormodifications in these components,have changes that alter the aerosol’sflow path, or contain other significantdeviations will be required to meet therequirements of Class I equivalentmethods, in the amendments to 40 CFRpart 53. Further, a provision has beenadded to require that sequential samplefilters stored in a sequential sampler beadequately covered and protected fromcontamination during storage periods inthe sampler.

A few commenters expressed concernabout who must carry out filter tests todetermine if they meet the filterspecifications. In response, the filterspecifications have been clarified toindicate that filter manufacturers shouldgenerally carry out most or all of thefilter performance tests in order tocertify that their filters meet the filterspecifications for the PM2.5 referencemethod. In addition, EPA conductsacceptance tests on filters procured forNAMS/SLAMS networks prior todistribution to State and local agencies.

Some commenters requestedadditional information on therequirement that an ID number beattached to each filter. Preliminaryinformation indicates that it is notpractical at this time for either filtermanufacturers or users to print an IDnumber directly on the filter. However,EPA is continuing to pursue this goal.In the meantime, alternative means,such as attaching an appropriate IDnumber to the filter’s storage container,will be necessary. Additional detailsand possible alternative filteridentification methods will be providedin new section 2.12 of the QualityAssurance Handbook for Air PollutionMeasurement Systems.

e. Filter handling/weighing/conditioning requirements. Manycommenters felt that the filter handlingrequirements for collected PM2.5

samples were too burdensome.However, handling of the exposed filterbetween retrieval from the sampler andcommencement of the conditioningperiod is expected to be one of the mostsignificant sources of PM2.5

measurement variability. Thus, EPAconcludes that specific requirements forthis activity are necessary, and thisposition was supported by severalcommenters.

Some commenters felt that thesamples should be kept cold until

analysis to prevent volatile losses. Inresponse to this concern, the restrictionon the maximum temperature exposurefor collected samples has been reducedfrom 32 to 25° C, and a recommendationhas been added for sampler operators tokeep the samples as cool as practicalbetween retrieval from the sampler anddelivery to the conditioningenvironment. Further, the length of timepermitted between retrieval of the filterand post-collection weighing isincreased from 10 to 30 days, providedthat the sample is maintained at 4° C orless between retrieval and the start ofthe conditioning period. The newsection 2.12 of the Quality AssuranceHandbook for Air PollutionMeasurement Systems will provideguidance and techniques for keepingsamples cool during this period andmay suggest devices to documentmaximum temperature exposure of thesample.

Commenters also requested additionalspecifications and guidance for fieldblanks. The EPA will provide additionalclarification and detailed proceduresand guidance regarding field blanks inthe new section 2.12 of the QualityAssurance Handbook for Air PollutionMeasurement Systems.

Other commenters felt that the filterweighing requirements were toorestrictive. Because filter weighing isone of the most significant sources ofPM2.5 measurement variability, specificrequirements and restrictions aredeemed necessary. However, inresponse to some of the concernsexpressed, the proposed requirementthat both pre- and post-weighings becarried out by the same analyst has beenreduced to a non-mandatoryrecommendation. Detailedrecommendations and guidance on filterweighing, based on informationobtained in current field tests, will beprovided in the new section 2.12 of theQuality Assurance Handbook for AirPollution Measurement Systems.

Several commenters questioned thefilter conditioning requirements, withsome requesting a lower humidityrange. Since humidity can profoundlyaffect the weight of the PM2.5 on thefilter, EPA maintains that filterconditioning requirements need to betight to control measurement variabilityand to ensure satisfactory precision. Butin response to at least one of theconcerns, the filter conditioninghumidity requirement has been changedto allow conditioning at a relativehumidity within ±5 RH percent of themean ambient humidity duringsampling (down to a minimum of 20 RHpercent) for samples collected at averageambient humidities lower than 30


percent. The EPA will provide furtherdetails on filter conditioning controls inthe new section 2.12 of the QualityAssurance Handbook for Air PollutionMeasurement Systems.

f. Sampler performance requirements.Several commenters addressed samplerperformance requirements, includingsampler flow control specifications,filter temperature control, samplerperformance under extreme conditions,and data reporting. In response toconcerns that various sampler flowcontrol specifications are too tight, EPAcontends that good flow control isnecessary to maintain uniformsampling, to ensure correct particle sizediscrimination, and to controlmeasurement variability. Samplermanufacturers have been able to meetthe specified flow control requirements,and field studies to date confirm thatprototype samplers are able to meetthese flow control requirements.

In response to comments about theambient temperature plus 3° C filtertemperature control requirement, EPAbelieves that fairly tight control of thesample filter temperature is necessary tominimize losses of semi-volatilecomponents over a wide temperaturerange, and tight temperature control hasbeen strongly recommended by theCASAC. Monitoring of the filtertemperature difference from ambienttemperature is necessary to verify thatthe sampler filter temperature control isfunctioning properly. Testing to dateindicates that the proposed 3° C (aboveambient temperature) limit is somewhatdifficult to meet; however, a 5° C limitcan be reasonably met. Therefore, thefilter temperature control requirementhas been relaxed slightly from 3° C tonot more than 5° C above the concurrentambient temperature. Ambient and filtertemperature sensors will requireperiodic calibration or verification ofaccuracy. In response to a frequentcomment, the method has been clarifiedto indicate that exceedance of the filtertemperature difference limit would notnecessarily invalidate the sample.

In response to concerns about theperformance of the sampler underextreme weather conditions (e.g., highor low temperatures, low pressures,high winds, high or low humidity, fog,dust storms), the EPA has establishedsampler specifications that are intendedto cover reasonably normalenvironmental conditions at about 95percent of expected monitoring sites.Qualification test requirements in 40CFR part 53 address most, if not all, ofthese operational requirements.Specification of the samplerperformance for sites with extremeenvironmental conditions would

substantially raise the cost of thesampler for other users, most of whomdo not require the extra capability.Users requiring operation of samplersunder extreme conditions areencouraged to develop supplementalspecifications for modified samplers tocover those specific conditions. Samplermanufacturers have indicated acommitment to respond to the need formodified samplers for such extremeconditions.

Although concerns were expressedthat the amount of data required to bereported from each sampler is excessive,EPA stresses that only a portion of thedata collected by the sampler needs tobe reported to AIRS. These limited datareporting requirements (i.e., ambientand filter temperature, barometricpressure, sample volume, variation insample run flow rate) are important toestablish or verify the reliability andconfidence of the PM2.5 measurementsand to aid in utilization of those data.The substantial amount of additionaldata generated by the sampler are of useto the site operator to provideconfirmation of a given sample’svalidity, and to aid in troubleshootingshould outlier measurements appear inthe monitoring data. A variety of currentelectronic devices and systems may beused to acquire and handle the data, andthese devices can easily accommodatethe amount of data required to bereported, as well as the additional,optional data. Printers, modemconnections, and alternative data outputconnections or devices are notprecluded.

4. Additional changes. Additionalclarifying changes have also been madethroughout 40 CFR part 50, Appendix L,based on comments received or recentlyobtained field test information. In 40CFR part 50, Appendix L, section 3.1,the lower concentration of the methodhas been revised from 1 to 2 µg/m3,based on the results of field blanksassociated with available field test data.In 40 CFR part 50, Appendix L, section3.3, the sample period specification hasbeen augmented to clarify that ameasured PM2.5 concentration for asample period less than 23 hours that isgreater than the NAAQS level(s) is to beconsidered a valid measurement forcomparison to the NAAQS, even thoughnot valid for other purposes. Sections 4(Accuracy) and 5 (Precision) have beenrevised to properly reflect associatedchanges to the data quality and methodperformance assessment requirementsset forth in 40 CFR part 58, AppendixA.

A provision has been added in 40 CFRpart 50, Appendix L, section 7.4.17 torequire sampler manufacturers to make

available computer software to inputsampler output data and translate thedata into a standard spreadsheet format(since no specific format is specified foroutput of the sample data acquired bythe sampler).

The requirements for the sampler todisplay current flow rate, temperature,filter temperature, and barometricpressure readings have been changed torequire updating of these readings atleast every 30 seconds. This change isbased on operational experience ofprototype samplers in 40 CFR part 50,Appendix L, section 7.4.5.1, and willmake it easier for the operators toperform status checks and calibrations.In 40 CFR part 50, Appendix L, section7.4.8.1, the requirements for the ambienttemperature sensor have been changedto specify an external sensor with apassive sun shield, to provide betteruniformity in the ambient temperaturemeasurements among different types ofreference method samplers. Thereference method has also been clarifiedto indicate that PM2.5 samples for whichthe sampler reported an out-of-specification (FLAG) occurrence duringor after the sample period are notnecessarily invalid, and that suchsamples should be reviewed by a qualityassurance officer (40 CFR part 50,Appendix L, section 10.12). Finally, anew reference has been added in section13 of the Act to provide applicablestandards for meteorologicalmeasurements and measurementsystems.

5. Decision on 40 CFR part 50,Appendix L. After fully considering thepublic comments on the proposed newreference method for PM2.5, EPA hasconcluded that the proposed design andperformance specifications for thereference sampler, with themodifications discussed in this unit,will achieve the design objectives setforth in the proposal and outlinedabove. Therefore, EPA is adopting thesampler and other method requirementsspecified in 40 CFR part 50, AppendixL as the reference method for measuringPM2.5 in the ambient air.

Since proposal, a series of field testshave been performed using prototypesamplers manufactured in accordancewith the proposed design andperformance specifications. The resultsof these field tests confirm that theprototype samplers perform inaccordance with design expectations.Operational experience gained throughthese field tests did, however, identifythe need for minor modifications asdiscussed above in this unit. Inaddition, EPA made other modificationsto the proposed design and performancespecification in response to public


comment as discussed above. As part ofthis process, EPA performed laboratorytests to ensure that the modificationsachieved the intended objective.

While the results of these field testsand laboratory tests were largelyconfirmatory in nature and did notindicate a need to alter the basic designand performance specifications, theydid identify areas that needed furtherrefinement. Given that these tests wereperformed, by necessity, during andafter the close of the public commentperiod and because the results were notavailable for placement in the docketuntil late in the rulemaking process,EPA is announcing, in a separateFederal Register notice being signedtoday, a supplemental comment periodfor the limited purpose of takingcomments on these field and laboratorytest results.

VII. Effective Date of the Revised PMStandards and Applicability of theCurrent PM10 Standards

In summary, the primary andsecondary NAAQS for PM have beenrevised by establishing annual and 24-hour PM2.5 standards; and by changingthe form of the existing 24-hour PM10

standards. The existing PM10 annualstandards have been retained. Section50.3 (reference conditions) of 40 CFRpart 50 has been revised to remove theadjustment of measured PM10

concentrations to standard conditions oftemperature and pressure with respectto the revised PM standards. (AlthoughEPA is retaining the current annualPM10 standards, the revision of 40 CFR50.3 potentially may affect the effectivestringency of the annual standards.) Anew Appendix M has been added to 40CFR part 50 that reflects the revision of40 CFR 50.3. A new Appendix N to 40CFR part 50 has been added to reflectthe forms of the PM2.5 and revised PM10

standards. Finally, a new Appendix L to40 CFR part 50 has been added thatspecifies the reference method formeasuring PM2.5 in the ambient air.

The revised PM NAAQS, the revisionsto 40 CFR 50.3, and the newAppendices M, N, and L to 40 CFR part50 will become effective September 16,1997. Inherent in the establishment ofthis revised set of PM standards andrelated provisions is the revocation ofthe current set of PM10 standards andassociated provisions. To provide for aneffective transition from the existing PMstandards to the revised PM standards—in light of the need to establish PM2.5

monitoring networks, designate areas,and develop control strategies forPM2.5—the Administrator hasdetermined that the effective date of therevocation of the current set of PM10

standards and associated provisionsshould be delayed so that the existingstandards and associated provisions willcontinue to apply for an interim period.The duration of the interim periodwould depend on whether the area inquestion has attained the current PM10

standards, as described below in thisunit.

First, section 172(e) of the Actprovides that, if the Administratorrelaxes a national primary ambient airquality standard, she shall, within 12months after the relaxation, promulgaterequirements applicable to all areas thathave not attained that standard as of thedate of the relaxation. Thoserequirements shall provide for controlsthat are not less stringent than thecontrols applicable to areas designatednonattainment before such relaxation.Although the set of revised PMstandards, viewed as a whole, is morestringent than the set of current PMstandards, it appears that the shift fromthe current PM10 standards to therevised PM10 standards, viewed in andof itself, represents a relaxation of thecurrent PM10 standards. As a result,section 172(e) of the Act requires EPAto issue a rule within 12 months toapply implementation requirements noless stringent than the currentlyapplicable requirements for those areasthat have not yet attained the currentPM10 standard(s) by today’spromulgation. However, the Act doesnot specifically provide how to ensurethat States with current PM10 problemsshould maintain the necessary publichealth protection in the interim betweenpromulgation of a relaxed standard andissuance of a rule under section l72(e)of the Act. For that reason, EPA believesthat it is both necessary and appropriateto defer the effective date of therevocation of the current PM10

standards, for areas that have notattained those standards, until EPAissues the rule called for by section172(e) of the Act.

Second, since it will take many yearsfor States to identify PM problemsunder the revised standards and todevelop effective means for addressingthose problems, EPA believes it isnecessary for even those areas that havealready attained the current PM10

standards (and hence are not subject tothe terms of section 172(e) of the Act)to continue their current PM10

implementation efforts for the purposeof protecting public health in thetransition to implementation of therevised standards.

In order to deal with both of thesecategories of areas—those that are notattaining the current PM10 standardsand those that are in attainment of the

current PM10 standards—EPA is takinga two-pronged approach towardsdeferral of the effective date of therevocation of the current PM10

standards. For those areas that are notattaining the current PM10 standards atthe time of the promulgation of therevised PM10 standards, the currentstandards will continue to apply untilEPA has completed its rulemakingunder section 172(e) of the Act toprevent backsliding in those areas. Thiswill assure that no backsliding canoccur in the interim period between thepromulgation of the revised standardsand the completion of the rulemakingunder section 172(e) of the Act. Forthose areas that are attaining the currentPM10 standards at the time ofpromulgation of the revised PM10

standards, the existing PM10 standardswill continue to apply until the areashave an approved SIP that includes anycontrol measures that had been adoptedand implemented at the State level tomeet the current PM10 NAAQS and havean approved section 110 SIP forpurposes of implementing the revisedPM standards. If an area has alreadyreceived approval of a PM10 SIPembodying all of the measures that hadbeen adopted and implemented at theState level, no further Part D submissionor approval would be necessary. If anarea has already submitted suchmeasures, EPA would need to takeaction to approve them. Finally, if anarea has not yet submitted suchmeasures to EPA for inclusion in theSIP, the area would need to submit themand EPA would need to approve them.This submission and approval wouldserve to satisfy both the area’s remainingsubpart D obligations and, in part, itsnew obligations under section 110(a)(1)of the Act regarding the implementationof the revised PM NAAQS. EPAemphasizes that it is not requiring anapproval of a modeled attainmentdemonstration for the current PM10

NAAQS, only an approval of the controlmeasures that had in fact been adoptedand implemented and that, therefore,were responsible for the area’sattainment of the current PM10

standards.The existing definition of reference

conditions and 40 CFR part 50,Appendices J and K will remain in forceas long as the current PM10 standardsapply to an area. Additional policiesand guidance for assuring an effectivetransition will be set forth in future EPAguidance, policies, and/or rules.

VIII. Regulatory and EnvironmentalImpact Analyses

As discussed in Unit IV of thispreamble, the Clean Air Act and judicial


95 It is worth noting that Federal rules that applynationally also play a role in reducing emissionsgoverned by NAAQS. For instance, EPA rules underTitle II of the Act require reductions in ozone-forming emissions from on and off-road vehiclesand the fuels that power them. When EPA issuessuch rules, it conducts the analysis required underthe RFA. For example, EPA performed regulatoryflexibility analyses for the reformulated gasolinerule issued under section 211(k) of the Act. See 59FR 7716, February 16, 1994.

decisions make clear that the economicand technological feasibility of attainingambient standards are not to beconsidered in setting NAAQS, althoughsuch factors may be considered in thedevelopment of State plans toimplement the standards. Accordingly,although, as described below, aRegulatory Impact Analysis (RIA) hasbeen prepared, neither the RIA nor theassociated contractor reports have beenconsidered in issuing this final rule.

A. Executive Order 12866Under Executive Order 12866, 58 FR

51735 (October 4, 1993), the Agencymust determine whether a regulatoryaction is ‘‘significant’’ and, therefore,subject to Office of Management andBudget (OMB) review and otherrequirements of the Executive Order.The order defines ‘‘significantregulatory action’’ as any regulatoryaction that is likely to result in a rulethat may:

(1) Have an annual effect on theeconomy of $100 million or more oradversely affect in a material way theeconomy, a sector of the economy,productivity, competition, jobs, theenvironment, public health or safety, orState, local, or tribal governments orcommunities.

(2) Create a serious inconsistency orotherwise interfere with an action takenor planned by another Agency.

(3) Materially alter the budgetaryimpact of entitlements, grants, user fees,or loan programs or the rights andobligations of recipients thereof.

(4) Raise novel legal or policy issuesarising out of legal mandates, thePresident’s priorities, or the principlesset forth in the Executive Order.

In view of its important policyimplications, this action has beenjudged to be a ‘‘significant regulatoryaction’’ within the meaning of theExecutive Order. As a result, undersection 6 of the Executive Order, EPAhas prepared an RIA, entitled‘‘Regulatory Impact Analysis forParticulate Matter and Ozone NationalAmbient Air Quality Standards andProposed Regional Haze Rule (July1997).’’ This RIA assesses the costs,economic impacts, and benefitsassociated with potential Stateimplementation strategies for attainingthe PM and O3 NAAQS and theproposed Regional Haze Rule. Changesmade in response to OMB suggestions orrecommendations will be documentedin the public docket and made availablefor public inspection at EPA’s Air andRadiation Docket Information Center(Docket No. A-95-58). The RIA will bepublicly available in hard copy bycontacting the U.S. Environmental

Protection Agency Library at the addressunder ‘‘Availability of RelatedInformation’’ and in electronic form asdiscussed above in ‘‘ElectronicAvailability.’’

B. Regulatory Flexibility AnalysisThe Regulatory Flexibility Act (RFA),

5 U.S.C. 601 et seq., provides that,whenever an agency is required topublish a general notice of rulemakingfor a proposal, the agency must preparean initial regulatory flexibility analysisfor the proposal unless the head of theagency certifies that the rule will not, ifpromulgated, have a significanteconomic impact on a substantialnumber of small entities (section605(b)). The EPA certified the proposedNAAQS rule based on its conclusionthat the rule would not establishrequirements applicable to smallentities and therefore would not have asignificant economic impact on smallentities within the meaning of the RFA.See 61 FR 65638, 65668 (PM proposal)and 61 FR 65716, 65764 (ozoneproposal), both published December 13,1996. Accordingly, the Agency did notprepare an initial regulatory flexibilityanalysis for the proposal, but it didconduct a more general analysis of thepotential impact on small entities ofpossible State strategies forimplementing any new or revisedNAAQS.

At the heart of EPA’s certification ofthe proposed NAAQS rule was theAgency’s interpretation of the word‘‘impact’’ as used in the RFA. Is the‘‘impact’’ to be analyzed under the RFAa rule’s impact on the small entities thatwill be subject to the rule’srequirements, or the rule’s impact onsmall entities in general, whether or notthey will be subject to the rule? In thecase of NAAQS rules, the questionarises because of the congressionallydesigned mixture of Federal and Stateresponsibilities in setting andimplementing the NAAQS.

As EPA explained in the proposal,NAAQS rules establish air qualitystandards that States are primarilyresponsible for meeting. Under section110 and Part D of Title I of the Act,every State develops a StateImplementation Plan (SIP) containingthe control measures that will achieve anewly promulgated NAAQS. States havebroad discretion in the choice of controlmeasures. As the U.S. Supreme Courtnoted in Train v. NRDC, 421 U.S. 60(1975), 95 S. Ct. 1470:

[P]rimary [NAAQS] deal with the quality ofoutdoor air and are fixed on a nationwidebasis at a level which the agency determineswill protect the public health. It is theattainment and maintenance of these

standards which section 110(a)(2)(A) requiresthat State plans provide. In complying withthis requirement, a State’s plan must include‘‘emission limitations’’ which are regulationsof the composition of substances emitted intothe ambient air from such sources as powerplants, service stations and the like. They arethe specific rules to which operators ofpollution sources are subject and which, ifenforced, should result in ambient air whichmeets the national standards.

The Agency is plainly charged by the Actwith the responsibility for setting thenational ambient air standards. Just asplainly, it is relegated to a secondary role inthe process of determining and enforcing thespecific, source-by-source emissionlimitations which are necessary if thenational standards are to be met. Under110(a)(2), the Agency is required to approvea State plan which provides for the timelyattainment and maintenance of the ambientair standards, and which also satisfies thatsections other general requirements. The Actgives the agency no authority to question thewisdom of a state’s choices of emissionlimitations if they are part of a plan whichsatisfies the standards of 110(a)(2) and theAgency may devise and promulgate a plan ofits own only if the State fails to submit animplementation plan which satisfies thosestandards. Section 110(c).

421 U.S. 60 at 78–79 (emphasis inoriginal). In short, NAAQS rulesthemselves do not establish any controlrequirements applicable to smallentities. State rules implementing theNAAQS may establish suchrequirements and the extent to whichthey do depends primarily on eachState’s strategy for meeting theNAAQS.95

To determine the properinterpretation of impact under the RFA,EPA considered the RFA’s statedpurpose, its requirements for regulatoryflexibility analyses, its legislativehistory, the amendments made by theSmall Business Regulatory EnforcementFairness Act of 1996 (SBREFA) (Pub. L.104–121), and caselaw. The EPAconcluded that all of these traditionaltools of statutory construction point inone direction—that an agency isrequired to assess the impact of a ruleon the small entities that will be subjectto the rule’s requirements, because thepurpose of a regulatory flexibilityanalysis is to consider ways of easing oreven waiving a rule’s requirements asthey will apply to small entities,consistent with the statute authorizing


the rule. That purpose cannot be servedin the case of the rules like the NAAQSthat do not have requirements thatapply to small entities.

More specifically, EPA noted that itsinterpretation of ‘‘impact’’ flows fromthe express purpose of the RFA itself.As the RFA’s ‘‘Findings and Purposes’’section (Pub. L. 96–354, section 2)makes clear, Congress enacted the RFAin 1980 out of concern that agencieswere writing one-size-fits-all regulationsthat in fact did not fit the size andresources of small entities. Congressnoted that it is generally easier for bigbusinesses to comply with regulations,and that small businesses are thereforeat a competitive disadvantage incomplying with uniform rules. Congressalso noted that small entities’ relativecontribution to the problem a rule issupposed to solve may not warrantapplying the same requirements to largeand small entities alike. In the RFAitself, Congress therefore stated:

It is the purpose of this Act to establish asa principle of regulatory issuance thatagencies shall endeavor, consistent with theobjectives of the rule and of applicablestatutes, to fit regulatory and informationalrequirements to the scale of the businesses,organizations, and governmentaljurisdictions subject to regulation.

(Pub. L. 96–354, section 2(b))The EPA further noted that the RFA

sections governing initial and finalregulatory flexibility analyses reflectthis statement of purpose. Sections 603and 604 of the RFA require that initialand final regulatory flexibility analysesidentify the types and estimate thenumbers of small entities ‘‘to which theproposed will apply’’ (sections 603(b)(3)and 604(a)(3) of the RFA). Similarly,they require a description of the‘‘projected reporting, recordkeeping,and other compliance requirements ofthe proposal, including an estimate ofthe classes of small entities which willbe subject to the requirement’’ (sections603(b)(4) and 604(a)(4)). At the core ofthe analyses is the requirement thatagencies identify and consider‘‘significant regulatory alternatives’’ thatwould ‘‘accomplish the stated objectivesof applicable statutes and whichminimize any significant economicimpact of the proposal on smallentities’’ (sections 603(c) and 604(a)(5)).Among the types of alternativesagencies are to consider are theestablishment of different ‘‘complianceor reporting requirements or timetables’’for small entities and the exemption ofsmall entities ‘‘from coverage of therule, or any part’’ of the rule (section603(c)(1) and (4) of the RFA). The RFAthus makes clear that regulatoryflexibility analyses are to focus on how

to minimize rule requirements on smallentities.

As EPA further explained, sinceregulatory flexibility analyses are notrequired for a rule that will not have a‘‘significant economic impact on asubstantial number of small entities’’, itmakes sense to interpret ‘‘impact’’ inlight of the requirements for suchanalyses. Regulatory flexibility analyses,as described in this unit, are to considerhow a rule will apply to small entitiesand how its requirements may beminimized with respect to smallentities. In this context, ‘‘impact’’ isappropriately interpreted to mean theimpact of a rule on the small entitiessubject to the rule’s requirements.

The Agency cited two Federal courtcases in support of its interpretation. InMid-Tex Elec. Co-op v. FERC, 773 F.2d327, 342 (D.C. Cir. 1985), petitionersclaimed that the RFA required anagency to analyze the effects of a rule onsmall entities that were not regulated bythe rule but might be indirectlyimpacted by it. Petitioners noted thatthe Small Business Administration(SBA) also interpreted the RFA torequire analysis of a rule’s impact onsmall entities not regulated by the rule,and argued that the court should deferto the SBA’s position in light of itscompliance monitoring role under theRFA. After reviewing the RFA’s‘‘Findings and Purposes’’ section, itslegislative history, and its requirementsfor regulatory flexibility analyses, theMid-Tex court rejected petitioners’interpretation. As the court explained:

The problem Congress stated it discernedwas the high cost to small entities ofcompliance with uniform regulations, andthe remedy Congress fashioned—carefulconsideration of those costs in regulatoryflexibility analyses—is accordingly limited tosmall entities subject to the proposedregulation * * *. [W]e conclude that anagency may properly certify that noregulatory flexibility analysis is necessarywhen it determines that the rule will nothave a significant economic impact on asubstantial number of small entities that aresubject to the requirements of the rule.

Id. at 342. Notably, Congress let thisinterpretation stand when it recentlyamended the RFA in enacting SBREFA.

The EPA also cited a recent caseaffirming the Mid-Tex court’sinterpretation. In United DistributionCompanies v. FERC, 88 F.3d 1105, 1170(D.C. Cir. 1996), the court noted that theMid-Tex court:

* * * conducted an extensive analysis ofRFA provisions governing when a regulatoryflexibility analysis is required and concludedthat no analysis is necessary when an agencydetermines ‘‘that the rule will not have asignificant economic impact on a substantial

number of small entities that are subject tothe requirements of the rule’’.Id., citing and quoting Mid-Tex(emphasis added by United Distributioncourt). The Agency went on to explainthat given the Federal/State partnershipfor attaining healthy air, the proposedNAAQS, if adopted, would not establishany requirements applicable to smallentities. Instead, any new or revisedstandard would establish levels of airquality that States would be primarilyresponsible for achieving by adoptingplans containing specific controlmeasures for that purpose. Theproposed NAAQS rule was thus notsusceptible to regulatory flexibilityanalysis as prescribed by the amendedRFA. Since it would establish norequirements applicable to smallentities, it afforded no opportunity forEPA to fashion for small entities lessburdensome compliance or reportingrequirements or timetables, orexemptions from all or part of the rule.For these reasons, EPA certified that theproposal ‘‘will not, if promulgated, havea significant economic impact on asubstantial number of small entities,’’within the meaning of the RFA. BecauseEPA was not required to prepare aninitial regulatory flexibility analysis forthe rule, it was also not required toconvene a Small Business AdvocacyReview Panel for the rule under section609(b) of the RFA, as added by SBREFA.

Notwithstanding its certification ofthe proposal, EPA recognized that theproposed NAAQS, if adopted, wouldbegin a process of State implementationthat could eventually lead to smallentities having to comply with new ordifferent control measures, dependingon the implementation plans developedby the States. EPA also recognized thatthe Act does not allow EPA to dictate orsecond-guess how States shouldexercise their discretion in regulating toattain any new or revised NAAQS.Under those circumstances, EPAconcluded that the best way to takeaccount of small entity concernsregarding any new or revised NAAQSwas to work with small entityrepresentatives and States to provideinformation and guidance on how Statescould address small entity concernswhen they write their implementationplans.

In line with this approach, as part ofRIA it prepared for the proposedNAAQS, EPA analyzed howhypothetical State plans forimplementing the proposal might affectsmall entities. The analysis wasnecessarily speculative and limited,since it depended on projections aboutwhat States might do several years inthe future and did not take into account


96The SIP requirements of subpart 4 of Part D ofTitle I of the Act apply to SIPs for areas designatedas not attaining NAAQS for PM10. Thoserequirements will not apply to SIPs to implementthe PM2.5 NAAQS. Further, to the extent SIPs forareas in nonattainment with the applicable PM10

NAAQS remain subject to subpart 4 requirements,there will be no incremental change in the impacton sources regulated by the States’ SIPs pursuantto those requirements as a result of thispromulgation.

any new strategies that might bedeveloped and recommended by theFACA subcommittee formed to helpdevise potential strategies forimplementing a new or revised NAAQS(see discussion of RIA and FACAprocess in this document). Nevertheless,the analysis provided as muchinformation on potential small entityimpacts as was reasonably available atthe time of the proposal.

The Agency also took steps to ensurethat small entities’ voices were heard inthe NAAQS rulemaking itself. With JereGlover, Chief Counsel for Advocacy ofthe SBA, EPA convened outreachmeetings modeled on the SBREFA panelprocess to solicit and convey smallentities’ concerns with the proposedNAAQS. Two meetings were held aspart of that process, on January 7 andFebruary 28, 1997, with a totalattendance of 41 representatives ofsmall businesses, small governments,and small nonprofit organizations. Bothmeetings were attended byrepresentatives of SBA and OMB, aswell as of EPA. The key concerns raisedby small entities at those meetingsrelated to the scientific foundation ofthe proposed NAAQS and the potentialcost of implementing it, the sameconcerns raised by other industrycommenters on the proposal. TheAgency produced a report on themeetings to ensure that small entityconcerns were part of the rulemakingrecord when EPA made its finaldecision on the proposal.

In light of States’ pivotal role inNAAQS implementation, EPA alsoundertook a number of additionalactivities to assist and encourage theStates to be sensitive to small entityimpacts as they implement any new orrevised NAAQS. With the SBA, EPAbegan an interagency panel process tocollect advice and recommendationsfrom small entity representatives onhow States could lessen any impacts onsmall entities. The EPA plans to issuematerials in two phases to help Statesdevelop their implementation plans. Inview of States’ discretion inimplementing the NAAQS, thesematerials will mostly take the form ofguidance, which is not subject to theRFA’s requirement for initial regulatoryflexibility analysis. (Under section 603of the RFA, that requirement appliesonly to binding rules that are requiredto undergo notice and commentrulemaking procedures.) But regardlessof the form such materials take, EPA isemploying panel procedures to ensurethat small entities have an opportunityto raise any concerns prior to thematerials being issued in draft form.

To supplement the input the Agencyreceives from the ongoing FACA process(described previously in this document),EPA also added more small entityrepresentatives to the Subcommittee onimplementation of any new or revisedNAAQS. These representatives haveformed a small entity caucus to developand bring to the Subcommittee afocused approach to small entity issues.These new Subcommittee members arealso part of the group in theaforementioned panel process. Bymeans of these various processes, EPAhopes to promote the consideration ofsmall entity concerns and advicethroughout the NAAQS implementationprocess.

In response to the proposal, a numberof commenters questioned EPA’sdecision to certify that the proposedNAAQS will not have a significantimpact on a substantial number of smallentities. Some commenters disagreedwith EPA’s view that the proposedNAAQS would not establish regulatoryrequirements applicable to smallentities. These commenters argued thata number of control requirementsapplicable to small entities wouldautomatically result from promulgationof the proposed NAAQS, such as newreasonable further progress, SIP andFederal Implementation Plan (FIP)requirements. Other commenters statedthat it is possible for EPA to assess theimpacts of the NAAQS revision onsmall entities and that, to a limitedextent, EPA has already done so.Further, a number of commentersargued that EPA has a legal obligationunder the RFA, as amended by SBREFA,to choose a NAAQS alternative thatminimizes the impact on small entities.Some commenters questioned EPA’sinterpretations of the Mid-Tex andUnited Distribution cases. In addition,other commenters stated that EPA’sposition regarding the NAAQS and theRFA is inconsistent with its pastpractice and the legislative history ofthe RFA. Finally, a few commentersnoted that the panel process EPAconducted for the proposed NAAQS didnot satisfy the requirements of SBREFA.

EPA disagrees that promulgation ofthe NAAQS will automatically result incontrol requirements applicable to smallentities that EPA can and must analyzeunder the RFA. As noted previously inthis unit, a NAAQS rule only establishesa standard of air quality that otherprovisions of the Act call on States (orin case of State inaction, the Federalgovernment) to achieve by adoptingimplementation plans containingspecific control measures for thatpurpose. Following promulgation of anew or revised NAAQS, section 110 of

the Act requires States and EPA toengage in a designation process todetermine what areas within eachState’s borders are attaining or notattaining the NAAQS. Under section110 and Parts C and D of Title I of theAct, States then conduct a planningprocess to develop and adopt their SIPS.Depending on an area’s designation forthe particular NAAQS, these and otherTitle I provisions of the Act require aState’s SIP to contain certain controlprograms in addition to the controlmeasures that the State decides are alsoneeded to attain and maintain theNAAQS.

The fact that the Act requires SIPs tocontain certain control programs undercertain circumstances does not meanthat EPA either can or must conduct aregulatory flexibility analysis of a ruleestablishing a NAAQS. Just from thestandpoint of feasibility, EPA cannotknow which areas will be subject towhat mandatory SIP programs untilafter the designation process iscompleted. Beyond that, any mandatorySIP programs are still implemented bythe States, and States have considerablediscretion in how they implement them.For instance, the reasonable furtherprogress requirement under section 172of the Act leaves States broad discretionto determine the rate of progress and thecontrol measures to achieve thatprogress.96 As a result, EPA cannot becertain where and how any mandatoryprograms will be implemented withrespect to small (or large) entities. Muchless can EPA know about how Stateswill exercise their discretion to developadditional controls needed to attain andmaintain the NAAQS.

Even if EPA could know exactly howany mandatory SIP programs wouldapply to small entities, the purpose ofthe RFA is not served by attempting aregulatory flexibility analysis of Stateimplementation of those programs. Asexplained previously in this unit, theRFA and the caselaw interpreting itclearly establish that the purpose of theRFA is to promote Federal agencyefforts to tailor a rule’s requirements tothe scale of the small entities that willbe subject to it. That purpose cannot beserved in the case of a NAAQS rulesince the rule does not establishrequirements applicable to small


97 If and when the Agency issues any rulesaddressing State implementation of any statutorilyrequired actions, EPA would analyze and addressthe impact of those rules on small entities asappropriate under the RFA.

entities. In promulgating a NAAQS, theonly choice before EPA concerns thelevel of the standard, not itsimplementation. While mandatory SIPprograms may ultimately follow frompromulgation of the NAAQS, there isnothing EPA can do in setting theNAAQS to tailor those programs as theyapply to small entities. Whether andhow the programs will apply inparticular nonattainment areas isbeyond the scope of the NAAQSrulemaking and, indeed, beyond EPA’sreach in any rulemaking to the extentthe applicability and terms of theprograms are prescribed by statute.97

Moreover, any mandatory SIP programsare supplemented by discretionary Statecontrols that EPA has no power to tailorunder the RFA or the Act (see Train v.NRDC, quoted previously in this unit).

The commenters’ suggestions forminimizing the potential impact of theNAAQS rule on small entities run afoulof both the RFA and the Act. Somesuggested that EPA set a less stringentstandard (or no standard at all in thecase of PM2.5) to reduce the chance thatsmall entities would become subject tonew or tighter SIP requirements. Otherssuggested that EPA require States toexempt small entities from new ortighter SIP requirements. However, asexplained previously in this document,the RFA neither requires nor authorizesEPA to set a less stringent NAAQS thanthe applicable Clean Air Act provisionsallow in order to reduce potential smallentity impacts. Indeed, the RFAprovides that any means of providingregulatory flexibility to small entities beconsistent with the statute authorizingthe rule. Moreover, even if EPA set aless stringent standard, States could stillexercise their discretion to obtain anyneeded emission reductions from smallentities. As the Supreme Court in Trainv. NRDC made clear, EPA has noauthority to forbid States from obtainingreductions from any particular categoryof stationary sources, including smallentities. See also, Virginia v. EPA, No.108 F.3d 1397, 1408 (D.C. Cir. 1997),quoting Union Electric v. EPA, 427 U.S.246, 269 (1976) (‘‘section 110 left to thestates the power to determine whichsources would be burdened byregulations and to what extent’’).

EPA’s approval of SIPs for the new orrevised NAAQS also will not establishnew requirements, but will insteadsimply approve requirements that aState is already imposing. And again,EPA does not have authority to

disapprove a State’s plan except to theextent that the plan fails to demonstrateattainment and maintenance of theNAAQS as required by Title I of theClean Air Act. In cases where EPApromulgates a FIP, EPA might establishcontrol requirements applicable to smallentities, and in such a circumstance,EPA would conduct the analysesrequired by the RFA.

Some commenters argued that underthe RFA as amended by SBREFA, EPAnow has an obligation to choose thealternative that minimizes the impact onsmall entities when setting the NAAQS.As indicated previously in this unit,EPA disagrees with the commenters’argument for the reasons stated in thisdocument’s discussion of the Agency’sauthority to consider costs and otherfactors not related to public health insetting and revising primary NAAQS. Ina nutshell, both the text and legislativehistory of the RFA make clear that theRFA does not override the substantiveprovisions of the statute authorizing therule, but only requires agencies toidentify and consider ways ofminimizing the economic impact onsmall entities subject to the rule in amanner consistent with the authorizingstatute.

Some commenters disagreed withEPA’s interpretation of the Mid-Tex andUnited Distribution cases. In particular,these commenters noted that in thosecases the relevant regulatory agency,Federal Energy Regulatory Commission(FERC), wholly lacked jurisdiction toregulate the small entities at issue.According to these commenters, EPAdoes have the ability and jurisdiction toregulate small entities in the case of theNAAQS, and therefore EPA’s relianceon Mid-Tex and United Distribution ismisplaced.

The commenters’ attempt todistinguish the FERC cases from theNAAQS rulemaking wholly overlooksthe courts’ reasoning, which in fact fullysupports EPA’s certification of theproposed NAAQS. As describedpreviously in this unit, the Mid-Texcourt exhaustively reviewed therelevant sections of the RFA and itslegislative history. Its analysis revealedthat Congress passed the RFA out ofconcern with one-size-fits-allregulations and fashioned a remedylimited to regulations that apply tosmall entities. This principle is fullyapplicable to the NAAQS, which createsno rule requirements that apply to smallentities.

The fact that FERC had no regulatoryauthority over the small entitiesindirectly affected by its rules played noessential role in the court’s rationale.FERC could (and apparently did in the

Mid-Tex rulemaking) estimate thepotential indirect impact of its rules onsmall entities. Presumably, FERC couldhave also mitigated any indirect impactby changing some aspect of the rule (orelse the small entities would have hadno incentive to sue the agency). Thecourt nevertheless found it unnecessaryfor FERC to do either, based on itsreading of the RFA as limited to analysisof a rule’s impact on the small entitiessubject to the rule’s requirements. Inreaching its decision, the court notedthat requiring agencies to ‘‘considerevery indirect effect that any regulationmight have on small businesses * * * isa very broad and ambitious agenda, * ** that Congress is unlikely to haveembarked on * * * without airing thematter.’’ Mid-Tex, 773 F.d. at 343.

The commenters also overstate EPA’sregulatory authority over small entitieswith respect to the regulation of criteriapollutants. Various provisions of theClean Air Act authorize EPA to regulatevarious types of sources at the Federallevel to accomplish specified goals.However, EPA’s authority to moregenerally regulate sources, includingsmall entities, in the manner of SIPs islimited to instances of State default ofSIP responsibilities. When that occurs,EPA may issue a FIP containing specificcontrol measures, and to the extent aproposed FIP would establish controlmeasures applicable to small entities,EPA would analyze the small entityimpact of those measures as required bythe RFA. In 1994, for example, EPAprepared an initial regulatory flexibilityanalysis when it proposed a FIP for LosAngeles. See 59 FR 23264 (May 5, 1994).

As noted previously in this unit,Congress let the Mid-Tex interpretationstand when it recently amended theRFA in enacting SBREFA. If it haddisagreed with the court’s decision, itwould have revised the relevantstatutory provisions or otherwiseindicated its disagreement when itenacted SBREFA. Instead, Congressactually reinforced the Mid-Tex court’sinterpretation of the RFA in enactingsection 212(a) of SBREFA. That sectionrequires that an agency issue a ‘‘smallentity compliance guide’’ for ‘‘each rule* * * for which an agency is requiredto prepare a final regulatory flexibilityanalysis under section 604’’ of the RFA.The guide is ‘‘to assist small entities incomplying with the rule’’ by‘‘explain[ing] the actions a small entityis required to take to comply’’ with therule (section 212(a) of SBREFA).Obviously, it makes no sense to preparea small entity compliance guide for arule that does not apply to smallentities. Thus SBREFA stands as furtherconfirmation that Congress intended the


98 As commenters pointed out, the RIA for theproposed PM NAAQS does state that ‘‘[t]hescreening analysis * * * provides enoughinformation for an initial regulatory flexibilityanalysis (RFA) if such an analysis were to be done.’’That statement was mistaken and was not made inthe RIA for the proposed ozone NAAQS. While bothRIAs attempted to gauge the potential impact onsmall entities of State implementation of theproposed NAAQS, neither could or did identify anyspecific control or information requirementscontained in the NAAQS rule that would apply tosmall entities. Indeed, both RIAs made clear thatthe impact being analyzed was that of potentialState measures to attain the NAAQS, and that suchan analysis was inherently speculative anduncertain. Thus, the RIAs actually confirm EPA’sstatement in the preambles for the proposedNAAQS that conducting a complete regulatoryflexibility analysis is not feasible for rules settingor revising a NAAQS.

RFA to address only rules that establishrequirements small entities must meet.Since SBREFA’s passage, the UnitedDistribution court has affirmed the Mid-Tex court’s interpretation.

Some commenters noted that EPA’sinformal panel process did not complywith the requirements of SBREFA. TheEPA did not convene a SBREFA panelbecause such a panel is not required forrules like the NAAQS that do not applyto small entities. Under the RFA asamended by SBREFA, since the Agencycertified the proposal, it was notrequired to convene a panel for it.Nevertheless, EPA conducted thevoluntary panel process describedpreviously in this unit, as well as othervoluntary small business outreachefforts. The process could not complywith the analytical requirements of theRFA for the reasons given in this unit.However, it could and did ensure thatEPA heard directly from small entitiesabout the NAAQS proposals.

A few commenters stated that EPA’sview of the NAAQS and the RFA isinconsistent with EPA’s past positionsregarding the RFA and NAAQSrevisions. Some commenters also citedthe RIA for the proposed NAAQS andnoted that this analysis demonstratesEPA’s ability to estimate the impact ofthe NAAQS on small entities, therebyundercutting EPA’s argument that it isnot able to perform a regulatoryflexibility analysis when setting theNAAQS.

Past Federal Register documentsmake clear that the nature of theNAAQS makes a regulatory flexibilityanalysis inapplicable to NAAQSrulemakings. For instance, in 1984, EPAstated that a ‘‘NAAQS for NOx by itselfhas no direct impact on small entities.However, it forces each State to designand implement control strategies forareas not in attainment.’’ See 49 FR6866, 6876 (February 23, 1984); see also,50 FR 37484, 37499 (September 13,1985); 50 FR 25532, 25542 (June 19,1985) (NAAQS for NO2 do not impactsmall entities directly). EPA stated againin 1987 that the NAAQS ‘‘themselves donot contain emission limits or otherpollution controls. Rather, such controlsare contained in state implementationplans.’’ See 52 FR 24634, 24654 (July 1,1987).

EPA has typically performed ananalysis to assess, to the extentpracticable, the potential impact ofretaining or revising the NAAQS onsmall entities, depending on possibleState strategies for implementing theNAAQS. These analyses have providedas much insight into the potential smallentity impacts of implementing revisedNAAQS as could be provided at the

NAAQS rulemaking stage. In someinstances, these preliminary analyseswere described as ‘‘regulatory flexibilityanalys[es]’’ or as analyses ‘‘pursuant tothis [Regulatory Flexibility] Act.’’ See,e.g., 52 FR 24634, 24654 (July 1, 1987);50 FR 37484, 37499 (September 13,1985).

However, these analyses were basedon hypothetical State control strategies,and EPA made the point on variousoccasions that any conclusions to bedrawn from such analyses werespeculative, given that the NAAQSthemselves do not impose requirementson small entities. Although these pastanalyses reflected the Agency’s bestefforts to evaluate potential impacts,they were not regulatory flexibilityanalyses containing the necessaryelements required by the RFA. Theseanalyses, for example, did not describethe reporting, recordkeeping, and othercompliance requirements of theproposed NAAQS rules that wouldapply to small entities, since theNAAQS rules did not apply to smallentities. Nor did they determine howthe proposed NAAQS rules could beeased or waived for small entities. Suchan analysis is not possible in the case ofthe NAAQS. To the extent EPA labeledthese analyses regulatory flexibilityanalyses in the past, that label wasinappropriate. EPA’s current practice isto describe such an analysis moreaccurately as a general analysis of thepotential cost impacts on small entities.See, e.g., 61 FR 65638, 65669, 65747(December 13, 1996) (current O3 and PMNAAQS proposals).98 EPA’s analyticalapproach to small entity impacts of theNAAQS has thus remained consistentover time.

One commenter noted that thelegislative history of the RFA suggeststhat the RFA was intended to apply tothe NAAQS. As noted previously in thisunit, EPA’s reading of both the RFA andSBREFA, based on the language of thestatute as amended and its legislative

histories and applicable caselaw, is thatthe RFA requirements at issue do notapply to the NAAQS. The legislativehistory cited by the commenter does notchange this conclusion.

In fact, the statement by SenatorCulver on which the commenter reliesdoes not indicate that the NAAQSshould be subject to regulatoryflexibility analyses. Rather, SenatorCulver uses the NAAQS as an exampleof the type of standard that agencieswould not change as a result of the RFA.According to Senator Culver, section606 of the RFA ‘‘succinctly states thatthis bill does not alter the substantivestandard contained in underlyingstatutes which defines the agency’smandate.’’ 126 Cong. Rec. S 21455(August 6, 1980) daily ed. After citingsection 109 of the Act, Senator Culvergoes on to describe EPA’s bubble policy(which addresses the limits onemissions from a particular facility) asthe type of flexible regulation thatagencies should consider, once EPA hasset a NAAQS. ‘‘The important point forpurposes of this discussion is that the‘bubble concept,’ a type of flexibleregulation, in no manner altered thebasic statutory substantive standard ofthe EPA * * *. No regulatory flexibilityanalysis alters the substantive standardotherwise applicable by law to agencyaction.’’ Id. Thus, contrary to thesuggestion of the commenter, SenatorCulver’s statement actually confirmsthat the time to consider regulatoryflexibility is when regulationsapplicable to sources are beingestablished, not when a NAAQS itself isbeing set.

Under section 604 of the RFA,whenever an agency promulgates a finalrule under section 553 of theAdministrative Procedure Act, afterbeing required by that section or anyother law to publish a general notice ofproposed rulemaking (NPRM), theagency is required to prepare a finalregulatory flexibility analysis. RFAsection 605(b) provides, however, thatsection 603 (re initial regulatoryflexibility analyses) and section 604 donot apply if the agency certifies that therule will not have a significanteconomic impact on a substantialnumber of small entities and publishessuch certification at the time ofpublication of the NPRM or at the timeof the final rule.

As noted above, EPA certified thisfinal rule at the time of the NPRM. Afterconsidering the public comments on thecertification, EPA continues to believethat this final rule will not have asignificant economic impact on asubstantial number of small entities forthe reasons explained above and that it


99 As noted in unit VIII.B., a NAAQS rule onlyestablishes a standard of air quality that otherprovisions of the Act call on States (or in the caseof State inaction, the Federal government) toachieve by adopting implementation planscontaining specific control measures for thepurpose. Thus, it is questionable whether theNAAQS itself imposes an enforceable duty and thuswhether it is a significant Federal mandate withinthe meaning of UMRA. EPA need not and does notreach this issue in this document. For the reasonsgiven in this unit, even if the NAAQS weredetermined to be a significant Federal mandate,EPA does not have any obligations under sections202 and 205 of UMRA, and EPA has met anyobligations it would have under section 204 ofUMRA.

100In addition to the estimates and assessmentsdescribed in section 202 of UMRA, writtenstatements are also to include an identification ofthe Federal law under which the rule ispromulgated (section 202(a)(1) of UMRA) and adescription of outreach efforts under section 204 ofUMRA (section 202(a)(5) of UMRA). Although theserequirements do not apply here because a writtenstatement is not required under section 202 ofUMRA, this preamble identifies the Federal lawunder which this rule is being promulgated and awritten statement describing EPA’s outreach effortswith State, local, and tribal governments will beplaced in the docket.

therefore appropriately certified therule. Further, as required by the CleanAir Act, EPA is promulgating this finalrule under section 307(d) of the CleanAir Act. For all the foregoing reasons,EPA has not prepared a final regulatoryflexibility analysis for the rule. TheAgency has nonetheless analyzed in thefinal RIA for the rule the potentialimpact on small entities of hypotheticalState plans for implementing theNAAQS. The Agency also plans to issueguidance to the States on reducing thepotential impact on small entities ofimplementing the NAAQS.

C. Impact on Reporting RequirementsThere are no reporting requirements

directly associated with the finalizationof ambient air quality standards undersection 109 of the Act (42 U.S.C. 7400).There are, however, reportingrequirements associated with relatedsections of the Act, particularly sections107, 110, 160, and 317 (42 U.S.C. 7407,7410, 7460, and 7617).

In EPA’s final revisions to the airquality surveillance requirements (40CFR part 58) for PM, the associated RIAaddresses the Paperwork Reduction Actrequirements through an InformationCollection Request.

D. Unfunded Mandates Reform ActTitle II of the Unfunded Mandates

Reform Act of 1995 (UMRA), Pub. L.104–4, establishes requirements forFederal agencies to assess the effects ofcertain regulatory actions on State,local, and tribal governments and theprivate sector. Under section 202 ofUMRA, EPA generally must prepare awritten statement, including a cost-benefit analysis, for proposed and finalrules with ‘‘Federal mandates’’ that mayresult in expenditures by State, local,and tribal governments, in the aggregate,or by the private sector, of $100 millionor more in any 1 year. This requirementdoes not apply if EPA is prohibited bylaw from considering section 202 ofUMRA estimates and analyses inadopting the rule in question. Beforepromulgating a final rule for which awritten statement is needed, section 205of UMRA generally requires EPA toidentify and consider a reasonablenumber of regulatory alternatives andadopt the least costly, most cost-effective, or least burdensomealternative that achieves the objectivesof the rule. These requirements do notapply when they are inconsistent withapplicable law. Moreover, section 205 ofUMRA allows EPA to adopt analternative other than the least costly,most cost-effective, or least burdensomealternative if the Administratorpublishes with the final rule an

explanation of why that alternative wasnot adopted. Before EPA establishes anyregulatory requirements that maysignificantly or uniquely affect smallgovernments, including tribalgovernments, it must have developedunder section 203 of UMRA a smallgovernment agency plan. The plan mustprovide for notifying potentiallyaffected small governments, enablingofficials of affected small governmentsto have meaningful and timely input inthe development of EPA regulatoryproposals with significant Federalintergovernmental mandates, andinforming, educating, and advisingsmall governments on compliance withthe regulatory requirements. Section 204of UMRA requires each agency todevelop ‘‘an effective process to permitelected officers of state, local and tribalgovernments * * * to providemeaningful and timely input’’ in thedevelopment of regulatory proposalscontaining a significant Federalintergovernmental mandate.99

The EPA has determined that theprovisions of sections 202 and 205 ofUMRA do not apply to thisdecision.‘‘Unless otherwise prohibitedby law,’’ EPA is to prepare a writtenstatement under section 202 of UMRAthat is to contain assessments andestimates of the costs and benefits of arule containing a Federal mandate.Congress clarified that ‘‘unlessotherwise prohibited by law’’ referred towhether an agency was prohibited fromconsidering the information in therulemaking process, not to whether anagency was prohibited from collectingthe information. The Conference Reporton UMRA states, ‘‘This section [202]does not require the preparation of anyestimate or analysis if the agency isprohibited by law from considering theestimate or analysis in adopting therule.’’ 141 Cong. Rec. H3063 (daily ed.March 13, 1995). Because the Clean AirAct prohibits EPA, when setting theNAAQS, from considering the types ofestimates and assessments described insection 202 of UMRA, UMRA does notrequire EPA to prepare a written

statement under section 202.100 Therequirements in section 205 of UMRAdo not apply because thoserequirements only apply to rules ‘‘forwhich a written statement is requiredunder section 202 * * *.’’

The EPA has determined that theprovisions of section 203 of UMRA donot apply to this decision. Section 203of UMRA only requires the developmentof a small government agency plan forrequirements with which smallgovernments might have to comply.Since setting the NAAQS does notestablish requirements with whichsmall governments might have tocomply, section 203 of UMRA does notapply. The EPA acknowledges,however, that any correspondingrevisions to associated SIP requirementsand air quality surveillancerequirements, 40 CFR parts 51 and 58,respectively, might result in sucheffects.Accordingly, EPA did addressunfunded mandates when it proposedrevisions to 40 CFR part 58, and will doso, as appropriate, when it proposes anyrevision to 40 CFR part 51.

With regard to the outreach describedin section 204 of UMRA, EPA did followa process for providing elected officialswith an opportunity for meaningful andtimely input into the proposed NAAQSrevisions, although EPA did notdescribe this process in the proposal.The EPA conducted a series of pre-proposal outreach meetings with Stateand local officials and theirrepresentatives that permitted theseofficials to provide meaningful andtimely input on issues related to theNAAQS and the monitoring issuesassociated with them. Beginning inJanuary, 1996, EPA briefed State andlocal air pollution control officials atnational meetings with State andTerritorial Air Pollution ProgramAdministrators (STAPPA) / Associationof Local Air Pollution Control Officials(ALAPCO) in Washington, DC, NorthCarolina, Chicago, and Nevada. TheEPA also held briefings for theWashington, DC representatives ofseveral State and local organizations,including National Conference of StateLegislators, U.S. Conference of Mayors,


101 One commenter argued that in reviewing theSO2 NAAQS, EPA determined that it need notrevise the S02 NAAQS, but could instead pursue analternative regulatory program under otherauthority. This commenter argued that EPA hassimilar flexibility in reviewing the PM and OzoneNAAQS, and thus UMRA requires EPA to identifythe least burdensome alternative (such as retainingthe current NAAQS) as part of that process. Asdiscussed more fully in Unit IV. of this preamble,EPA does not agree that it has flexibility to choosesuch an alternative; nor does EPA agree with thecommenter’s characterization of the action it tookin deciding not to revise the SO2 NAAQS. In fact,in deciding not to revise the SO2 NAAQS, EPAdetermined, for reasons independent of section 303of the Clean Air Act that a NAAQS revision was notwarranted. See 61 FR 25566, 25575 (May 22, 1996).

National Governors Association,National League of Cities, and STAPPA/ALAPCO. EPA also held separatebriefings and discussions with State andlocal officials at meetings set up by theNational Governors Association, theU.S. Conference of Mayors and theCouncil of State Governments. The EPAalso conducted in-depth briefings ateach EPA regional office and regionalstaff also had several meetings anddiscussions with their Statecounterparts about the standards. Theefforts described in this paragraph ofthis preamble, which provided electedofficials with opportunity formeaningful and timely input into theproposed NAAQS revisions, met anyrequirements imposed by section 204 ofUMRA. The docket will contain awritten statement describing theseoutreach efforts, including a summary ofthe comments and concerns presentedby State, local, and tribal governmentsand a summary of EPA’s evaluation ofthose comments and concerns.

Several commenters disagreed withEPA that sections 202, 203, and 205 ofUMRA do not apply to this decision.These commenters argued that EPA isnot prohibited from considering costs insetting NAAQS under the Clean Air Actand applicable judicial decisions. Somecommenters also expressed the viewthat there is no conflict between UMRAand the Clean Air Act with regard to theNAAQS. These commenters argued thatUMRA and the NAAQS can beharmonized by reading UMRA as aninformation gathering statute and thatEPA should therefore perform theanalyses required by UMRA, regardlessof whether costs may be considered.Finally, at least one commenter arguedthat in past NAAQS reviews, EPA didnot dispute its UMRA obligations.

As discussed more fully in Unit IV. ofthis preamble, EPA is prohibited fromconsidering cost in setting the NAAQS.Given that fact (as noted in Unit IV. ofthis preamble), sections 202 and 205 ofUMRA do not apply.101 As theConference Report clarifies, UMRA

itself states that the section 202estimates and analyses are not requiredin cases such as the NAAQS, where anagency is prohibited by law fromconsidering section 202 estimates andanalyses. Reading UMRA in the mannersuggested by the commenters wouldeffectively read this provision out ofUMRA; UMRA contains an exceptionfor rules like the NAAQS, it must begiven effect.

With regard to EPA’s positionregarding UMRA in previous NAAQSreview exercises, EPA simply madeplain in those situations that because itdid not plan on revising the NAAQS, itdetermined, without further review, thatsections 202, 203, and 205 of UMRA didnot apply. EPA thus stated that:

Because the Administrator has decided notto revise the existing primary NAAQS forSO2, this action will not impose any newexpenditures on governments or on theprivate sector, or establish any newregulatory requirements affecting smallgovernments. Accordingly, EPA hasdetermined that the provisions of sections202, 203 and 205 do not apply to this finaldecision.

61 FR 25566, 25577, May 22, 1996; seealso 61 FR 52852, 52856, October 8,1996 (Same statement for NO2 NAAQS).As this statement makes clear, EPA onlydetermined that sections 202, 203, and205 of UMRA did not apply to theNAAQS when EPA fails to revise thestandard. Having made thatdetermination, EPA had no reason tocatalog additional bases for findingUMRA inapplicable. Nothing in thatstatement was intended to precludeEPA, or precludes EPA, fromconcluding for other reasons (such asthose discussed in this unit) that UMRAalso does not apply when EPA in factrevises an applicable NAAQS.

E. Environmental JusticeExecutive Order 12848 (58 FR 7629,

February 11, 1994) requires that eachFederal agency make achievingenvironmental justice part of its missionby identifying and addressing, asappropriate, disproportionately highand adverse human health orenvironmental effects of its programs,policies, and activities on minoritiesand low-income populations. Theserequirements have been addressed tothe extent practicable in the RIA citedin this unit.

F. Submission to Congress and theComptroller General

Under 5 U.S.C. 801(a)(1)(A), as addedby the Small Business RegulatoryEnforcement Fairness Act of 1996(SBREFA), EPA submitted a reportcontaining this rule and other required

information to the U.S. Senate, the U.S.House of Representatives, and theComptroller General of the UnitedStates prior to publication of the rule inthis issue of the Federal Register. Thisrule is a ‘‘major rule’’ for purposes ofSBREFA.

IX. Response to Petition forAdministrator Browner’s Rescusal

On March 13, 1997, the WashingtonLegal Foundation (WLF), filed a petitionwith EPA asking that I, Carol Browner,disqualify myself in rulemakingregarding the NAAQS for PM andozone. The petition claims that mypublic statements indicate a ‘‘clear andconvincing showing’’ that I had‘‘already decided to revise the NAAQSfor PM and ozone’’ and that I therefore‘‘could not give meaningfulconsideration‘‘ to comments adverse tothe proposed rule. On May 12, 1997,EPA’s General Counsel, Jonathan Z.Cannon, sent a letter to WLF regardingthe petition. This letter and the WLFpetition were then placed in the docketsfor the proposed ozone and PMstandards pending ‘‘consideration andfinal response in connection with theAgency’s final actions.’’

Contrary to WLF’s assertions, I havemaintained an open mind throughoutthese proceedings, and have basedtoday’s decisions on the rulemakingrecord—including consideration ofcomments opposed to the proposal. Thelaw does not require the Administratorof EPA to disqualify herself merely forexpressing views on a proposedregulation; in fact, it is part of myresponsibility to engage in the publicdebate on the proposals. Moreover, theassertions in WLF’s petition do notaccurately represent my views. Thepetition takes quotes out of context andrepeatedly misinterprets my statements.For example, WLF quotes a statementthat I made at the Children’sEnvironmental Health NetworkResearch Conference as an indicationthat I had ‘‘prejudged the issue.’’However, my statement that ‘‘I will notbe swayed’’ did not refer to adopting theNAAQS as proposed. Instead, as is clearfrom reviewing the entire speech, I wasaddressing my broader concern aboutchildren’s health and the range of EPAstandards affecting children’s health. Ialso appeared at several congressionalhearings and testified before members ofCongress, some of whom were stronglyopposed to the proposals. At thosehearings, I explained the basis for theproposals and put forward the reasonswhy I concluded the proposals wereappropriate, given the informationbefore me at the time. At the same time,I made clear that I took very seriously


my obligation to keep an open mind,and to consider fully and fairly allsignificant comments that the Agencyreceived. For these reasons and others,as set forth in Mr. Cannon’s May 12,1997 response to WLF, which I adopt infull, I have decided not to recuse myselffrom any aspect of considering revisionsto the NAAQS for ozone and PM.Accordingly, I am hereby denyingWLF’s petition.

X. References

(1) Abbey, D.E.; Mills, P.K.; Petersen, F.F.;Beeson, W.L. (1991) Long-term ambientconcentrations of total suspendedparticulates and oxidants as related toincidence of chronic disease in CaliforniaSeventh-Day Adventists. EnvironmentalHealth Perspectives 94:43–50.

(2) Abt Associates (1996a) ProposedMethodology for PM Risk Analyses inSelected Cities (Draft). Prepared by AbtAssociates, Inc., Hampden Square, Suite 500,4800 Montgomery Lane, Bethesda, MD20814–5341. February 12, 1996.

(3) Abt Associates (1996b) A ParticulateMatter Risk Analysis for Philadelphia andLos Angeles. Prepared by Abt Associates forUS EPA, OAQPS, Hampden Square, Suite500, 4800 Montgomery Lane, Bethesda, MD20814–5341. July 3, 1996 (UpdatedNovember 1996).

(4) Abt Associates (1997a) Revision ofMortality Incidence Estimates Based on Popeet al. (1995) in the Abt Particulate MatterRisk Assessment Report. Prepared by AbtAssociates, Inc., Hampden Square, Suite 500,4800 Montgomery Lane, Bethesda, MD29814–5341. June 5, 1997.

(5) Abt Associates (1997b) Revision ofMortality Incidence Estimates Based on Popeet al. (1995) in the Abt Particulate MatterRisk Assessment Report. Prepared by AbtAssociates, Inc., Hampden Square, Suite 500,4800 Montgomery Lane, Bethesda, MD29814–5341. June 6, 1997.

(6) American Industrial Health Council.(1997) Comments on EPA’s Proposed Rule forNational Ambient Air Quality Standards forParticulate Matter. Docket No. A–95–54, IV–D–2340. March 3, 1997.

(7) American Petroleum Institute. (1997)Comments of the American PetroleumInstitute on the Proposed National AmbientAir Quality Standards for Particulate Matter.Docket No. A–95–54, IV–D–2247. March 12,1997.

(8) Borja-Aburto, V.H.; Loomis, D.P.;Bangdiwala, S.I.; Shy, C.M.; Rascon-Pacheco,R.A. (1997) Ozone, suspended particulates,and daily mortality in Mexico City. AmericanJournal of Epidemiology 145:258–268.

(9) Braun-Fahrlander, C.; Ackermann-Liebrich, U.; Schwartz, J.; Gnehm, H.P.;Rutishauser, M.; Wanner, H.U. (1992) Airpollution and respiratory symptoms inpreschool children. American Review ofRespiratory Disease 145:42–47.

(10) Brown, K. (1997) Predictability ofpersonal exposure to airborne particulatematter from ambient concentrations in fourcommunities. In: Comments of the AmericanPetroleum Institute on the proposed nationalambient air quality standards for particulate

matter (Appendix F). Docket No. A–95–54,IV–D–2247.

(11) California Environmental ProtectionAgency. (1997) Comments on the proposednational ambient air quality standards.Docket No. A–95–54, IV–D–2251. March 11,1997.

(12) Chestnut, L.G.; Rowe, R.D. (1990)Preservation values for visibility in thenational parks. Washington, DC: U.S.Environmental Protection Agency.

(13) Chestnut, L.G.; Dennis, R.L.; Latimer,D.A. (1994) Economic benefits ofimprovements in visibility: acid rainprovisions of the 1990 Clean Air ActAmendments. Proceedings of the aerosolsand atmospheric optics: radiative balanceand visual air quality. Air & WasteManagement International SpecialtyConference, pp. 791–802.

(14) Cooper, J.A.; Tawney, C.W. (1996)Comments on Office of Air Quality Planningand Standards recommended PM2.5

concentrations equivalent to current PM10

standards. In: Comments of the AmericanIron and Steel Institute. Docket No. A–95–54,IV–D–2242. March 12, 1997.

(15) Costa, D.L.; Dreher, K.L. (1997)Bioavailable transition metals in particulatematter mediate cardiopulmonary injury inhealthy and compromised animal models.Environmental Health Perspectives (In press)

(16) Damberg, R.; Polkowsky, B. (1996)Methodology for Estimating VisibilityImprovements Due to Reductions in FineParticle Concentrations. Memorandum to thedocket for the PM National Ambient AirQuality Standards review. September 1996.

(17) Davis, J.M.; Sacks, J.; Saltzman, N.;Smith, R.L.; Styer, P. (1996) Airborneparticulate matter and daily mortality inBirmingham, Alabama. Technical Report #55,National Institute of Statistical Sciences, P.O.Box 14162, Research Triangle Park, NC27709.

(18) Delfino RJ, Murphy-Moulton AM,Burnett RT, Brook JR, Becklake MR. (1997)Effects of air pollution on emergency roomvisits for respiratory illnesses in Montreal,Quebec. American Journal of Respiratory andCritical Care Medicine 155:568–576

(19) Dickey, J.H. (1997) Letter to PresidentBill Clinton. Docket No. A–95–54, IV–D–2301. March 12, 1997.

(20) Dockery, D.W.; Cunningham, J.;Damokosh, A.I.; Neas, L.M.; Spengler, J.D.;Koutrakis, P.; Ware, J.H.; Raizenne, M.;Speizer, F.E. (1996) Health effects of acidaerosols on North American children:respiratory symptoms. Environmental HealthPerspectives 104:500–505.

(21) Dockery, D.W.; Pope, C.A., III; Wu, W.;Spengler, J.D.; Ware, J.H.; Fay, M.E.; Ferris,B.G., Jr.; Speizer, F.E. (1993) An associationbetween air pollution and mortality in sixU.S. cities. New England Journal of Medicine329:1753–1759.

(22) Dockery, D.W.; Schwartz, J.; Spengler,J.D. (1992) Air pollution and daily mortality:associations with particulates and acidaerosols. Environmental Research 59:362–373.

(23) Dockery, D.W.; Speizer, F.E.; Stram,D.O.; Ware, J.H.; Spengler, J.D.; Ferris, B.G.,Jr. (1989) Effects of inhalable particles onrespiratory health of children. AmericanReview of Respiratory Disease 139:587–594

(24) Fairley, D. (1990) The relationship ofdaily mortality to suspended particulates inSanta Clara County, 1980–86. EnvironmentalHealth Perspectives 89:159–168.

(25) Fitz-Simons, T.; Mintz, D.; Wayland,M. (1996) Proposed methodology forpredicting PM2.5 from PM10 values to assessthe impact of alternative forms and levels ofthe PM National Ambient Air QualityStandards. Document transmitted tomembers of the Clean Air Act ScientificAdvisory Committee on June 26, 1996.

(26) Freas, W.P. (1997) Memorandum tothe files: Ratios of Particulate Matter AnnualArithmetic Mean to Annual Geometric MeanConcentrations. Docket No. A–95–54, IV–B–2. May 29, 1997.

(27) Friedlander, S.K. (1982) Letter fromSheldon K. Friedlander, Chair, Clean AirScience Advisory Committee toAdministrator Anne. M. Gorsuch. Clean AirScientific Advisory Committee Review andClosure of the Office of Air Quality Planningand Standards Staff Paper for ParticulateMatter. January 29, 1982.

(28) Gamble, J.F.; Lewis, R.J. (1996) Healthand respirable particulate (PM10) airpollution: a causal or statistical association?Environmental Health Perspectives 104:838–850.

(29) Grand Canyon Visibility TransportCommission (1996) Recommendations forimproving Western vistas. Report of theGrand Canyon Visibility TransportCommission to the U.S. EnvironmentalProtection Agency. June 1996.

(30) Godleski, J.J.; Sioutas, C.; Katler, M.;Catalano, P.; Koutrakis, P. (1996) Death frominhalation of concentrated ambient airparticles in animal models of pulmonarydisease. Proceedings of the Second AnnualColloquium on Particulate Air Pollution andHuman Health. Lee, J.; Phalen, R.; eds. 4:136–143.

(31) Gordian, M.E.; Ozkaynak, H,; Xue, J.;Morris, S.S.; Spengler, J.D. (1996) Particulateair pollution and respiratory disease inAnchorage, Alaska. Environmental HealthPerspectives 104:290–297.

(32) Hefflin, B.J.; Jalaludin, B.; McClure, E.;Cobb, N.; Johnson, C.A.; Jecha, L.; Etzel, R.A.(1994) Surveillance for dust storms andrespiratory diseases in Washington State,1991. Archives of Environmental Health49:170–174.

(33) Health Effects Institute. (1997)Particulate Air Pollution and Daily Mortality:Analyses of the Effects of Weather andMultiple Air Pollutants (The Phase I.B Reportof the Particle Epidemiology EvaluationProject). Health Effects Institute, Cambridge,MA.

(34) Health Effects Institute. (1995)Particulate Air Pollution and Daily Mortality:Replication and Valication of SelectedStudies (The Phase I.A Report of the ParticleEpidemiology Evaluation Project). HealthEffects Institute, Cambridge, MA.

(35) Hill, A.B. (1965) The environment anddisease: associations or causation?Proceedings of the Royal Society of Medicine58:295–300.

(36) Ito, K.; Kinney, P.L.; Thurston, G.D.(1995) Variations in PM10 concentrationswithin two metropolitan areas and theirimplications for health effects analyses.Inhalation Toxicology 7:735–745.


(37) Katsouyanni, K.; Touloumi, G.; Spix,C.; Schwartz, J.; Balducci, F.; Medina, S.;Rossi, G.; Wojtyniak, B.; Sunyer, J.;Bacharova, L.; Schouten, J.P.; Ponka, A.;Anderson, H.R. (1997) Short-term effects ofambient sulphur dioxide and particulatematter on mortality in 12 European cities:results from the APHEA project. BritishMedical Journal (In press)

(38) Killingsworth, C.R.; Alessandrini, F.;Krishna Murthy, G.G.; Catalano, P.J.;Paulauskis, J.D.; Godleski, J.J. (1997)Inflammation, chemokine expression, anddeath in monocrotaline-treated rats followingfuel oil fly ash inhalation. InhalationToxicology (In press)

(39) Koman, P.D. (1996) Memorandum tothe files regarding PM2.5 Air Quality Valuesfrom PM Health Studies. September 30, 1996.

(40) Koman, P.D. (1997) SupplementalInformation Regarding PM2.5 Air QualityValues from Key Short-term Fine ParticleExposure Studies. June 30, 1997.

(41) Li, Y.; Roth, H.D. (1995) Dailymortality analysis by using differentregression models in Philadelphia County,1973–1990. Inhalation Toxicology 7:45–58.

(42) Lipfert, F.W., Wyzga, R.E. (1997) Airpollution and mortality: the implications ofuncertainties in regression modeling andexposure measurement. Journal of Air &Waste Management 47:517–523.

(43) Lipfert, F.W.; Urch, R.B. (1997)Personal exposures of asthmatics to airpollution in Toronto and implications forepidemiological studies. Submitted asattachment to comments on the proposedNational Ambient Air Quality Standard forparticulate matter by R.E. Wyzga, Docket No.A–95–54, IV–D–2672. March 11, 1997.

(44) Lipfert, F.W. (1995) Estimating airpollution-mortality risks from cross-sectionalstudies: prospective vs. ecologic studydesigns. In: Particulate matter: health andregulatory issues: proceedings of aninternational specialty conference; April;Pittsburgh, PA. Pittsburgh, PA: Air & WasteManagement Association; p. 78–102. (Air &Waste Management Association PublicationVIP–49)

(45) Lipfert, F.W.; Wyzga, R.E. (1995) Airpollution and mortality: issues anduncertainties. Journal of Air & WasteManagement Association 45:949–966.

(46) Lippmann, M.; Shy, C.; Stolwijk, J.;Speizer, F. (1996) Letter to AdministratorCarol M. Browner. Supplement to the ClosureLetter from the Clean Air Scientific AdvisoryCommittee. March 20, 1996.

(47) Lippman, M. (1986) Letter fromMorton Lippmann, Chair, Clean Air ActScientific Advisory Committee, toAdministrator Lee M. Thomas.

(48) Lipsett, M.; Hurley, S.; Ostro, B. (1997)Air Pollution and emergency room visits forasthma in Santa Clara County, California.Environmental Health Perspectives105(2):216–222

(49) Lyon, J.L.; Mori, M.; Gao, R. (1995) Isthere a causal association between excessmortality and exposure to PM10 air pollution?Additional analyses by location, year, seasonand cause of death. Inhalation Toxicology7:603–614.

(50) Martin, A.E.; Bradley, W.H. (1960)Mortality, fog and atmospheric pollution: an

investigation during the winter of 1958–1959.Monthly Bulletin of the Minister of Health,Public Health Laboratory Service (GB) 19:56–73.

(51) Ministry of Health - London- HerMajesty’s Stationary Office (1954) Reports onPublic Health and Medical Subjects No. 95.Mortality and Morbidity During the LondonFog of December 1952.

(52) Moolgavkar, S.H.; Luebeck, E.G.;Anderson, E.L. (1997) Air pollution andhospital admissions for respiratory causes inMinneapolis-St. Paul and Birmingham.Epidemiology 8:364–370.

(53) Moolgavkar, S.H.; Luebeck, E.G. (1996)Particulate air pollution and mortality: acritical review of the evidence. Epidemiology7:420–428.

(54) Moolgavkar, S.H.; Luebeck, E.G.; Hall,T.A.; Anderson, E.L. (1995a) Air pollutionand daily mortality in Philadelphia.Epidemiology 6:476–484.

(55) Moolgavkar, S.H.; Luebeck, E.G.; Hall,T.A.; Anderson, E.L. (1995b) Particulate airpollution, sulfur dioxide, and daily mortality:a reanalysis of the Steubenville data.Inhalation Toxicology 7:35–44.

(56) Neas L.M.; Dockery D.W.; Koutrakis P.;Tollerud, D.J.; Speizer, F.E. (1995) Theassociation of ambient air pollution withtwice daily peak expiratory flow ratemeasurements in children. American Journalof Epidemiology 141:111–122.

(57) National Mining Association. (1997)Comments of the National MiningAssociation on the proposed revisions to thenational ambient air quality standards forparticulate matter. Docket No. A–95–54, IV–D–2158. March 12, 1997.

(58) Ozkaynak, H.; Spengler, J.D. (1996)The role of outdoor particulate matter inassessing total human exposure. In: Particlesin Our Air: Concentrations and HealthEffects. Harvard University Press. Wilson, R;Spengler, J.D.; eds.

(59) Peters, A.; Wichmann, H-E.; Tuch, T.;Heinrich, J.; Heyder, J. (1997) Respiratoryeffects are associated with the number ofultrafine particles. American Journal ofRespiratory Critical Care Medicine 155:(Inpress).

(60) Pacific Gas and Electric Company(1997) Comments on the proposed PM2.5 fineparticulate standard. Docket No. A–95–54,IV–D–2183. March 11, 1997.

(61) Pope, C.A., III, Kalkstein, J.S. (1996)Synoptic weather modeling and estimates ofthe exposure-response relationship betweendaily mortality and particulate air pollution.Environmental Health Perspectives 104:414–420.

(62) Pope, C.A., III; Thun, M.J.; Namboori,M.M.; Dockery, D.W.; Evans, J.S.; Speizer,F.E.; Heath, D.W., Jr. (1995) Particulate airpollution as a predictor of mortality in aprospective study of U.S. adults. AmericanJournal of Respiratory and Critical CareMedicine 151:669–674.

(63) Pope, C.A., III; Schwartz, J.; Ransom,M.R. (1992) Daily mortality and PM10

pollution in Utah Valley. ArchivesEnvironmental Health 47:211–217.

(64) Pope, C.A., III; Dockery, D.W. (1992)Acute health effects of PM10 pollution onsymptomatic and asymptomatic children.American Review of Respiratory Disease145:1123–1128.

(65) Raizenne, M.; Neas, L.M.; Damokosh,A.I.; Dockery, D.W.; Spengler, J.D.; Koutrakis,P.; Ware, J.H.; Speizer, F.E. (1996) Healtheffects of acid aerosols on North Americanchildren: pulmonary function.Environmental Health Perspectives 104:506–514.

(66) Roth, H.D.; Li, Y. (1997) Analysis ofthe association between air pollutants withmortality and hospital admissions inBirmingham, Alabama: 1986–1990.(Submitted for publication)

(67) Roth, H.D. et al. (1997) Analysis of theassociation between total particulate matterand daily mortality in the Czech Republic:1986–1994. Submitted as an attachment tocomments of R.E. Wyzga, Docket No. A–95–54, IV–D–2672. March 11, 1997.

(68) Sacks, J.; Karr, A.F.; Smith, R.L.; Davis,J.M. (1997) Comment on Scientific Input toDecision-Making on Airborne ParticulateStandards. Docket No. A–95–54, IV–D–14,298. March 12, 1997.

(69) Samet, J.M.; Zeger, S.L.; Kelsall, J.E.;Xu, J. (1996a) Air pollution and mortality inPhiladelphia 1973–1988: Report to the HealthEffects Institute on Phase I.B of the ParticleEpidemiology Evaluation Project. HealthEffects Institute, Cambridge, MA.

(70) Samet, J.M.; Zeger, S.L.; Kelsall, J.E.;Xu, J.; Kalkstein, L.S. (1996b) Weather, airpollution, and mortality in Philadelphia,1973–1980: Report to the Health EffectsInstitute on Phase I.B of the ParticleEpidemiology Evaluation Project. HealthEffects Institute, Cambridge, MA.

(71) Samet, J.M.; Zeger, S.L.; Berthane, K.(1995) The association of mortality andparticulate air pollution. In: Particulate AirPollution and Daily Mortality: Replicationand Valication of Selected Studies (ThePhase I.A Report of the Particle EpidemiologyEvaluation Project) pp. 3–104. Health EffectsInstitute, Cambridge, MA.

(72) Scarlett, J.F.; Abbott, K.J.; Peacock,J.L.; Strachan, D.P.; Anderson, H.R. (1996)Acute effects of summer air pollution onrespiratory function in primary schoolchildren in southern England. Thorax51:1109–1114.

(73) Schulze, W.D.; Brookshire, D.S.;Walther, E.G.; MacFarland, K. K.; Thayer,M.A.; Whitworth, R.L.; Ben–David, S.; Malm,W.; Molenar, Jr. (1983) The economicbenefits of preserving visibility in thenational parklands of the southwest. NaturalResources Journal 23:149–173.

(74) Schwartz, J.; Dockery, D.W.; Neas,L.M. (1996) Is daily mortality associatedspecifically with fine particles? Journal of Air& Waste Management Association 46:927–939.

(75) Schwartz, J.; Dockery, D.W.; Neas,L.M.; Wypij, D.; Ware, J.H.; Spengler, J.D.;Koutrakis, P.; Speizer, F.E.; Ferris, Jr., B.G.(1994) Acute effects of summer air pollutionon respiratory symptom reporting inchildren. American Journal of Respiratoryand Critical Care Medicine 150:1234–1242.

(76) Schwartz, J. (1994) Air pollution andhospital admissions for the elderly inBirmingham, Alabama. American Journal ofEpidemiology 139:589–598.

(77) Schwartz, J. (1993) Air pollution anddaily mortality in Birmingham, Alabama.American Journal of Epidemiology 137:1136–1147.


(78) Schwartz, J; Dockery, D.W. (1992a)Increased mortality in Philadelphiaassociated with daily air pollutionconcentrations. American Review ofRespiratory Disease 145:600–604.

(79) Schwartz, J.; Dockery, D.W. (1992b)Particulate air pollution and daily mortalityin Steubenville, Ohio. American Journal ofEpidemiology 135:12–19

(80) Sisler, J.; Malm, W.; Molenar, J.;Gebhardt, K. (1996) Spatial and SeasonalPatterns and Long Term Variability of theChemical Composition of the Haze in theU.S.: An Analysis of Data from the IMPROVENetwork. Fort Collins, CO: CooperativeInstitute for Research in the Atmosphere,Colorado State University. July 1996.

(81) Styer, P.; McMillan, N.; Gao, F.; Davis,J.; Sacks, J. (1995) Effect of outdoor airborneparticulate matter on daily death counts.Environmental Health Perspectives 103:490–497.

(82) Swiss EKL. (1996) Report No. 270.Swiss Federal Commission of Air Hygiene(EKL). Docket No. A–95–54, IV–I–59.

(83) Thurston, G.D. (1997) Letter toPresident William J. Clinton. Docket No. A–95–54, IV–F–97. January 10, 1997.

(84) Thurston, G.D.; Ito, K.; Hayes, C.G.;Bates, D.V.; Lippmann, M. (1994) Respiratoryhospital admissions and summertime hazeair pollution in Toronto, Ontario:consideration of the role of acid aerosols.Environmental Research 55:271–290.

(85) Utility Air Regulatory Group. (1997)Comments on the proposed air qualitystandards. Docket No. A–95–54, IV–D–2250.March 14, 1997

(86) United Kingdom Department ofEnvironment. (1997) The United KingdomNational Air Quality Strategy. UnitedKingdom of the Environment. Scottish Office.Docket No. A–95–54, IV–I–58. March 1997.

(87) U.S. Department of Health, Educationand Welfare (1964) Smoking and health:report of the Advisory Committee to theSurgeon General of the Public Health Service.Washington, DC: Public Health Service; p.60.

(88) U.S. Environmental Protection Agency(1996a) Air Quality Criteria for ParticulateMatter. Research Triangle Park, NC: NationalCenter for Environmental Assessment. Officeof Research and Development. April 12,1996.

(89) U.S. Environmental Protection Agency(1996b) Review of the National Ambient AirQuality Standards for Particulate Matter:Policy Assessment of Scientific andTechnical Information—Office of Air QualityPlanning and Standards Staff Paper. Office ofAir Quality Planning and Standards. Office ofAir and Radiation. July 1996.

(90) U.S. Environmental Protection Agency(1996c) Transcript of the Clean Air ScientificAdvisory Committee’s Review of theParticulate Matter Staff Paper Meetings heldon May 6–7, 1996 in Chapel Hill, NC.

(91) U.S. Environmental Protection Agency(1993) Office of Air Quality Planning andStandards Effects of the 1990 Clean Air ActAmendments on Visibility in Class I Areas:An EPA Report to Congress. ResearchTriangle Park, NC.

(92) Valdberg, P.A. (1997) Causality has notbeen demonstrated between outdoor levels of

particulate matter (PM) and daily mortalityand morbidity. In: Comments of the EngineManufacturers Association on the proposedrevisions to NAAQS for particulate matterand ozone. Docket No. A–95–54, IV–D–2328.March 11, 1997.

(93) Ware, J.H.; Ferris, B.G., Jr.; Dockery,D.W.; Spengler, J.D.; Stram, D.O.; Speizer,F.E. (1986) Effects of ambient sulfur oxidesand suspended particles on respiratoryhealth of children. American Review ofRespiratory Disease 133:834–842.

(94) World Health Organization. (1997)Update and Revision of the WHO Air QualityGuidelines for Europe. European Centre forEnvironment and Health. (In press). DocketNo. A–95–54, IV–I–60.

(95) Wolff, G.T. (1996a) Letter from GeorgeT. Wolff, Chair, Clean Air Scientific AdvisoryCommittee, to Administrator Carol M.Browner. Closure letter on draft Air QualityCriteria for Particulate Matter. March 15,1996.

(96) Wolff, G.T. (1996b) Letter from GeorgeT. Wolff, Chair, Clean Air Scientific AdvisoryCommittee, to Administrator Carol M.Browner. Clsoure letter on draft OAQPS StaffPaper (Review of the National Ambient AirQuality Standards for Particulate Matter:policy Assessment of Scientific andTechnical Information). June 13, 1996.

(97) Woodruff, T.J.; Grillo, J.; Schoendorf,K.C.; 1997. The relationship between selectedcauses of postneonatal infant mortality andparticulate air pollution in the United States.Environmental Health Perspectives 105:(Inpress)

(98) Wordley, J.; Walters, S.; Ayres, J.R.(1997) Short-term variations in hospitaladmissions and mortality and particulate airpollution. Occupational and EnvironmentalMedicine 54:108–116.

(99) Wyzga, R.E.; Lipfert, F.W. (1995)Temperature-pollution interactions withdaily mortality in Philadelphia. In:Particulate matter: health and regulatoryissues: proceedings of an internationalspecialty conference; April; Pittsburgh, PA.Air & Waste Management Association,Pittsburgh, PA (Air & Waste ManagementAssociation Publication VIP–49)

List of Subjects in 40 CFR Part 50Environmental protection, Air

pollution control, Carbon monoxide,Lead, Nitrogen dioxide, Ozone,Particulate matter, Sulfur oxides.

Dated: July 16, 1997.

Carol M. Browner,Administrator.

Therefore, 40 CFR Chapter I isamended as follows:

PART 50—NATIONAL PRIMARY ANDSECONDARY AMBIENT AIR QUALITYSTANDARDS

1. The authority citation for part 50continues to read as follows:

Authority: Secs. 109 and 301(a), Clean AirAct, as amended (42 U.S.C. 7409, 7601(a)).

2. Section 50.3 is revised to read asfollows:

§ 50.3 Reference conditions.All measurements of air quality that

are expressed as mass per unit volume(e.g., micrograms per cubic meter) otherthan for the particulate matter (PM10

and PM2.5) standards contained in § 50.7shall be corrected to a referencetemperature of 25 °C and a referencepressure of 760 millimeters of mercury(1,013.2 millibars). Measurements ofPM10 and PM2.5 for purposes ofcomparison to the standards containedin § 50.7 shall be reported based onactual ambient air volume measured atthe actual ambient temperature andpressure at the monitoring site duringthe measurement period.

3. Section 50.6 is amended by revisingthe section heading and addingparagraph (d) to read as follows:

§ 50.6 National primary and secondaryambient air quality standards for PM10.

* * * * *(d) The PM10 standards set forth in

this section will no longer apply to anarea not attaining these standards as ofSeptember 16, 1997, once EPA takesfinal action to promulgate a rulepursuant to section 172(e) of the CleanAir Act, as amended (42 U.S.C. 7472(e))applicable to the area. The PM10

standards set forth in this section willno longer apply to an area attainingthese standards as of September 16,1997, once EPA approves a StateImplementation Plan (SIP) applicable tothe area containing all PM10 controlmeasures adopted and implemented bythe state prior to September 16, 1997,and a section 110 SIP implementing thePM standards published on July 18,1997. SIP approvals are codified in 40CFR part 52.

4. Section 50.7 is added to read asfollows:

§ 50.7 National primary and secondaryambient air quality standards for particulatematter.

(a) The national primary andsecondary ambient air quality standardsfor particulate matter are:

(1) 15.0 micrograms per cubic meter(µg/m3) annual arithmetic meanconcentration, and 65 µg/m3 24-houraverage concentration measured in theambient air as PM2.5 (particles with anaerodynamic diameter less than or equalto a nominal 2.5 micrometers) by either:

(i) A reference method based onAppendix L of this part and designatedin accordance with part 53 of thischapter; or

(ii) An equivalent method designatedin accordance with part 53 of thischapter.

(2) 50 micrograms per cubic meter(µg/m3) annual arithmetic mean


concentration, and 150 µg/m3 24-houraverage concentration measured in theambient air as PM10 (particles with anaerodynamic diameter less than or equalto a nominal 10 micrometers) by either:

(i) A reference method based onAppendix M of this part and designatedin accordance with part 53 of thischapter; or

(ii) An equivalent method designatedin accordance with part 53 of thischapter.

(b) The annual primary and secondaryPM2.5 standards are met when theannual arithmetic mean concentration,as determined in accordance withAppendix N of this part, is less than orequal to 15.0 micrograms per cubicmeter.

(c) The 24-hour primary andsecondary PM2.5 standards are met whenthe 98th percentile 24-hourconcentration, as determined inaccordance with Appendix N of thispart, is less than or equal to 65micrograms per cubic meter.

(d) The annual primary and secondaryPM10 standards are met when theannual arithmetic mean concentration,as determined in accordance withAppendix N of this part, is less than orequal to 50 micrograms per cubic meter.

(e) The 24-hour primary andsecondary PM10 standards are met whenthe 99th percentile 24-hourconcentration, as determined inaccordance with Appendix N of thispart, is less than or equal to 150micrograms per cubic meter.

5. Appendix K is revised (forconformity with the format of the otherappendices in this part) to read asfollows:

Appendix K to Part 50—Interpretationof the National Ambient Air QualityStandards for Particulate Matter

1.0 General.(a) This appendix explains the

computations necessary for analyzingparticulate matter data to determineattainment of the 24-hour and annualstandards specified in 40 CFR 50.6. For theprimary and secondary standards, particulatematter is measured in the ambient air as PM10

(particles with an aerodynamic diameter lessthan or equal to a nominal 10 micrometers)by a reference method based on appendix Jof this part and designated in accordancewith part 53 of this chapter, or by anequivalent method designated in accordancewith part 53 of this chapter. The requiredfrequency of measurements is specified inpart 58 of this chapter.

(b) The terms used in this appendix aredefined as follows:

Average refers to an arithmetic mean. Allparticulate matter standards are expressed interms of expected annual values: Expectednumber of exceedances per year for the 24-hour standards and expected annualarithmetic mean for the annual standards.

Daily value for PM10 refers to the 24-houraverage concentration of PM10 calculated ormeasured from midnight to midnight (localtime).

Exceedance means a daily value that isabove the level of the 24-hour standard afterrounding to the nearest 10 µg/m3 (i.e., valuesending in 5 or greater are to be rounded up).

Expected annual value is the numberapproached when the annual values from anincreasing number of years are averaged, inthe absence of long-term trends in emissionsor meteorological conditions.

Year refers to a calendar year.(c) Although the discussion in this

appendix focuses on monitored data, thesame principles apply to modeling data,subject to EPA modeling guidelines.2.0 Attainment Determinations.

2.1 24-Hour Primary and SecondaryStandards.

(a) Under 40 CFR 50.6(a) the 24-hourprimary and secondary standards are attainedwhen the expected number of exceedancesper year at each monitoring site is less thanor equal to one. In the simplest case, thenumber of expected exceedances at a site isdetermined by recording the number ofexceedances in each calendar year and thenaveraging them over the past 3 calendaryears. Situations in which 3 years of data arenot available and possible adjustments forunusual events or trends are discussed insections 2.3 and 2.4 of this appendix.Further, when data for a year are incomplete,it is necessary to compute an estimatednumber of exceedances for that year byadjusting the observed number ofexceedances. This procedure, performed bycalendar quarter, is described in section 3.0of this appendix. The expected number ofexceedances is then estimated by averagingthe individual annual estimates for the past3 years.

(b) The comparison with the allowableexpected exceedance rate of one per year ismade in terms of a number rounded to thenearest tenth (fractional values equal to orgreater than 0.05 are to be rounded up; e.g.,an exceedance rate of 1.05 would be roundedto 1.1, which is the lowest rate fornonattainment).

2.2 Annual Primary and SecondaryStandards. Under 40 CFR 50.6(b), the annualprimary and secondary standards are attainedwhen the expected annual arithmetic meanPM10 concentration is less than or equal tothe level of the standard. In the simplest case,the expected annual arithmetic mean isdetermined by averaging the annualarithmetic mean PM10 concentrations for thepast 3 calendar years. Because of thepotential for incomplete data and thepossible seasonality in PM10 concentrations,the annual mean shall be calculated byaveraging the four quarterly means of PM10

concentrations within the calendar year. Theequations for calculating the annualarithmetic mean are given in section 4.0 ofthis appendix. Situations in which 3 years ofdata are not available and possibleadjustments for unusual events or trends arediscussed in sections 2.3 and 2.4 of thisappendix. The expected annual arithmeticmean is rounded to the nearest 1 µg/m3

before comparison with the annual standards

(fractional values equal to or greater than 0.5are to be rounded up).

2.3 Data Requirements.(a) 40 CFR 58.13 specifies the required

minimum frequency of sampling for PM10.For the purposes of making comparisonswith the particulate matter standards, all dataproduced by National Air MonitoringStations (NAMS), State and Local AirMonitoring Stations (SLAMS) and other sitessubmitted to EPA in accordance with the Part58 requirements must be used, and aminimum of 75 percent of the scheduledPM10 samples per quarter are required.

(b) To demonstrate attainment of either theannual or 24-hour standards at a monitoringsite, the monitor must provide sufficient datato perform the required calculations ofsections 3.0 and 4.0 of this appendix. Theamount of data required varies with thesampling frequency, data capture rate and thenumber of years of record. In all cases, 3years of representative monitoring data thatmeet the 75 percent criterion of the previousparagraph should be utilized, if available,and would suffice. More than 3 years may beconsidered, if all additional representativeyears of data meeting the 75 percent criterionare utilized. Data not meeting these criteriamay also suffice to show attainment;however, such exceptions will have to beapproved by the appropriate RegionalAdministrator in accordance with EPAguidance.

(c) There are less stringent datarequirements for showing that a monitor hasfailed an attainment test and thus hasrecorded a violation of the particulate matterstandards. Although it is generally necessaryto meet the minimum 75 percent data capturerequirement per quarter to use thecomputational equations described insections 3.0 and 4.0 of this appendix, thiscriterion does not apply when less data issufficient to unambiguously establishnonattainment. The following examplesillustrate how nonattainment can bedemonstrated when a site fails to meet thecompleteness criteria. Nonattainment of the24-hour primary standards can be establishedby the observed annual number ofexceedances (e.g., four observed exceedancesin a single year), or by the estimated numberof exceedances derived from the observednumber of exceedances and the requirednumber of scheduled samples (e.g., twoobserved exceedances with every other daysampling). Nonattainment of the annualstandards can be demonstrated on the basisof quarterly mean concentrations developedfrom observed data combined with one-halfthe minimum detectable concentrationsubstituted for missing values. In both cases,expected annual values must exceed thelevels allowed by the standards.

2.4 Adjustment for Exceptional Eventsand Trends.

(a) An exceptional event is anuncontrollable event caused by naturalsources of particulate matter or an event thatis not expected to recur at a given location.Inclusion of such a value in the computationof exceedances or averages could result ininappropriate estimates of their respectiveexpected annual values. To reduce the effectof unusual events, more than 3 years of


representative data may be used.Alternatively, other techniques, such as theuse of statistical models or the use ofhistorical data could be considered so thatthe event may be discounted or weightedaccording to the likelihood that it will recur.The use of such techniques is subject to theapproval of the appropriate RegionalAdministrator in accordance with EPAguidance.

(b) In cases where long–term trends inemissions and air quality are evident,mathematical techniques should be appliedto account for the trends to ensure that theexpected annual values are notinappropriately biased by unrepresentativedata. In the simplest case, if 3 years of dataare available under stable emissionconditions, this data should be used. In theevent of a trend or shift in emission patterns,either the most recent representative year(s)could be used or statistical techniques ormodels could be used in conjunction withprevious years of data to adjust for trends.The use of less than 3 years of data, and anyadjustments are subject to the approval of theappropriate Regional Administrator inaccordance with EPA guidance.3.0 Computational Equations for the 24-hour Standards.

3.1 Estimating Exceedances for a Year.(a) If PM10 sampling is scheduled less

frequently than every day, or if somescheduled samples are missed, a PM10 valuewill not be available for each day of the year.To account for the possible effect ofincomplete data, an adjustment must bemade to the data collected at each monitoringlocation to estimate the number ofexceedances in a calendar year. In thisadjustment, the assumption is made that thefraction of missing values that would haveexceeded the standard level is identical tothe fraction of measured values above thislevel. This computation is to be made for allsites that are scheduled to monitorthroughout the entire year and meet theminimum data requirements of section 2.3 ofthis appendix. Because of possible seasonalimbalance, this adjustment shall be appliedon a quarterly basis. The estimate of theexpected number of exceedances for thequarter is equal to the observed number ofexceedances plus an increment associatedwith the missing data. The followingequation must be used for thesecomputations:

Equation 1

e v v n N n v N nq q q q q q q q q= + ( ) × −( )[ ] = ×

where:

eq=the estimated number of exceedances forcalendar quarter q;

vq=the observed number of exceedances forcalendar quarter q;

Nq=the number of days in calendar quarter q;

nq=the number of days in calendar quarter qwith PM10 data; and

q=the index for calendar quarter, q=1, 2, 3 or4.

(b) The estimated number of exceedancesfor a calendar quarter must be rounded to the

nearest hundredth (fractional values equal toor greater than 0.005 must be rounded up).

(c) The estimated number of exceedancesfor the year, e, is the sum of the estimates foreach calendar quarter.

Equation 2

e eqq

==

∑1

4

(d) The estimated number of exceedancesfor a single year must be rounded to onedecimal place (fractional values equal to orgreater than 0.05 are to be rounded up). Theexpected number of exceedances is thenestimated by averaging the individual annualestimates for the most recent 3 or morerepresentative years of data. The expectednumber of exceedances must be rounded toone decimal place (fractional values equal toor greater than 0.05 are to be rounded up).

(e) The adjustment for incomplete data willnot be necessary for monitoring or modelingdata which constitutes a complete record,i.e., 365 days per year.

(f) To reduce the potential foroverestimating the number of expectedexceedances, the correction for missing datawill not be required for a calendar quarter inwhich the first observed exceedance hasoccurred if:

(1) There was only one exceedance in thecalendar quarter;

(2) Everyday sampling is subsequentlyinitiated and maintained for 4 calendarquarters in accordance with 40 CFR 58.13;and

(3) Data capture of 75 percent is achievedduring the required period of everydaysampling. In addition, if the first exceedanceis observed in a calendar quarter in whichthe monitor is already sampling every day,no adjustment for missing data will be madeto the first exceedance if a 75 percent datacapture rate was achieved in the quarter inwhich it was observed.

Example 1

a. During a particular calendar quarter, 39out of a possible 92 samples were recorded,with one observed exceedance of the 24-hourstandard. Using Equation 1, the estimatednumber of exceedances for the quarter is:eq=1×92/39=2.359 or 2.36.

b. If the estimated exceedances for theother 3 calendar quarters in the year were2.30, 0.0 and 0.0, then, using Equation 2, theestimated number of exceedances for the yearis 2.36+2.30+0.0+0.0 which equals 4.66 or4.7. If no exceedances were observed for the2 previous years, then the expected numberof exceedances is estimated by: (1/3)×(4.7+0+0)=1.57 or 1.6. Since 1.6 exceedsthe allowable number of expectedexceedances, this monitoring site would failthe attainment test.

Example 2

In this example, everyday sampling wasinitiated following the first observedexceedance as required by 40 CFR 58.13.Accordingly, the first observed exceedancewould not be adjusted for incompletesampling. During the next three quarters, 1.2exceedances were estimated. In this case, the

estimated exceedances for the year would be1.0+1.2+0.0+0.0 which equals 2.2. If, asbefore, no exceedances were observed for thetwo previous years, then the estimatedexceedances for the 3–year period wouldthen be (1/3)×(2.2+0.0+0.0)=0.7, and themonitoring site would not fail the attainmenttest.

3.2 Adjustments for Non-ScheduledSampling Days.

(a) If a systematic sampling schedule isused and sampling is performed on days inaddition to the days specified by thesystematic sampling schedule, e.g., duringepisodes of high pollution, then anadjustment must be made in the eqution forthe estimation of exceedances. Such anadjustment is needed to eliminate the bias inthe estimate of the quarterly and annualnumber of exceedances that would occur ifthe chance of an exceedance is different forscheduled than for non-scheduled days, aswould be the case with episode sampling.

(b) The required adjustment treats thesystematic sampling schedule as a stratifiedsampling plan. If the period from onescheduled sample until the day preceding thenext scheduled sample is defined as asampling stratum, then there is one stratumfor each scheduled sampling day. An averagenumber of observed exceedances iscomputed for each of these sampling strata.With nonscheduled sampling days, theestimated number of exceedances is definedas:

Equation 3

e N m v kq q q j jj

mq

= ( ) × ( )=∑

1

where:eq=the estimated number of exceedances for

the quarter;

Nq=the number of days in the quarter;

mq=the number of strata with samples duringthe quarter;

vj=the number of observed exceedances instratum j; and

kj=the number of actual samples in stratumj.

(c) Note that if only one sample value isrecorded in each stratum, then Equation 3reduces to Equation 1.

Example 3

A monitoring site samples according to asystematic sampling schedule of one sampleevery 6 days, for a total of 15 scheduledsamples in a quarter out of a total of 92possible samples. During one 6-day period,potential episode levels of PM10 weresuspected, so 5 additional samples weretaken. One of the regular scheduled sampleswas missed, so a total of 19 samples in 14sampling strata were measured. The one 6-day sampling stratum with 6 samplesrecorded 2 exceedances. The remainder ofthe quarter with one sample per stratumrecorded zero exceedances. Using Equation 3,the estimated number of exceedances for thequarter is:eq=(92/14)×(2/6+0+. . .+0)=2.19.


4.0 Computational Equations for AnnualStandards.

4.1 Calculation of the Annual ArithmeticMean. (a) An annual arithmetic mean valuefor PM10 is determined by averaging thequarterly means for the 4 calendar quartersof the year. The following equation is to beused for calculation of the mean for acalendar quarter:

Equation 4

x n xq q ii

nq

= ( ) ×=∑1

1

where:x̄q= the quarterly mean concentration for

quarter q, q=1, 2, 3, or 4,

nq= the number of samples in the quarter,and

xi= the ith concentration value recorded inthe quarter.

(b) The quarterly mean, expressed in µg/m3, must be rounded to the nearest tenth(fractional values of 0.05 should be roundedup).

(c) The annual mean is calculated by usingthe following equation:

Equation 5

x xqq

= ( ) ×=

∑14

1

4

where:

x̄=the annual mean; and

x̄q=the mean for calendar quarter q.

(d) The average of quarterly means must berounded to the nearest tenth (fractionalvalues of 0.05 should be rounded up).

(e) The use of quarterly averages tocompute the annual average will not benecessary for monitoring or modeling datawhich results in a complete record, i.e., 365days per year.

(f) The expected annual mean is estimatedas the average of three or more annual means.This multi-year estimate, expressed in µg/m3,shall be rounded to the nearest integer forcomparison with the annual standard(fractional values of 0.5 should be roundedup).

Example 4

Using Equation 4, the quarterly means arecalculated for each calendar quarter. If thequarterly means are 52.4, 75.3, 82.1, and 63.2µg/m 3, then the annual mean is:x̄ = (1/4)×(52.4+75.3+82.1+63.2)= 68.25 or68.3.

4.2 Adjustments for Non-scheduledSampling Days. (a) An adjustment in thecalculation of the annual mean is needed ifsampling is performed on days in addition tothe days specified by the systematic samplingschedule. For the same reasons given in thediscussion of estimated exceedances, undersection 3.2 of this appendix, the quarterlyaverages would be calculated by using thefollowing equation:

Equation 6

x m x kqq j

m

i

k

ij j

q j

=

× ( )

= =1

1 1Σ Σ

where:

x̄q=the quarterly mean concentration forquarter q, q=1, 2, 3, or 4;

xij=the ith concentration value recorded instratum j;

kj=the number of actual samples in stratumj; and

mq=the number of strata with data in thequarter.

(b) If one sample value is recorded in eachstratum, Equation 6 reduces to a simplearithmetic average of the observed values asdescribed by Equation 4.

Example 5

a. During one calendar quarter, 9observations were recorded. These sampleswere distributed among 7 sampling strata,with 3 observations in one stratum. Theconcentrations of the 3 observations in thesingle stratum were 202, 242, and 180 µg/m3.The remaining 6 observed concentrationswere 55, 68, 73, 92, 120, and 155 µg/m3.Applying the weighting factors specified inEquation 6, the quarterly mean is:x̄q = (1/7) × [(1/3) × (202 + 242 + 180) + 155+ 68 + 73 + 92 + 120 + 155] = 110.1

b. Although 24–hour measurements arerounded to the nearest 10 µg/m3 fordeterminations of exceedances of the 24–hour standard, note that these values arerounded to the nearest 1 µg/m3 for thecalculation of means.

6. Appendix L is added to read asfollows:

Appendix L to Part 50—ReferenceMethod For the Determination of FineParticulate Matter as PM2.5 in theAtmosphere

1.0 Applicability.1.1 This method provides for the

measurement of the mass concentration offine particulate matter having anaerodynamic diameter less than or equal toa nominal 2.5 micrometers (PM2.5) in ambientair over a 24-hour period for purposes ofdetermining whether the primary andsecondary national ambient air qualitystandards for fine particulate matter specifiedin § 50.6 of this part are met. Themeasurement process is considered to benondestructive, and the PM2.5 sampleobtained can be subjected to subsequentphysical or chemical analyses. Qualityassessment procedures are provided in part58, Appendix A of this chapter, and qualityassurance guidance are provided inreferences 1, 2, and 3 in section 13.0 of thisappendix.

1.2 This method will be considered areference method for purposes of part 58 ofthis chapter only if:

(a) The associated sampler meets therequirements specified in this appendix andthe applicable requirements in part 53 of thischapter, and

(b) The method and associated samplerhave been designated as a reference methodin accordance with part 53 of this chapter.

1.3 PM2.5 samplers that meet nearly allspecifications set forth in this method buthave minor deviations and/or modificationsof the reference method sampler will bedesignated as ‘‘Class I’’ equivalent methodsfor PM2.5 in accordance with part 53 of thischapter.2.0 Principle.

2.1 An electrically powered air samplerdraws ambient air at a constant volumetricflow rate into a specially shaped inlet andthrough an inertial particle size separator(impactor) where the suspended particulatematter in the PM2.5 size range is separated forcollection on a polytetrafluoroethylene(PTFE) filter over the specified samplingperiod. The air sampler and other aspects ofthis reference method are specified eitherexplicitly in this appendix or generally withreference to other applicable regulations orquality assurance guidance.

2.2 Each filter is weighed (after moistureand temperature conditioning) before andafter sample collection to determine the netgain due to collected PM2.5. The total volumeof air sampled is determined by the samplerfrom the measured flow rate at actualambient temperature and pressure and thesampling time. The mass concentration ofPM2.5 in the ambient air is computed as thetotal mass of collected particles in the PM2.5

size range divided by the actual volume of airsampled, and is expressed in micrograms percubic meter of air (µg/m3).3.0 PM2.5 Measurement Range.

3.1 Lower concentration limit. The lowerdetection limit of the mass concentrationmeasurement range is estimated to beapproximately 2 µg/am3, based on notedmass changes in field blanks in conjunctionwith the 24 m3 nominal total air samplevolume specified for the 24-hour sample.

3.2 Upper concentration limit. The upperlimit of the mass concentration range isdetermined by the filter mass loading beyondwhich the sampler can no longer maintainthe operating flow rate within specifiedlimits due to increased pressure drop acrossthe loaded filter. This upper limit cannot bespecified precisely because it is a complexfunction of the ambient particle sizedistribution and type, humidity, theindividual filter used, the capacity of thesampler flow rate control system, andperhaps other factors. Nevertheless, allsamplers are estimated to be capable ofmeasuring 24-hour PM2.5 massconcentrations of at least 200 µg/m3 whilemaintaining the operating flow rate withinthe specified limits.

3.3 Sample period. The required sampleperiod for PM2.5 concentration measurementsby this method shall be 1,380 to 1500minutes (23 to 25 hours). However, when asample period is less than 1,380 minutes, themeasured concentration (as determined bythe collected PM2.5 mass divided by theactual sampled air volume), multiplied bythe actual number of minutes in the sampleperiod and divided by 1,440, may be used asif it were a valid concentration measurementfor the specific purpose of determining aviolation of the NAAQS. This value assumes


that the PM2.5 concentration is zero for theremaining portion of the sample period andtherefore represents the minimumconcentration that could have been measuredfor the full 24-hour sample period.Accordingly, if the value thus calculated ishigh enough to be an exceedance, such anexceedance would be a valid exceedance forthe sample period. When reported to AIRS,this data value should receive a special codeto identify it as not to be commingled withnormal concentration measurements or usedfor other purposes.4.0 Accuracy.

4.1 Because the size and volatility of theparticles making up ambient particulatematter vary over a wide range and the massconcentration of particles varies with particlesize, it is difficult to define the accuracy ofPM2.5 measurements in an absolute sense.The accuracy of PM2.5 measurements istherefore defined in a relative sense,referenced to measurements provided by thisreference method. Accordingly, accuracyshall be defined as the degree of agreementbetween a subject field PM2.5 sampler and acollocated PM2.5 reference method auditsampler operating simultaneously at themonitoring site location of the subjectsampler and includes both random(precision) and systematic (bias) errors. Therequirements for this field sampler auditprocedure are set forth in part 58, AppendixA of this chapter.

4.2 Measurement system bias. Results ofcollocated measurements where theduplicate sampler is a reference methodsampler are used to assess a portion of themeasurement system bias according to theschedule and procedure specified in part 58,Appendix A of this chapter.

4.3 Audits with reference method samplersto determine system accuracy and bias.According to the schedule and procedurespecified in part 58, Appendix A of thischapter, a reference method sampler isrequired to be located at each of selectedPM2.5 SLAMS sites as a duplicate sampler.The results from the primary sampler and theduplicate reference method sampler are usedto calculate accuracy of the primary sampleron a quarterly basis, bias of the primarysampler on an annual basis, and bias of asingle reporting organization on an annualbasis. Reference 2 in section 13.0 of thisappendix provides additional informationand guidance on these reference methodaudits.

4.4 Flow rate accuracy and bias. Part 58,Appendix A of this chapter requires that theflow rate accuracy and bias of individualPM2.5 samplers used in SLAMS monitoringnetworks be assessed periodically via auditsof each sampler’s operational flow rate. Inaddition, part 58, Appendix A of this chapterrequires that flow rate bias for each referenceand equivalent method operated by eachreporting organization be assessed quarterlyand annually. Reference 2 in section 13.0 ofthis appendix provides additionalinformation and guidance on flow rateaccuracy audits and calculations for accuracyand bias.5.0 Precision. A data quality objective of 10percent coefficient of variation or better hasbeen established for the operational precisionof PM2.5 monitoring data.

5.1 Tests to establish initial operationalprecision for each reference method samplerare specified as a part of the requirements fordesignation as a reference method under§ 53.58 of this chapter.

5.2 Measurement System Precision.Collocated sampler results, where theduplicate sampler is not a reference methodsampler but is a sampler of the samedesignated method as the primary sampler,are used to assess measurement systemprecision according to the schedule andprocedure specified in part 58, Appendix Aof this chapter. Part 58, Appendix A of thischapter requires that these collocatedsampler measurements be used to calculatequarterly and annual precision estimates foreach primary sampler and for eachdesignated method employed by eachreporting organization. Reference 2 in section13.0 of this appendix provides additionalinformation and guidance on thisrequirement.6.0 Filter for PM2.5 Sample Collection. Anyfilter manufacturer or vendor who sells oroffers to sell filters specifically identified foruse with this PM2.5 reference method shallcertify that the required number of filtersfrom each lot of filters offered for sale as suchhave been tested as specified in this section6.0 and meet all of the following design andperformance specifications.

6.1 Size. Circular, 46.2 mm diameter ±0.25mm.

6.2 Medium. Polytetrafluoroethylene (PTFETeflon), with integral support ring.

6.3 Support ring. Polymethylpentene(PMP) or equivalent inert material, 0.38 ±0.04mm thick, outer diameter 46.2 mm ±0.25mm, and width of 3.68 mm ( ±0.00, -0.51mm).

6.4 Pore size. 2 µm as measured by ASTMF 316–94.

6.5 Filter thickness. 30 to 50 µm.6.6 Maximum pressure drop (clean filter).

30 cm H2O column @ 16.67 L/min clean airflow.

6.7 Maximum moisture pickup. Not morethan 10 µg weight increase after 24-hourexposure to air of 40 percent relativehumidity, relative to weight after 24-hourexposure to air of 35 percent relativehumidity.

6.8 Collection efficiency. Greater than 99.7percent, as measured by the DOP test (ASTMD 2986–91) with 0.3 µm particles at thesampler’s operating face velocity.

6.9 Filter weight stability. Filter weight lossshall be less than 20 µg, as measured in eachof the following two tests specified insections 6.9.1 and 6.9.2 of this appendix. Thefollowing conditions apply to both of thesetests: Filter weight loss shall be the averagedifference between the initial and the finalfilter weights of a random sample of testfilters selected from each lot prior to sale.The number of filters tested shall be not lessthan 0.1 percent of the filters of eachmanufacturing lot, or 10 filters, whichever isgreater. The filters shall be weighed underlaboratory conditions and shall have had noair sample passed through them, i.e., filterblanks. Each test procedure must includeinitial conditioning and weighing, the test,and final conditioning and weighing.Conditioning and weighing shall be in

accordance with sections 8.0 through 8.2 ofthis appendix and general guidance providedin reference 2 of section 13.0 of thisappendix.

6.9.1 Test for loose, surface particlecontamination. After the initial weighing,install each test filter, in turn, in a filtercassette (Figures L–27, L–28, and L–29 of thisappendix) and drop the cassette from aheight of 25 cm to a flat hard surface, suchas a particle-free wood bench. Repeat twotimes, for a total of three drop tests for eachtest filter. Remove the test filter from thecassette and weigh the filter. The averagechange in weight must be less than 20 µg.

6.9.2 Test for temperature stability. Afterweighing each filter, place the test filters ina drying oven set at 40 °C ±2 °C for not lessthan 48 hours. Remove, condition, andreweigh each test filter. The average changein weight must be less than 20 µg.

6.10 Alkalinity. Less than 25microequivalents/gram of filter, as measuredby the guidance given in reference 2 insection 13.0 of this appendix.

6.11 Supplemental requirements. Althoughnot required for determination of PM2.5 massconcentration under this reference method,additional specifications for the filter must bedeveloped by users who intend to subjectPM2.5 filter samples to subsequent chemicalanalysis. These supplemental specificationsinclude background chemical contaminationof the filter and any other filter parametersthat may be required by the method ofchemical analysis. All such supplementalfilter specifications must be compatible withand secondary to the primary filterspecifications given in this section 6.0 of thisappendix.7.0 PM2.5 Sampler.

7.1 Configuration. The sampler shallconsist of a sample air inlet, downtube,particle size separator (impactor), filterholder assembly, air pump and flow ratecontrol system, flow rate measurementdevice, ambient and filter temperaturemonitoring system, barometric pressuremeasurement system, timer, outdoorenvironmental enclosure, and suitablemechanical, electrical, or electronic controlcapability to meet or exceed the design andfunctional performance as specified in thissection 7.0 of this appendix. Theperformance specifications require that thesampler:

(a) Provide automatic control of samplevolumetric flow rate and other operationalparameters.

(b) Monitor these operational parameters aswell as ambient temperature and pressure.

(c) Provide this information to the sampleroperator at the end of each sample period indigital form, as specified in Table L–1 ofsection 7.4.19 of this appendix.

7.2 Nature of specifications. The PM2.5

sampler is specified by a combination ofdesign and performance requirements. Thesample inlet, downtube, particle sizediscriminator, filter cassette, and the internalconfiguration of the filter holder assembly arespecified explicitly by design figures andassociated mechanical dimensions,tolerances, materials, surface finishes,assembly instructions, and other necessaryspecifications. All other aspects of the


sampler are specified by required operationalfunction and performance, and the design ofthese other aspects (including the design ofthe lower portion of the filter holderassembly) is optional, subject to acceptableoperational performance. Test procedures todemonstrate compliance with both the designand performance requirements are set forthin subpart E of part 53 of this chapter.

7.3 Design specifications. Except asindicated in this section 7.3 of this appendix,these components must be manufactured orreproduced exactly as specified, in an ISO9001-registered facility, with registrationinitially approved and subsequentlymaintained during the period ofmanufacture. See § 53.1(t) of this chapter forthe definition of an ISO-registered facility.Minor modifications or variances to one ormore components that clearly would notaffect the aerodynamic performance of theinlet, downtube, impactor, or filter cassettewill be considered for specific approval. Anysuch proposed modifications shall bedescribed and submitted to the EPA forspecific individual acceptability either aspart of a reference or equivalent methodapplication under part 53 of this chapter orin writing in advance of such an intendedapplication under part 53 of this chapter.

7.3.1 Sample inlet assembly. The sampleinlet assembly, consisting of the inlet,downtube, and impactor shall be configuredand assembled as indicated in Figure L–1 ofthis appendix and shall meet all associatedrequirements. A portion of this assemblyshall also be subject to the maximum overallsampler leak rate specification under section7.4.6 of this appendix.

7.3.2 Inlet. The sample inlet shall befabricated as indicated in Figures L–2through L–18 of this appendix and shall meetall associated requirements.

7.3.3 Downtube. The downtube shall befabricated as indicated in Figure L–19 of thisappendix and shall meet all associatedrequirements.

7.3.4 Impactor.7.3.4.1 The impactor (particle size

separator) shall be fabricated as indicated inFigures L–20 through L–24 of this appendixand shall meet all associated requirements.Following the manufacture and finishing ofeach upper impactor housing (Figure L–21 ofthis appendix), the dimension of theimpaction jet must be verified by themanufacturer using Class ZZ go/no-go pluggauges that are traceable to NIST.

7.3.4.2 Impactor filter specifications:(a) Size. Circular, 35 to 37 mm diameter.(b) Medium. Borosilicate glass fiber,

without binder.(c) Pore size. 1 to 1.5 micrometer, as

measured by ASTM F 316–80.(d) Thickness. 300 to 500 micrometers.7.3.4.3 Impactor oil specifications:(a) Composition.

Tetramethyltetraphenyltrisiloxane, single-compound diffusion oil.

(b) Vapor pressure. Maximum 2 x 10-8 mmHg at 25 °C.

(c) Viscosity. 36 to 40 centistokes at 25 °C.(d) Density. 1.06 to 1.07 g/cm3 at 25 °C.(e) Quantity. 1 mL ±0.1 mL.7.3.5 Filter holder assembly. The sampler

shall have a sample filter holder assembly to

adapt and seal to the down tube and to holdand seal the specified filter, under section 6.0of this appendix, in the sample air stream ina horizontal position below the downtubesuch that the sample air passes downwardthrough the filter at a uniform face velocity.The upper portion of this assembly shall befabricated as indicated in Figures L–25 andL–26 of this appendix and shall accept andseal with the filter cassette, which shall befabricated as indicated in Figures L–27through L–29 of this appendix.

(a) The lower portion of the filter holderassembly shall be of a design andconstruction that:

(1) Mates with the upper portion of theassembly to complete the filter holderassembly,

(2) Completes both the external air seal andthe internal filter cassette seal such that allseals are reliable over repeated filterchangings, and

(3) Facilitates repeated changing of thefilter cassette by the sampler operator.

(b) Leak–test performance requirements forthe filter holder assembly are included insection 7.4.6 of this appendix.

(c) If additional or multiple filters arestored in the sampler as part of an automaticsequential sample capability, all such filters,unless they are currently and directlyinstalled in a sampling channel or samplingconfiguration (either active or inactive), shallbe covered or (preferably) sealed in such away as to:

(1) Preclude significant exposure of thefilter to possible contamination oraccumulation of dust, insects, or othermaterial that may be present in the ambientair, sampler, or sampler ventilation air duringstorage periods either before or aftersampling; and

(2) To minimize loss of volatile or semi-volatile PM sample components duringstorage of the filter following the sampleperiod.

7.3.6 Flow rate measurement adapter. Aflow rate measurement adapter as specifiedin Figure L–30 of this appendix shall befurnished with each sampler.

7.3.7 Surface finish. All internal surfacesexposed to sample air prior to the filter shallbe treated electrolytically in a sulfuric acidbath to produce a clear, uniform anodizedsurface finish of not less than 1000 mg/ft2

(1.08 mg/cm2) in accordance with militarystandard specification (mil. spec.) 8625F,Type II, Class 1 in reference 4 of section 13.0of this appendix. This anodic surface coatingshall not be dyed or pigmented. Followinganodization, the surfaces shall be sealed byimmersion in boiling deionized water for notless than 15 minutes. Section 53.51(d)(2) ofthis chapter should also be consulted.

7.3.8 Sampling height. The sampler shallbe equipped with legs, a stand, or othermeans to maintain the sampler in a stable,upright position and such that the center ofthe sample air entrance to the inlet, duringsample collection, is maintained in ahorizontal plane and is 2.0 ±0.2 meters abovethe floor or other horizontal supportingsurface. Suitable bolt holes, brackets, tie-downs, or other means should be provided tofacilitate mechanically securing the sampleto the supporting surface to prevent topplingof the sampler due to wind.

7.4 Performance specifications.7.4.1 Sample flow rate. Proper operation of

the impactor requires that specific airvelocities be maintained through the device.Therefore, the design sample air flow ratethrough the inlet shall be 16.67 L/min (1.000m3/hour) measured as actual volumetric flowrate at the temperature and pressure of thesample air entering the inlet.

7.4.2 Sample air flow rate control system.The sampler shall have a sample air flow ratecontrol system which shall be capable ofproviding a sample air volumetric flow ratewithin the specified range, under section7.4.1 of this appendix, for the specified filter,under section 6.0 of this appendix, at anyatmospheric conditions specified, undersection 7.4.7 of this appendix, at a filterpressure drop equal to that of a clean filterplus up to 75 cm water column (55 mm Hg),and over the specified range of supply linevoltage, under section 7.4.15.1 of thisappendix. This flow control system shallallow for operator adjustment of theoperational flow rate of the sampler over arange of at least ±15 percent of the flow ratespecified in section 7.4.1 of this appendix.

7.4.3 Sample flow rate regulation. Thesample flow rate shall be regulated such thatfor the specified filter, under section 6.0 ofthis appendix, at any atmospheric conditionsspecified, under section 7.4.7 of thisappendix, at a filter pressure drop equal tothat of a clean filter plus up to 75 cm watercolumn (55 mm Hg), and over the specifiedrange of supply line voltage, under section7.4.15.1 of this appendix, the flow rate isregulated as follows:

7.4.3.1 The volumetric flow rate, measuredor averaged over intervals of not more than5 minutes over a 24-hour period, shall notvary more than ±5 percent from the specified16.67 L/min flow rate over the entire sampleperiod.

7.4.3.2 The coefficient of variation (samplestandard deviation divided by the mean) ofthe flow rate, measured over a 24-hourperiod, shall not be greater than 2 percent.

7.4.3.3 The amplitude of short-term flowrate pulsations, such as may originate fromsome types of vacuum pumps, shall beattenuated such that they do not causesignificant flow measurement error or affectthe collection of particles on the particlecollection filter.

7.4.4 Flow rate cut off. The sampler’ssample air flow rate control system shallterminate sample collection and stop allsample flow for the remainder of the sampleperiod in the event that the sample flow ratedeviates by more than 10 percent from thesampler design flow rate specified in section7.4.1 of this appendix for more than 60seconds. However, this sampler cut-offprovision shall not apply during periodswhen the sampler is inoperative due to atemporary power interruption, and theelapsed time of the inoperative period shallnot be included in the total sample timemeasured and reported by the sampler, undersection 7.4.13 of this appendix.

7.4.5 Flow rate measurement.7.4.5.1 The sampler shall provide a means

to measure and indicate the instantaneoussample air flow rate, which shall bemeasured as volumetric flow rate at the


temperature and pressure of the sample airentering the inlet, with an accuracy of ±2percent. The measured flow rate shall beavailable for display to the sampler operatorat any time in either sampling or standbymodes, and the measurement shall beupdated at least every 30 seconds. Thesampler shall also provide a simple means bywhich the sampler operator can manuallystart the sample flow temporarily during non-sampling modes of operation, for the purposeof checking the sample flow rate or the flowrate measurement system.

7.4.5.2 During each sample period, thesampler’s flow rate measurement systemshall automatically monitor the samplevolumetric flow rate, obtaining flow ratemeasurements at intervals of not greater than30 seconds.

(a) Using these interval flow ratemeasurements, the sampler shall determineor calculate the following flow-relatedparameters, scaled in the specifiedengineering units:

(1) The instantaneous or interval-averageflow rate, in L/min.

(2) The value of the average sample flowrate for the sample period, in L/min.

(3) The value of the coefficient of variation(sample standard deviation divided by theaverage) of the sample flow rate for thesample period, in percent.

(4) The occurrence of any time intervalduring the sample period in which themeasured sample flow rate exceeds a rangeof ±5 percent of the average flow rate for thesample period for more than 5 minutes, inwhich case a warning flag indicator shall beset.

(5) The value of the integrated total samplevolume for the sample period, in m3.

(b) Determination or calculation of thesevalues shall properly exclude periods whenthe sampler is inoperative due to temporaryinterruption of electrical power, undersection 7.4.13 of this appendix, or flow ratecut off, under section 7.4.4 of this appendix.

(c) These parameters shall be accessible tothe sampler operator as specified in Table L–1 of section 7.4.19 of this appendix. Inaddition, it is strongly encouraged that theflow rate for each 5-minute interval duringthe sample period be available to the operatorfollowing the end of the sample period.

7.4.6 Leak test capability.7.4.6.1 External leakage. The sampler shall

include an external air leak-test capabilityconsisting of components, accessoryhardware, operator interface controls, awritten procedure in the associatedOperation/Instruction Manual, under section7.4.18 of this appendix, and all othernecessary functional capability to permit andfacilitate the sampler operator toconveniently carry out a leak test of thesampler at a field monitoring site withoutadditional equipment. The samplercomponents to be subjected to this leak testinclude all components and theirinterconnections in which external airleakage would or could cause an error in thesampler’s measurement of the total volume ofsample air that passes through the samplefilter.

(a) The suggested technique for theoperator to use for this leak test is as follows:

(1) Remove the sampler inlet and installsthe flow rate measurement adapter suppliedwith the sampler, under section 7.3.6 of thisappendix.

(2) Close the valve on the flow ratemeasurement adapter and use the sampler airpump to draw a partial vacuum in thesampler, including (at least) the impactor,filter holder assembly (filter in place), flowmeasurement device, and interconnectionsbetween these devices, of at least 55 mm Hg(75 cm water column), measured at a locationdownstream of the filter holder assembly.

(3) Plug the flow system downstream ofthese components to isolate the componentsunder vacuum from the pump, such as witha built-in valve.

(4) Stop the pump.(5) Measure the trapped vacuum in the

sampler with a built-in pressure measuringdevice.

(6) (i) Measure the vacuum in the samplerwith the built-in pressure measuring deviceagain at a later time at least 10 minutes afterthe first pressure measurement.

(ii) Caution: Following completion of thetest, the adaptor valve should be openedslowly to limit the flow rate of air into thesampler. Excessive air flow rate may blow oilout of the impactor.

(7) Upon completion of the test, open theadaptor valve, remove the adaptor and plugs,and restore the sampler to the normaloperating configuration.

(b) The associated leak test procedure shallrequire that for successful passage of this test,the difference between the two pressuremeasurements shall not be greater than thenumber of mm of Hg specified for thesampler by the manufacturer, based on theactual internal volume of the sampler, thatindicates a leak of less than 80 mL/min.

(c) Variations of the suggested technique oran alternative external leak test techniquemay be required for samplers whose designor configuration would make the suggestedtechnique impossible or impractical. Thespecific proposed external leak testprocedure, or particularly an alternative leaktest technique, proposed for a particularcandidate sampler may be described andsubmitted to the EPA for specific individualacceptability either as part of a reference orequivalent method application under part 53of this chapter or in writing in advance ofsuch an intended application under part 53of this chapter.

7.4.6.2 Internal, filter bypass leakage. Thesampler shall include an internal, filterbypass leak-check capability consisting ofcomponents, accessory hardware, operatorinterface controls, a written procedure in theOperation/Instruction Manual, and all othernecessary functional capability to permit andfacilitate the sampler operator toconveniently carry out a test for internal filterbypass leakage in the sampler at a fieldmonitoring site without additionalequipment. The purpose of the test is todetermine that any portion of the sampleflow rate that leaks past the sample filterwithout passing through the filter isinsignificant relative to the design flow ratefor the sampler.

(a) The suggested technique for theoperator to use for this leak test is as follows:

(1) Carry out an external leak test asprovided under section 7.4.6.1 of thisappendix which indicates successful passageof the prescribed external leak test.

(2) Install a flow-impervious membranematerial in the filter cassette, either with orwithout a filter, as appropriate, whicheffectively prevents air flow through thefilter.

(3) Use the sampler air pump to draw apartial vacuum in the sampler, downstreamof the filter holder assembly, of at least 55mm Hg (75 cm water column).

(4) Plug the flow system downstream of thefilter holder to isolate the components undervacuum from the pump, such as with a built-in valve.

(5) Stop the pump.(6) Measure the trapped vacuum in the

sampler with a built-in pressure measuringdevice.

(7) Measure the vacuum in the samplerwith the built-in pressure measuring deviceagain at a later time at least 10 minutes afterthe first pressure measurement.

(8) Remove the flow plug and membraneand restore the sampler to the normaloperating configuration.

(b) The associated leak test procedure shallrequire that for successful passage of this test,the difference between the two pressuremeasurements shall not be greater than thenumber of mm of Hg specified for thesampler by the manufacturer, based on theactual internal volume of the portion of thesampler under vacuum, that indicates a leakof less than 80 mL/min.

(c) Variations of the suggested technique oran alternative internal, filter bypass leak testtechnique may be required for samplerswhose design or configuration would makethe suggested technique impossible orimpractical. The specific proposed internalleak test procedure, or particularly analternative internal leak test techniqueproposed for a particular candidate samplermay be described and submitted to the EPAfor specific individual acceptability either aspart of a reference or equivalent methodapplication under part 53 of this chapter orin writing in advance of such intendedapplication under part 53 of this chapter.

7.4.7 Range of operational conditions. Thesampler is required to operate properly andmeet all requirements specified in thisappendix over the following operationalranges.

7.4.7.1 Ambient temperature. -30 to +45 °C(Note: Although for practical reasons, thetemperature range over which samplers arerequired to be tested under part 53 of thischapter is -20 to +40 °C, the sampler shall bedesigned to operate properly over this widertemperature range.).

7.4.7.2 Ambient relative humidity. 0 to 100percent.

7.4.7.3 Barometric pressure range. 600 to800 mm Hg.

7.4.8 Ambient temperature sensor. Thesampler shall have capability to measure thetemperature of the ambient air surroundingthe sampler over the range of -30 to +45 °C,with a resolution of 0.1 °C and accuracy of±2.0 °C, referenced as described in reference3 in section 13.0 of this appendix, with andwithout maximum solar insolation.


7.4.8.1 The ambient temperature sensorshall be mounted external to the samplerenclosure and shall have a passive, naturallyventilated sun shield. The sensor shall belocated such that the entire sun shield is atleast 5 cm above the horizontal plane of thesampler case or enclosure (disregarding theinlet and downtube) and external to thevertical plane of the nearest side orprotuberance of the sampler case orenclosure. The maximum temperaturemeasurement error of the ambienttemperature measurement system shall beless than 1.6 °C at 1 m/s wind speed and1000 W/m2 solar radiation intensity.

7.4.8.2 The ambient temperature sensorshall be of such a design and mounted insuch a way as to facilitate its convenientdismounting and immersion in a liquid forcalibration and comparison to the filtertemperature sensor, under section 7.4.11 ofthis appendix.

7.4.8.3 This ambient temperaturemeasurement shall be updated at least every30 seconds during both sampling andstandby (non-sampling) modes of operation.A visual indication of the current (mostrecent) value of the ambient temperaturemeasurement, updated at least every 30seconds, shall be available to the sampleroperator during both sampling and standby(non-sampling) modes of operation, asspecified in Table L–1 of section 7.4.19 ofthis appendix.

7.4.8.4 This ambient temperaturemeasurement shall be used for the purposeof monitoring filter temperature deviationfrom ambient temperature, as required bysection 7.4.11 of this appendix, and may beused for purposes of effecting filtertemperature control, under section 7.4.10 ofthis appendix, or computation of volumetricflow rate, under sections 7.4.1 to 7.4.5 of thisappendix, if appropriate.

7.4.8.5 Following the end of each sampleperiod, the sampler shall report themaximum, minimum, and averagetemperature for the sample period, asspecified in Table L–1 of section 7.4.19 ofthis appendix.

7.4.9 Ambient barometric sensor. Thesampler shall have capability to measure thebarometric pressure of the air surroundingthe sampler over a range of 600 to 800 mmHg referenced as described in reference 3 insection 13.0 of this appendix; also see part53, subpart E of this chapter. This barometricpressure measurement shall have a resolutionof 5 mm Hg and an accuracy of ±10 mm Hgand shall be updated at least every 30seconds. A visual indication of the value ofthe current (most recent) barometric pressuremeasurement, updated at least every 30seconds, shall be available to the sampleroperator during both sampling and standby(non-sampling) modes of operation, asspecified in Table L–1 of section 7.4.19 ofthis appendix. This barometric pressuremeasurement may be used for purposes ofcomputation of volumetric flow rate, undersections 7.4.1 to 7.4.5 of this appendix, ifappropriate. Following the end of a sampleperiod, the sampler shall report themaximum, minimum, and mean barometricpressures for the sample period, as specifiedin Table L–1 of section 7.4.19 of thisappendix.

7.4.10 Filter temperature control (samplingand post-sampling). The sampler shallprovide a means to limit the temperature riseof the sample filter (all sample filters forsequential samplers), from insolation andother sources, to no more 5 °C above thetemperature of the ambient air surroundingthe sampler, during both sampling and post-sampling periods of operation. The post-sampling period is the non-sampling periodbetween the end of the active samplingperiod and the time of retrieval of the samplefilter by the sampler operator.

7.4.11 Filter temperature sensor(s).7.4.11.1 The sampler shall have the

capability to monitor the temperature of thesample filter (all sample filters for sequentialsamplers) over the range of -30 to +45 °Cduring both sampling and non-samplingperiods. While the exact location of thistemperature sensor is not explicitly specified,the filter temperature measurement systemmust demonstrate agreement, within 1 °C,with a test temperature sensor located within1 cm of the center of the filter downstreamof the filter during both sampling and non-sampling modes, as specified in the filtertemperature measurement test described inpart 53, subpart E of this chapter. This filtertemperature measurement shall have aresolution of 0.1 °C and accuracy of ±1.0 °C,referenced as described in reference 3 insection 13.0 of this appendix. Thistemperature sensor shall be of such a designand mounted in such a way as to facilitateits reasonably convenient dismounting andimmersion in a liquid for calibration andcomparison to the ambient temperaturesensor under section 7.4.8 of this appendix.

7.4.11.2 The filter temperaturemeasurement shall be updated at least every30 seconds during both sampling andstandby (non-sampling) modes of operation.A visual indication of the current (mostrecent) value of the filter temperaturemeasurement, updated at least every 30seconds, shall be available to the sampleroperator during both sampling and standby(non-sampling) modes of operation, asspecified in Table L–1 of section 7.4.19 ofthis appendix.

7.4.11.3 For sequential samplers, thetemperature of each filter shall be measuredindividually unless it can be shown, asspecified in the filter temperaturemeasurement test described in § 53.57 of thischapter, that the temperature of each filtercan be represented by fewer temperaturesensors.

7.4.11.4 The sampler shall also provide awarning flag indicator following anyoccurrence in which the filter temperature(any filter temperature for sequentialsamplers) exceeds the ambient temperatureby more than 5 °C for more than 30consecutive minutes during either thesampling or post-sampling periods ofoperation, as specified in Table L–1 ofsection 7.4.19 of this appendix, under section10.12 of this appendix, regarding samplevalidity when a warning flag occurs. It isfurther recommended (not required) that thesampler be capable of recording themaximum differential between the measuredfilter temperature and the ambienttemperature and its time and date of

occurrence during both sampling and post-sampling (non-sampling) modes of operationand providing for those data to be accessibleto the sampler operator following the end ofthe sample period, as suggested in Table L–1 of section 7.4.19 of this appendix.

7.4.12 Clock/timer system.(a) The sampler shall have a programmable

real-time clock timing/control system that:(1) Is capable of maintaining local time and

date, including year, month, day-of-month,hour, minute, and second to an accuracy of±1.0 minute per month.

(2) Provides a visual indication of thecurrent system time, including year, month,day-of-month, hour, and minute, updated atleast each minute, for operator verification.

(3) Provides appropriate operator controlsfor setting the correct local time and date.

(4) Is capable of starting the samplecollection period and sample air flow at aspecific, operator-settable time and date, andstopping the sample air flow and terminatingthe sampler collection period 24 hours (1440minutes) later, or at a specific, operator-settable time and date.

(b) These start and stop times shall bereadily settable by the sampler operator towithin ±1.0 minute. The system shall providea visual indication of the current start andstop time settings, readable to ±1.0 minute,for verification by the operator, and the startand stop times shall also be available via thedata output port, as specified in Table L–1 ofsection 7.4.19 of this appendix. Uponexecution of a programmed sample periodstart, the sampler shall automatically reset allsample period information and warning flagindications pertaining to a previous sampleperiod. Refer also to section 7.4.15.4 of thisappendix regarding retention of current dateand time and programmed start and stoptimes during a temporary electrical powerinterruption.

7.4.13 Sample time determination. Thesampler shall be capable of determining theelapsed sample collection time for each PM2.5

sample, accurate to within ±1.0 minute,measured as the time between the start of thesampling period, under section 7.4.12 of thisappendix and the termination of the sampleperiod, under section 7.4.12 of this appendixor section 7.4.4 of this appendix. Thiselapsed sample time shall not includeperiods when the sampler is inoperative dueto a temporary interruption of electricalpower, under section 7.4.15.4 of thisappendix. In the event that the elapsedsample time determined for the sampleperiod is not within the range specified forthe required sample period in section 3.3 ofthis appendix, the sampler shall set awarning flag indicator. The date and time ofthe start of the sample period, the value ofthe elapsed sample time for the sampleperiod, and the flag indicator status shall beavailable to the sampler operator followingthe end of the sample period, as specified inTable L–1 of section 7.4.19 of this appendix.

7.4.14 Outdoor environmental enclosure.The sampler shall have an outdoor enclosure(or enclosures) suitable to protect the filterand other non-weatherproof components ofthe sampler from precipitation, wind, dust,extremes of temperature and humidity; tohelp maintain temperature control of the


filter (or filters, for sequential samplers); andto provide reasonable security for samplercomponents and settings.

7.4.15 Electrical power supply.7.4.15.1 The sampler shall be operable and

function as specified herein when operatedon an electrical power supply voltage of 105to 125 volts AC (RMS) at a frequency of 59to 61 Hz. Optional operation as specified atadditional power supply voltages and/orfrequencies shall not be precluded by thisrequirement.

7.4.15.2 The design and construction of thesampler shall comply with all applicableNational Electrical Code and UnderwritersLaboratories electrical safety requirements.

7.4.15.3 The design of all electrical andelectronic controls shall be such as toprovide reasonable resistance to interferenceor malfunction from ordinary or typicallevels of stray electromagnetic fields (EMF)as may be found at various monitoring sitesand from typical levels of electrical transientsor electronic noise as may often oroccasionally be present on various electricalpower lines.

7.4.15.4 In the event of temporary loss ofelectrical supply power to the sampler, thesampler shall not be required to sample orprovide other specified functions duringsuch loss of power, except that the internalclock/timer system shall maintain its localtime and date setting within ±1 minute perweek, and the sampler shall retain all othertime and programmable settings and all datarequired to be available to the sampleroperator following each sample period for atleast 7 days without electrical supply power.When electrical power is absent at theoperator-set time for starting a sample periodor is interrupted during a sample period, thesampler shall automatically start or resumesampling when electrical power is restored,if such restoration of power occurs before theoperator-set stop time for the sample period.

7.4.15.5 The sampler shall have thecapability to record and retain a record of the

year, month, day-of-month, hour, and minuteof the start of each power interruption ofmore than 1 minute duration, up to 10 suchpower interruptions per sample period.(More than 10 such power interruptions shallinvalidate the sample, except where anexceedance is measured, under section 3.3 ofthis appendix.) The sampler shall provide forthese power interruption data to be availableto the sampler operator following the end ofthe sample period, as specified in Table L–1 of section 7.4.19 of this appendix.

7.4.16 Control devices and operatorinterface. The sampler shall havemechanical, electrical, or electronic controls,control devices, electrical or electroniccircuits as necessary to provide the timing,flow rate measurement and control,temperature control, data storage andcomputation, operator interface, and otherfunctions specified. Operator-accessiblecontrols, data displays, and interface devicesshall be designed to be simple,straightforward, reliable, and easy to learn,read, and operate under field conditions. Thesampler shall have provision for operatorinput and storage of up to 64 characters ofnumeric (or alphanumeric) data for purposesof site, sampler, and sample identification.This information shall be available to thesampler operator for verification and changeand for output via the data output port alongwith other data following the end of a sampleperiod, as specified in Table L–1 of section7.4.19 of this appendix. All data required tobe available to the operator following asample collection period or obtained duringstandby mode in a post-sampling period shallbe retained by the sampler until reset, eithermanually by the operator or automatically bythe sampler upon initiation of a new samplecollection period.

7.4.17 Data output port requirement. Thesampler shall have a standard RS–232C dataoutput connection through which digital datamay be exported to an external data storageor transmission device. All information

which is required to be available at the endof each sample period shall be accessiblethrough this data output connection. Theinformation that shall be accessible thoughthis output port is summarized in Table L–1 of section 7.4.19 of this appendix. Since nospecific format for the output data isprovided, the sampler manufacturer orvendor shall make available to samplerpurchasers appropriate computer softwarecapable of receiving exported sampler dataand correctly translating the data into astandard spreadsheet format and optionallyany other formats as may be useful tosampler users. This requirement shall notpreclude the sampler from offering othertypes of output connections in addition tothe required RS–232C port.

7.4.18 Operation/instruction manual. Thesampler shall include an associatedcomprehensive operation or instructionmanual, as required by part 53 of thischapter, which includes detailed operatinginstructions on the setup, operation,calibration, and maintenance of the sampler.This manual shall provide complete anddetailed descriptions of the operational andcalibration procedures prescribed for fielduse of the sampler and all instrumentsutilized as part of this reference method. Themanual shall include adequate warning ofpotential safety hazards that may result fromnormal use or malfunction of the method anda description of necessary safety precautions.The manual shall also include a cleardescription of all procedures pertaining toinstallation, operation, periodic andcorrective maintenance, and troubleshooting,and shall include parts identificationdiagrams.

7.4.19 Data reporting requirements. Thevarious information that the sampler isrequired to provide and how it is to beprovided is summarized in the followingTable L–1.

TABLE L–1.—SUMMARY OF INFORMATION TO BE PROVIDED BY THE SAMPLER

Information to be pro-vided

AppendixL sectionreference

Availability Format

Anytime1 End of pe-riod2

Visual dis-play3

Data out-put4 Digital reading5 Units

Flow rate, 30-secondmaximum interval.

7.4.5.1 .... ✔ .................... ✔ * XX.X ........................... L/min

Flow rate, average forthe sample period.

7.4.5.2 .... * ✔ * ✔ XX.X ........................... L/min

Flow rate, CV, forsample period.

7.4.5.2 .... * ✔ * ✔0 XX.X ........................... %

Flow rate, 5-min. aver-age out of spec.(FLAG6).

7.4.5.2 .... ✔ ✔ ✔ ✔0 On/Off .........................

Sample volume, total .. 7.4.5.2 .... * ✔ ✔ ✔0 XX.X ........................... m3

Temperature, ambient,30-second interval.

7.4.8 ....... ✔ .................... ✔ .................... XX.X ........................... °C

Temperature, ambient,min., max., averagefor the sample pe-riod.

7.4.8 ....... * ✔ ✔ ✔0 XX.X ........................... °C

Baro pressure, ambi-ent, 30-second inter-val.

7.4.9 ....... ✔ .................... ✔ .................... XXX ............................ mm Hg


TABLE L–1.—SUMMARY OF INFORMATION TO BE PROVIDED BY THE SAMPLER—Continued

Information to be pro-vided

AppendixL sectionreference

Availability Format

Anytime1 End of pe-riod2

Visual dis-play3

Data out-put4 Digital reading5 Units

Baro pressure, ambi-ent, min., max., av-erage for the sampleperiod.

7.4.9 ....... * ✔ ✔ ✔0 XXX ............................ mm Hg

Filter temperature, 30-second interval.

7.4.11 ..... ✔ .................... ✔ .................... XX.X ........................... °C

Filter temperature dif-ferential, 30-secondinterval, out of spec.(FLAG6).

7.4.11 ..... * ✔ ✔ ✔0 On/Off .........................

Filter temperature,maximum differentialfrom ambient, date,time of occurrence.

7.4.11 ..... * * * * X.X, YY/MM/DDHH:mm.

°C, Yr./Mon./Day Hrs.min

Date and time ............. 7.4.12 ..... ✔ .................... ✔ .................... YY/MM/DD HH:mm .... Yr./Mon./Day Hrs. minSample start and stop

time settings.7.4.12 ..... ✔ ✔ ✔ ✔ YY/MM/DD HH:mm .... Yr./Mon./Day Hrs. min

Sample period starttime.

7.4.12 ..... .................... ✔ ✔ ✔0 YYYY/MM/DD HH:mm Yr./Mon./Day Hrs. min

Elapsed sample time .. 7.4.13 ..... * ✔ ✔ ✔0 HH:mm ....................... Hrs. minElapsed sample time,

out of spec. (FLAG6).7.4.13 ..... .................... ✔ ✔ ✔0 On/Off .........................

Power interruptions >1min., start time offirst 10.

7.4.15.5 .. * ✔ * ✔ 1HH:mm, 2HH:mm,etc ....

Hrs. min

User-entered informa-tion, such as sam-pler and site identi-fication.

7.4.16 ..... ✔ ✔ ✔ ✔0 As entered ..................

✔ Provision of this information is required.Provision of this information is optional. If information related to the entire sample period is optionally provided prior to the end of the sample

period, the value provided should be the value calculated for the portion of the sampler period completed up to the time the information is pro-vided.

0 Indicates that this information is also required to be provided to the AIRS data bank; see § § 58.26 and 58.35 of this chapter.

1 Information is required to be available to the operator at any time the sampler is operating, whether sampling or not.2 Information relates to the entire sampler period and must be provided following the end of the sample period until reset manually by the oper-

ator or automatically by the sampler upon the start of a new sample period.3 Information shall be available to the operator visually.4 Information is to be available as digital data at the sampler’s data output port specified in section 7.4.16 of this appendix following the end of

the sample period until reset manually by the operator or automatically by the sampler upon the start of a new sample period.5 Digital readings, both visual and data output, shall have not less than the number of significant digits and resolution specified.6 Flag warnings may be displayed to the operator by a single-flag indicator or each flag may be displayed individually. Only a set (on) flag

warning must be indicated; an off (unset) flag may be indicated by the absence of a flag warning. Sampler users should refer to section 10.12 ofthis appendix regarding the validity of samples for which the sampler provided an associated flag warning.

8.0 Filter Weighing. See reference 2 in section13.0 of this appendix, for additional, moredetailed guidance.

8.1 Analytical balance. The analyticalbalance used to weigh filters must be suitablefor weighing the type and size of filtersspecified, under section 6.0 of this appendix,and have a readability of ±1 µg. The balanceshall be calibrated as specified by themanufacturer at installation and recalibratedimmediately prior to each weighing session.See reference 2 in section 13.0 of thisappendix for additional guidance.

8.2 Filter conditioning. All sample filtersused shall be conditioned immediately beforeboth the pre- and post-sampling weighings asspecified below. See reference 2 in section13.0 of this appendix for additional guidance.

8.2.1 Mean temperature. 20 - 23 °C.8.2.2 Temperature control. ±2 °C over 24

hours.8.2.3 Mean humidity. Generally, 30–40

percent relative humidity; however, where it

can be shown that the mean ambient relativehumidity during sampling is less than 30percent, conditioning is permissible at amean relative humidity within ±5 relativehumidity percent of the mean ambientrelative humidity during sampling, but notless than 20 percent.

8.2.4 Humidity control. ±5 relativehumidity percent over 24 hours.

8.2.5 Conditioning time. Not less than 24hours.

8.3 Weighing procedure.8.3.1 New filters should be placed in the

conditioning environment immediately uponarrival and stored there until the pre-sampling weighing. See reference 2 in section13.0 of this appendix for additional guidance.

8.3.2 The analytical balance shall belocated in the same controlled environmentin which the filters are conditioned. Thefilters shall be weighed immediatelyfollowing the conditioning period without

intermediate or transient exposure to otherconditions or environments.

8.3.3 Filters must be conditioned at thesame conditions (humidity within ±5 relativehumidity percent) before both the pre- andpost-sampling weighings.

8.3.4 Both the pre- and post-samplingweighings should be carried out on the sameanalytical balance, using an effectivetechnique to neutralize static charges on thefilter, under reference 2 in section 13.0 of thisappendix. If possible, both weighings shouldbe carried out by the same analyst.

8.3.5 The pre-sampling (tare) weighingshall be within 30 days of the samplingperiod.

8.3.6 The post-sampling conditioning andweighing shall be completed within 240hours (10 days) after the end of the sampleperiod, unless the filter sample is maintainedat 4 °C or less during the entire time betweenretrieval from the sampler and the start of theconditioning, in which case the period shall


not exceed 30 days. Reference 2 in section13.0 of this appendix has additional guidanceon transport of cooled filters.

8.3.7 Filter blanks.8.3.7.1 New field blank filters shall be

weighed along with the pre-sampling (tare)weighing of each lot of PM2.5 filters. Theseblank filters shall be transported to thesampling site, installed in the sampler,retrieved from the sampler without sampling,and reweighed as a quality control check.

8.3.7.2 New laboratory blank filters shall beweighed along with the pre-sampling (tare)weighing of each set of PM2.5 filters. Theselaboratory blank filters should remain in thelaboratory in protective containers during thefield sampling and should be reweighed as aquality control check.

8.3.8 Additional guidance for proper filterweighing and related quality assuranceactivities is provided in reference 2 in section13.0 of this appendix.9.0 Calibration. Reference 2 in section 13.0of this appendix contains additionalguidance.

9.1 General requirements.9.1.1 Multipoint calibration and single-

point verification of the sampler’s flow ratemeasurement device must be performedperiodically to establish and maintaintraceability of subsequent flow measurementsto a flow rate standard.

9.1.2 An authoritative flow rate standardshall be used for calibrating or verifying thesampler’s flow rate measurement device withan accuracy of ±2 percent. The flow ratestandard shall be a separate, stand-alonedevice designed to connect to the flow ratemeasurement adapter, Figure L–30 of thisappendix. This flow rate standard must haveits own certification and be traceable to aNational Institute of Standards andTechnology (NIST) primary standard forvolume or flow rate. If adjustments to thesampler’s flow rate measurement systemcalibration are to be made in conjunctionwith an audit of the sampler’s flowmeasurement system, such adjustments shallbe made following the audit. Reference 2 insection 13.0 of this appendix containsadditional guidance.

9.1.3 The sampler’s flow rate measurementdevice shall be re-calibrated afterelectromechanical maintenance or transportof the sampler.

9.2 Flow rate calibration/verificationprocedure.

9.2.1 PM2.5 samplers may employ varioustypes of flow control and flow measurementdevices. The specific procedure used forcalibration or verification of the flow ratemeasurement device will vary depending onthe type of flow rate controller and flow ratemeasurement employed. Calibration shall bein terms of actual ambient volumetric flowrates (Qa), measured at the sampler’s inletdowntube. The generic procedure given hereserves to illustrate the general steps involvedin the calibration of a PM2.5 sampler. Thesampler operation/instruction manualrequired under section 7.4.18 of thisappendix and the Quality AssuranceHandbook in reference 2 in section 13.0 ofthis appendix provide more specific anddetailed guidance for calibration.

9.2.2 The flow rate standard used for flowrate calibration shall have its own

certification and be traceable to a NISTprimary standard for volume or flow rate. Acalibration relationship for the flow ratestandard, e.g., an equation, curve, or familyof curves relating actual flow rate (Qa) to theflow rate indicator reading, shall beestablished that is accurate to within 2percent over the expected range of ambienttemperatures and pressures at which the flowrate standard may be used. The flow ratestandard must be re-calibrated or re-verifiedat least annually.

9.2.3 The sampler flow rate measurementdevice shall be calibrated or verified byremoving the sampler inlet and connectingthe flow rate standard to the sampler’sdowntube in accordance with the operation/instruction manual, such that the flow ratestandard accurately measures the sampler’sflow rate. The sampler operator shall firstcarry out a sampler leak check and confirmthat the sampler passes the leak test and thenverify that no leaks exist between the flowrate standard and the sampler.

9.2.4 The calibration relationship betweenthe flow rate (in actual L/min) indicated bythe flow rate standard and by the sampler’sflow rate measurement device shall beestablished or verified in accordance with thesampler operation/instruction manual.Temperature and pressure corrections to theflow rate indicated by the flow rate standardmay be required for certain types of flow ratestandards. Calibration of the sampler’s flowrate measurement device shall consist of atleast three separate flow rate measurements(multipoint calibration) evenly spaced withinthe range of -10 percent to +10 percent of thesampler’s operational flow rate, section 7.4.1of this appendix. Verification of thesampler’s flow rate shall consist of one flowrate measurement at the sampler’soperational flow rate. The sampler operation/instruction manual and reference 2 in section13.0 of this appendix provide additionalguidance.

9.2.5 If during a flow rate verification thereading of the sampler’s flow rate indicatoror measurement device differs by ±2 percentor more from the flow rate measured by theflow rate standard, a new multipointcalibration shall be performed and the flowrate verification must then be repeated.

9.2.6 Following the calibration orverification, the flow rate standard shall beremoved from the sampler and the samplerinlet shall be reinstalled. Then the sampler’snormal operating flow rate (in L/min) shallbe determined with a clean filter in place. Ifthe flow rate indicated by the sampler differsby ±2 percent or more from the requiredsampler flow rate, the sampler flow rate mustbe adjusted to the required flow rate, undersection 7.4.1 of this appendix.

9.3 Periodic calibration or verification ofthe calibration of the sampler’s ambienttemperature, filter temperature, andbarometric pressure measurement systems isalso required. Reference 3 of section 13.0 ofthis appendix contains additional guidance.

10.0 PM2.5 Measurement Procedure Thedetailed procedure for obtaining valid PM2.5

measurements with each specific samplerdesignated as part of a reference method forPM2.5 under part 53 of this chapter shall beprovided in the sampler-specific operation or

instruction manual required by section 7.4.18of this appendix. Supplemental guidance isprovided in section 2.12 of the QualityAssurance Handbook listed in reference 2 insection 13.0 of this appendix. The genericprocedure given here serves to illustrate thegeneral steps involved in the PM2.5 samplecollection and measurement, using a PM2.5

reference method sampler.10.1 The sampler shall be set up,

calibrated, and operated in accordance withthe specific, detailed guidance provided inthe specific sampler’s operation orinstruction manual and in accordance with aspecific quality assurance program developedand established by the user, based onapplicable supplementary guidance providedin reference 2 in section 13.0 of thisappendix.

10.2 Each new sample filter shall beinspected for correct type and size and forpinholes, particles, and other imperfections.Unacceptable filters should be discarded. Aunique identification number shall beassigned to each filter, and an informationrecord shall be established for each filter. Ifthe filter identification number is not orcannot be marked directly on the filter,alternative means, such as a number-identified storage container, must beestablished to maintain positive filteridentification.

10.3 Each filter shall be conditioned in theconditioning environment in accordancewith the requirements specified in section8.2 of this appendix.

10.4 Following conditioning, each filtershall be weighed in accordance with therequirements specified in section 8.0 of thisappendix and the presampling weightrecorded with the filter identificationnumber.

10.5 A numbered and preweighed filtershall be installed in the sampler followingthe instructions provided in the sampleroperation or instruction manual.

10.6 The sampler shall be checked andprepared for sample collection in accordancewith instructions provided in the sampleroperation or instruction manual and with thespecific quality assurance programestablished for the sampler by the user.

10.7 The sampler’s timer shall be set tostart the sample collection at the beginningof the desired sample period and stop thesample collection 24 hours later.

10.8 Information related to the samplecollection (site location or identificationnumber, sample date, filter identificationnumber, and sampler model and serialnumber) shall be recorded and, ifappropriate, entered into the sampler.

10.9 The sampler shall be allowed tocollect the PM2.5 sample during the set 24-hour time period.

10.10 Within 96 hours of the end of thesample collection period, the filter, whilestill contained in the filter cassette, shall becarefully removed from the sampler,following the procedure provided in thesampler operation or instruction manual andthe quality assurance program, and placed ina protective container. This protectivecontainer shall be made of metal and containno loose material that could be transferred tothe filter. The protective container shall hold


the filter cassette securely such that the covershall not come in contact with the filter’ssurfaces. Reference 2 in section 13.0 of thisappendix contains additional information.

10.11 The total sample volume in actual m3

for the sampling period and the elapsedsample time shall be obtained from thesampler and recorded in accordance with theinstructions provided in the sampleroperation or instruction manual. All samplerwarning flag indications and otherinformation required by the local qualityassurance program shall also be recorded.

10.12 All factors related to the validity orrepresentativeness of the sample, such assampler tampering or malfunctions, unusualmeteorological conditions, constructionactivity, fires or dust storms, etc. shall berecorded as required by the local qualityassurance program. The occurrence of a flagwarning during a sample period shall notnecessarily indicate an invalid sample butrather shall indicate the need for specificreview of the QC data by a quality assuranceofficer to determine sample validity.

10.13 After retrieval from the sampler, theexposed filter containing the PM2.5 sampleshould be transported to the filterconditioning environment as soon as possibleideally to arrive at the conditioningenvironment within 24 hours forconditioning and subsequent weighing.During the period between filter retrievalfrom the sampler and the start of theconditioning, the filter shall be maintained ascool as practical and continuously protectedfrom exposure to temperatures over 25 °C.See section 8.3.6 of this appendix regardingtime limits for completing the post-sampling

weighing. See reference 2 in section 13.0 ofthis appendix for additional guidance ontransporting filter samplers to theconditioning and weighing laboratory.

10.14. The exposed filter containing thePM2.5 sample shall be re-conditioned in theconditioning environment in accordancewith the requirements specified in section8.2 of this appendix.

10.15. The filter shall be reweighedimmediately after conditioning in accordancewith the requirements specified in section8.0 of this appendix, and the postsamplingweight shall be recorded with the filteridentification number.

10.16 The PM2.5 concentration shall becalculated as specified in section 12.0 of thisappendix.11.0 Sampler Maintenance

The sampler shall be maintained asdescribed by the sampler’s manufacturer inthe sampler-specific operation or instructionmanual required under section 7.4.18 of thisappendix and in accordance with the specificquality assurance program developed andestablished by the user based on applicablesupplementary guidance provided inreference 2 in section 13.0 of this appendix.12.0 Calculations

12.1 (a) The PM2.5 concentration iscalculated as:

PM2.5 = (Wf - Wi)/Va

where:PM2.5 = mass concentration of PM2.5, µg/

m3;Wf, Wi = final and initial weights,

respectively, of the filter used to collect thePM2.5 particle sample, µg;

Va = total air volume sampled in actualvolume units, as provided by the sampler,m3.

(b) Note: Total sample time must bebetween 1,380 and 1,500 minutes (23 and 25hrs) for a fully valid PM2.5 sample; however,see also section 3.3 of this appendix.13.0 References.

1. Quality Assurance Handbook for AirPollution Measurement Systems, Volume I,Principles. EPA/600/R–94/038a, April 1994.Available from CERI, ORD Publications, U.S.Environmental Protection Agency, 26 WestMartin Luther King Drive, Cincinnati, Ohio45268.

2. Copies of section 2.12 of the QualityAssurance Handbook for Air PollutionMeasurement Systems, Volume II, AmbientAir Specific Methods, EPA/600/R–94/038b,are available from Department E (MD-77B),U.S. EPA, Research Triangle Park, NC 27711.

3. Quality Assurance Handbook for AirPollution Measurement Systems, Volume IV:Meteorological Measurements, (RevisedEdition) EPA/600/R–94/038d, March, 1995.Available from CERI, ORD Publications, U.S.Environmental Protection Agency, 26 WestMartin Luther King Drive, Cincinnati, Ohio45268.

4. Military standard specification (mil.spec.) 8625F, Type II, Class 1 as listed inDepartment of Defense Index ofSpecifications and Standards (DODISS),available from DODSSP–Customer Service,Standardization Documents Order Desk, 700Robbins Avenue, Building 4D, Philadelphia,PA 1911–5094.14.0 Figures L–1 through L–30 to AppendixL.
































7. Appendix M is added to read asfollows:

Appendix M to Part 50—ReferenceMethod for the Determination ofParticulate Matter as PM10 in theAtmosphere

1.0 Applicability.1.1 This method provides for the

measurement of the mass concentration ofparticulate matter with an aerodynamicdiameter less than or equal to a nominal 10micrometers (PM1O) in ambient air over a 24-hour period for purposes of determiningattainment and maintenance of the primaryand secondary national ambient air qualitystandards for particulate matter specified in§ 50.6 of this chapter. The measurementprocess is nondestructive, and the PM10

sample can be subjected to subsequentphysical or chemical analyses. Qualityassurance procedures and guidance areprovided in part 58, Appendices A and B ofthis chapter and in references 1 and 2 ofsection 12.0 of this appendix.2.0 Principle.

2.1 An air sampler draws ambient air ata constant flow rate into a specially shapedinlet where the suspended particulate matteris inertially separated into one or more sizefractions within the PM10 size range. Eachsize fraction in the PM1O size range is thencollected on a separate filter over thespecified sampling period. The particle sizediscrimination characteristics (samplingeffectiveness and 50 percent cutpoint) of thesampler inlet are prescribed as performancespecifications in part 53 of this chapter.

2.2 Each filter is weighed (after moistureequilibration) before and after use todetermine the net weight (mass) gain due tocollected PM10. The total volume of airsampled, measured at the actual ambienttemperature and pressure, is determinedfrom the measured flow rate and thesampling time. The mass concentration ofPM10 in the ambient air is computed as thetotal mass of collected particles in the PM10

size range divided by the volume of airsampled, and is expressed in micrograms peractual cubic meter (µg/m3).

2.3 A method based on this principle willbe considered a reference method only if theassociated sampler meets the requirementsspecified in this appendix and therequirements in part 53 of this chapter, andthe method has been designated as areference method in accordance with part 53of this chapter.3.0 Range.

3.1 The lower limit of the massconcentration range is determined by therepeatability of filter tare weights, assumingthe nominal air sample volume for thesampler. For samplers having an automaticfilter-changing mechanism, there may be noupper limit. For samplers that do not have anautomatic filter-changing mechanism, theupper limit is determined by the filter massloading beyond which the sampler no longermaintains the operating flow rate withinspecified limits due to increased pressuredrop across the loaded filter. This upper limitcannot be specified precisely because it is acomplex function of the ambient particle size

distribution and type, humidity, filter type,and perhaps other factors. Nevertheless, allsamplers should be capable of measuring 24-hour PM10 mass concentrations of at least 300µg/m3 while maintaining the operating flowrate within the specified limits.4.0 Precision.

4.1 The precision of PM10 samplers mustbe 5 µg/m3 for PM10 concentrations below 80µg/m3 and 7 percent for PM10 concentrationsabove 80 µg/m3, as required by part 53 of thischapter, which prescribes a test procedurethat determines the variation in the PM10

concentration measurements of identicalsamplers under typical sampling conditions.Continual assessment of precision viacollocated samplers is required by part 58 ofthis chapter for PM10 samplers used incertain monitoring networks.5.0 Accuracy.

5.1 Because the size of the particlesmaking up ambient particulate matter variesover a wide range and the concentration ofparticles varies with particle size, it isdifficult to define the absolute accuracy ofPM10 samplers. Part 53 of this chapterprovides a specification for the samplingeffectiveness of PM10 samplers. Thisspecification requires that the expected massconcentration calculated for a candidatePM10 sampler, when sampling a specifiedparticle size distribution, be within ±10percent of that calculated for an idealsampler whose sampling effectiveness isexplicitly specified. Also, the particle size for50 percent sampling effectiveness is requiredto be 10±0.5 micrometers. Otherspecifications related to accuracy apply toflow measurement and calibration, filtermedia, analytical (weighing) procedures, andartifact. The flow rate accuracy of PM10

samplers used in certain monitoringnetworks is required by part 58 of thischapter to be assessed periodically via flowrate audits.6.0 Potential Sources of Error.

6.1 Volatile Particles. Volatile particlescollected on filters are often lost duringshipment and/or storage of the filters prior tothe post-sampling weighing 3. Althoughshipment or storage of loaded filters issometimes unavoidable, filters should bereweighed as soon as practical to minimizethese losses.

6.2 Artifacts. Positive errors in PM10

concentration measurements may result fromretention of gaseous species on filters 4, 5.Such errors include the retention of sulfurdioxide and nitric acid. Retention of sulfurdioxide on filters, followed by oxidation tosulfate, is referred to as artifact sulfateformation, a phenomenon which increaseswith increasing filter alkalinity 6. Little or noartifact sulfate formation should occur usingfilters that meet the alkalinity specification insection 7.2.4 of this appendix, Artifact nitrateformation, resulting primarily from retentionof nitric acid, occurs to varying degrees onmany filter types, including glass fiber,cellulose ester, and many quartz fiberfilters 5, 7, 8, 9, 10. Loss of true atmosphericparticulate nitrate during or followingsampling may also occur due to dissociationor chemical reaction. This phenomenon hasbeen observed on Teflon filters 8 andinferred for quartz fiber filters 11, 12. The

magnitude of nitrate artifact errors in PM10

mass concentration measurements will varywith location and ambient temperature;however, for most sampling locations, theseerrors are expected to be small.

6.3 Humidity. The effects of ambienthumidity on the sample are unavoidable. Thefilter equilibration procedure in section 9.0 ofthis appendix is designed to minimize theeffects of moisture on the filter medium.

6.4 Filter Handling. Careful handling offilters between presampling andpostsampling weighings is necessary to avoiderrors due to damaged filters or loss ofcollected particles from the filters. Use of afilter cartridge or cassette may reduce themagnitude of these errors. Filters must alsomeet the integrity specification in section7.2.3 of this appendix.

6.5 Flow Rate Variation. Variations in thesampler’s operating flow rate may alter theparticle size discrimination characteristics ofthe sampler inlet. The magnitude of this errorwill depend on the sensitivity of the inlet tovariations in flow rate and on the particledistribution in the atmosphere during thesampling period. The use of a flow controldevice, under section 7.1.3 of this appendix,is required to minimize this error.

6.6 Air Volume Determination. Errors inthe air volume determination may result fromerrors in the flow rate and/or sampling timemeasurements. The flow control deviceserves to minimize errors in the flow ratedetermination, and an elapsed time meter,under section 7.1.5 of this appendix, isrequired to minimize the error in thesampling time measurement.7.0 Apparatus.

7.1 PM10 Sampler.7.1.1 The sampler shall be designed to:(a) Draw the air sample into the sampler

inlet and through the particle collection filterat a uniform face velocity.

(b) Hold and seal the filter in a horizontalposition so that sample air is drawndownward through the filter.

(c) Allow the filter to be installed andremoved conveniently.

(d) Protect the filter and sampler fromprecipitation and prevent insects and otherdebris from being sampled.

(e) Minimize air leaks that would causeerror in the measurement of the air volumepassing through the filter.

(f) Discharge exhaust air at a sufficientdistance from the sampler inlet to minimizethe sampling of exhaust air.

(g) Minimize the collection of dust fromthe supporting surface.

7.1.2 The sampler shall have a sample airinlet system that, when operated within aspecified flow rate range, provides particlesize discrimination characteristics meetingall of the applicable performancespecifications prescribed in part 53 of thischapter. The sampler inlet shall show nosignificant wind direction dependence. Thelatter requirement can generally be satisfiedby an inlet shape that is circularlysymmetrical about a vertical axis.

7.1.3 The sampler shall have a flowcontrol device capable of maintaining thesampler’s operating flow rate within the flowrate limits specified for the sampler inlet overnormal variations in line voltage and filterpressure drop.


7.1.4 The sampler shall provide a meansto measure the total flow rate during thesampling period. A continuous flow recorderis recommended but not required. The flowmeasurement device shall be accurate to ±2percent.

7.1.5 A timing/control device capable ofstarting and stopping the sampler shall beused to obtain a sample collection period of24 ±1 hr (1,440 ±60 min). An elapsed timemeter, accurate to within ±15 minutes, shallbe used to measure sampling time. Thismeter is optional for samplers withcontinuous flow recorders if the samplingtime measurement obtained by means of therecorder meets the ±15 minute accuracyspecification.

7.1.6 The sampler shall have anassociated operation or instruction manual asrequired by part 53 of this chapter whichincludes detailed instructions on thecalibration, operation, and maintenance ofthe sampler.

7.2 Filters.7.2.1 Filter Medium. No commercially

available filter medium is ideal in all respectsfor all samplers. The user’s goals in samplingdetermine the relative importance of variousfilter characteristics, e.g., cost, ease ofhandling, physical and chemicalcharacteristics, etc., and, consequently,determine the choice among acceptablefilters. Furthermore, certain types of filtersmay not be suitable for use with somesamplers, particularly under heavy loadingconditions (high mass concentrations),because of high or rapid increase in the filterflow resistance that would exceed thecapability of the sampler’s flow controldevice. However, samplers equipped withautomatic filter-changing mechanisms mayallow use of these types of filters. Thespecifications given below are minimumrequirements to ensure acceptability of thefilter medium for measurement of PM10 massconcentrations. Other filter evaluationcriteria should be considered to meetindividual sampling and analysis objectives.

7.2.2 Collection Efficiency. ≥99 percent,as measured by the DOP test (ASTM–2986)with 0.3 µm particles at the sampler’soperating face velocity.

7.2.3 Integrity. ±5 µg/m3 (assumingsampler’s nominal 24-hour air samplevolume). Integrity is measured as the PM10

concentration equivalent corresponding tothe average difference between the initial andthe final weights of a random sample of testfilters that are weighed and handled underactual or simulated sampling conditions, buthave no air sample passed through them, i.e.,filter blanks. As a minimum, the testprocedure must include initial equilibrationand weighing, installation on an inoperativesampler, removal from the sampler, and finalequilibration and weighing.

7.2.4 Alkalinity. <25 microequivalents/gram of filter, as measured by the proceduregiven in reference 13 of section 12.0 of thisappendix following at least two monthsstorage in a clean environment (free fromcontamination by acidic gases) at roomtemperature and humidity.

7.3 Flow Rate Transfer Standard. Theflow rate transfer standard must be suitablefor the sampler’s operating flow rate and

must be calibrated against a primary flow orvolume standard that is traceable to theNational Institute of Standard andTechnology (NIST). The flow rate transferstandard must be capable of measuring thesampler’s operating flow rate with anaccuracy of ±2 percent.

7.4 Filter Conditioning Environment.7.4.1 Temperature range. 15 to 30 C.7.4.2 Temperature control. ±3 C.7.4.3 Humidity range. 20% to 45% RH.7.4.4 Humidity control. ±5% RH.7.5 Analytical Balance. The analytical

balance must be suitable for weighing thetype and size of filters required by thesampler. The range and sensitivity requiredwill depend on the filter tare weights andmass loadings. Typically, an analyticalbalance with a sensitivity of 0.1 mg isrequired for high volume samplers (flow rates>0.5 m3/min). Lower volume samplers (flowrates <0.5 m3/min) will require a moresensitive balance.8.0 Calibration.

8.1 General Requirements.8.1.1 Calibration of the sampler’s flow

measurement device is required to establishtraceability of subsequent flow measurementsto a primary standard. A flow rate transferstandard calibrated against a primary flow orvolume standard shall be used to calibrate orverify the accuracy of the sampler’s flowmeasurement device.

8.1.2 Particle size discrimination byinertial separation requires that specific airvelocities be maintained in the sampler’s airinlet system. Therefore, the flow rate throughthe sampler’s inlet must be maintainedthroughout the sampling period within thedesign flow rate range specified by themanufacturer. Design flow rates are specifiedas actual volumetric flow rates, measured atexisting conditions of temperature andpressure (Qa).

8.2 Flow Rate Calibration Procedure.8.2.1 PM10 samplers employ various types

of flow control and flow measurementdevices. The specific procedure used for flowrate calibration or verification will varydepending on the type of flow controller andflow rate indicator employed. Calibration isin terms of actual volumetric flow rates (Qa)to meet the requirements of section 8.1 of thisappendix. The general procedure given hereserves to illustrate the steps involved in thecalibration. Consult the samplermanufacturer’s instruction manual andreference 2 of section 12.0 of this appendixfor specific guidance on calibration.Reference 14 of section 12.0 of this appendixprovides additional information on variousother measures of flow rate and theirinterrelationships.

8.2.2 Calibrate the flow rate transferstandard against a primary flow or volumestandard traceable to NIST. Establish acalibration relationship, e.g., an equation orfamily of curves, such that traceability to theprimary standard is accurate to within 2percent over the expected range of ambientconditions, i.e., temperatures and pressures,under which the transfer standard will beused. Recalibrate the transfer standardperiodically.

8.2.3 Following the samplermanufacturer’s instruction manual, remove

the sampler inlet and connect the flow ratetransfer standard to the sampler such that thetransfer standard accurately measures thesampler’s flow rate. Make sure there are noleaks between the transfer standard and thesampler.

8.2.4 Choose a minimum of three flowrates (actual m3/min), spaced over theacceptable flow rate range specified for theinlet, under section 7.1.2 of the appendix,that can be obtained by suitable adjustmentof the sampler flow rate. In accordance withthe sampler manufacturer’s instructionmanual, obtain or verify the calibrationrelationship between the flow rate (actualm3/min) as indicated by the transfer standardand the sampler’s flow indicator response.Record the ambient temperature andbarometric pressure. Temperature andpressure corrections to subsequent flowindicator readings may be required forcertain types of flow measurement devices.When such corrections are necessary,correction on an individual or daily basis ispreferable. However, seasonal averagetemperature and average barometric pressurefor the sampling site may be incorporatedinto the sampler calibration to avoid dailycorrections. Consult the samplermanufacturer’s instruction manual andreference 2 in section 12.0 of this appendixfor additional guidance.

8.2.5 Following calibration, verify thatthe sampler is operating at its design flowrate (actual m3/min) with a clean filter inplace.

8.2.6 Replace the sampler inlet.9.0 Procedure.

9.1 The sampler shall be operated inaccordance with the specific guidanceprovided in the sampler manufacturer’sinstruction manual and in reference 2 insection 12.0 of this appendix. The generalprocedure given here assumes that thesampler’s flow rate calibration is based onflow rates at ambient conditions (Qa) andserves to illustrate the steps involved in theoperation of a PM10 sampler.

9.2 Inspect each filter for pinholes,particles, and other imperfections. Establisha filter information record and assign anidentification number to each filter.

9.3 Equilibrate each filter in theconditioning environment (see 7.4) for atleast 24 hours.

9.4 Following equilibration, weigh eachfilter and record the presampling weight withthe filter identification number.

9.5 Install a preweighed filter in thesampler following the instructions providedin the sampler manufacturer’s instructionmanual.

9.6 (a) Turn on the sampler and allow itto establish run-temperature conditions.Record the flow indicator reading and, ifneeded, the ambient temperature andbarometric pressure. Determine the samplerflow rate (actual m3/min) in accordance withthe instructions provided in the samplermanufacturer’s instruction manual.

(b) Note: No onsite temperature or pressuremeasurements are necessary if the sampler’sflow indicator does not require temperatureor pressure corrections or if seasonal averagetemperature and average barometric pressurefor the sampling site are incorporated into


the sampler calibration, under section 8.2.4of this appendix. If individual or dailytemperature and pressure corrections arerequired, ambient temperature andbarometric pressure can be obtained by on-site measurements or from a nearby weatherstation. Barometric pressure readingsobtained from airports must be stationpressure, not corrected to sea level, and mayneed to be corrected for differences inelevation between the sampling site and theairport.

9.7 If the flow rate is outside theacceptable range specified by themanufacturer, check for leaks, and ifnecessary, adjust the flow rate to thespecified setpoint. Stop the sampler.

9.8 Set the timer to start and stop thesampler at appropriate times. Set the elapsedtime meter to zero or record the initial meterreading.

9.9 Record the sample information (sitelocation or identification number, sampledate, filter identification number, andsampler model and serial number).

9.10 Sample for 24±1 hours.9.11 Determine and record the average

flow rate (Q̄a) in actual m3/min for thesampling period in accordance with theinstructions provided in the samplermanufacturer’s instruction manual. Recordthe elapsed time meter final reading and, ifneeded, the average ambient temperature andbarometric pressure for the sampling period,in note following section 9.6 of thisappendix.

9.12 Carefully remove the filter from thesampler, following the samplermanufacturer’s instruction manual. Touchonly the outer edges of the filter.

9.13 Place the filter in a protective holderor container, e.g., petri dish, glassineenvelope, or manila folder.

9.14 Record any factors such asmeteorological conditions, constructionactivity, fires or dust storms, etc., that mightbe pertinent to the measurement on the filterinformation record.

9.15 Transport the exposed sample filterto the filter conditioning environment assoon as possible for equilibration andsubsequent weighing.

9.16 Equilibrate the exposed filter in theconditioning environment for at least 24hours under the same temperature andhumidity conditions used for presamplingfilter equilibration (see section 9.3 of thisappendix).

9.17 Immediately after equilibration,reweigh the filter and record thepostsampling weight with the filteridentification number.10.0 Sampler Maintenance.

10.1 The PM10 sampler shall bemaintained in strict accordance with themaintenance procedures specified in thesampler manufacturer’s instruction manual.11.0 Calculations.

11.1 Calculate the total volume of airsampled as:

V = Qat

where:

V = total air sampled, at ambient temperatureand pressure,m3;

Qa = average sample flow rate at ambienttemperature and pressure, m3/min; and

t = sampling time, min.

11.2 (a) Calculate the PM10 concentrationas:

PM10 = (Wf¥Wi)×106/V

where:

PM10 = mass concentration of PM10, µg/m3;

Wf, Wi = final and initial weights of filtercollecting PM1O particles, g; and

106 = conversion of g to µg.

(b) Note: If more than one size fraction inthe PM10 size range is collected by thesampler, the sum of the net weight gain byeach collection filter [Σ(Wf¥Wi)] is used tocalculate the PM10 mass concentration.12.0 References.

1. Quality Assurance Handbook for AirPollution Measurement Systems, Volume I,Principles. EPA–600/9–76–005, March 1976.Available from CERI, ORD Publications, U.S.Environmental Protection Agency, 26 WestSt. Clair Street, Cincinnati, OH 45268.

2. Quality Assurance Handbook for AirPollution Measurement Systems, Volume II,Ambient Air Specific Methods. EPA–600/4–77–027a, May 1977. Available from CERI,ORD Publications, U.S. EnvironmentalProtection Agency, 26 West St. Clair Street,Cincinnati, OH 45268.

3. Clement, R.E., and F.W. Karasek. SampleComposition Changes in Sampling andAnalysis of Organic Compounds in Aerosols.Int. J. Environ. Analyt. Chem., 7:109, 1979.

4. Lee, R.E., Jr., and J. Wagman. ASampling Anomaly in the Determination ofAtmospheric Sulfate Concentration. Amer.Ind. Hyg. Assoc. J., 27:266, 1966.

5. Appel, B.R., S.M. Wall, Y. Tokiwa, andM. Haik. Interference Effects in SamplingParticulate Nitrate in Ambient Air. Atmos.Environ., 13:319, 1979.

6. Coutant, R.W. Effect of EnvironmentalVariables on Collection of AtmosphericSulfate. Environ. Sci. Technol., 11:873, 1977.

7. Spicer, C.W., and P. Schumacher.Interference in Sampling AtmosphericParticulate Nitrate. Atmos. Environ., 11:873,1977.

8. Appel, B.R., Y. Tokiwa, and M. Haik.Sampling of Nitrates in Ambient Air. Atmos.Environ., 15:283, 1981.

9. Spicer, C.W., and P.M. Schumacher.Particulate Nitrate: Laboratory and FieldStudies of Major Sampling Interferences.Atmos. Environ., 13:543, 1979.

10. Appel, B.R. Letter to Larry Purdue, U.S.EPA, Environmental Monitoring and SupportLaboratory. March 18, 1982, Docket No. A–82–37, II–I–1.

11. Pierson, W.R., W.W. Brachaczek, T.J.Korniski, T.J. Truex, and J.W. Butler. ArtifactFormation of Sulfate, Nitrate, and HydrogenIon on Backup Filters: Allegheny MountainExperiment. J. Air Pollut. Control Assoc.,30:30, 1980.

12. Dunwoody, C.L. Rapid Nitrate LossFrom PM10 Filters. J. Air Pollut. ControlAssoc., 36:817, 1986.

13. Harrell, R.M. Measuring the Alkalinityof Hi-Vol Air Filters. EMSL/RTP–SOP–QAD–534, October 1985. Available from the U.S.Environmental Protection Agency, EMSL/QAD, Research Triangle Park, NC 27711.

14. Smith, F., P.S. Wohlschlegel, R.S.C.Rogers, and D.J. Mulligan. Investigation ofFlow Rate Calibration Procedures AssociatedWith the High Volume Method forDetermination of Suspended Particulates.EPA–600/4–78–047, U.S. EnvironmentalProtection Agency, Research Triangle Park,NC 27711, 1978.

8. Appendix N is added to read asfollows:

Appendix N to Part 50—Interpretationof the National Ambient Air QualityStandards for Particulate Matter

1.0 General.(a) This appendix explains the data

handling conventions and computationsnecessary for determining when the annualand 24-hour primary and secondary nationalambient air quality standards for PMspecified in § 50.7 of this chapter are met.Particulate matter is measured in the ambientair as PM10 and PM2.5 (particles with anaerodynamic diameter less than or equal toa nominal 10 and 2.5 micrometers,respectively) by a reference method based onAppendix M of this part for PM10 and onAppendix L of this part for PM2.5, asapplicable, and designated in accordancewith part 53 of this chapter, or by anequivalent method designated in accordancewith part 53 of this chapter. Data handlingand computation procedures to be used inmaking comparisons between reported PM10

and PM2.5 concentrations and the levels ofthe PM standards are specified in thefollowing sections.

(b) Data resulting from uncontrollable ornatural events, for example structural fires orhigh winds, may require specialconsideration. In some cases, it may beappropriate to exclude these data becausethey could result in inappropriate values tocompare with the levels of the PM standards.In other cases, it may be more appropriate toretain the data for comparison with the levelof the PM standards and then allow the EPAto formulate the appropriate regulatoryresponse. Whether to exclude, retain, ormake adjustments to the data affected byuncontrollable or natural events is subject tothe approval of the appropriate RegionalAdministrator.

(c) The terms used in this appendix aredefined as follows:

Average and mean refer to an arithmeticmean.

Daily value for PM refers to the 24-houraverage concentration of PM calculated ormeasured from midnight to midnight (localtime) for PM10 or PM2.5.

Designated monitors are those monitoringsites designated in a State PM MonitoringNetwork Description for spatial averaging inareas opting for spatial averaging inaccordance with part 58 of this chapter.

98th percentile (used for PM2.5) means thedaily value out of a year of monitoring databelow which 98 percent of all values in thegroup fall.


99th percentile (used for PM10) means thedaily value out of a year of monitoring databelow which 99 percent of all values in thegroup fall.

Year refers to a calendar year.(d) Sections 2.1 and 2.5 of this appendix

contain data handling instructions for theoption of using a spatially averaged networkof monitors for the annual standard. If spatialaveraging is not considered for an area, thenthe spatial average is equivalent to the annualaverage of a single site and is treatedaccordingly in subsequent calculations. Forexample, paragraph (a)(3) of section 2.1 ofthis appendix could be eliminated since thespatial average would be equivalent to theannual average.2.0 Comparisons with the PM2.5 Standards.

2.1 Annual PM2.5 Standard.(a) The annual PM2.5 standard is met when

the 3-year average of the spatially averagedannual means is less than or equal to 15.0 µg/m3. The 3-year average of the spatiallyaveraged annual means is determined byaveraging quarterly means at each monitor toobtain the annual mean PM2.5 concentrationsat each monitor, then averaging across alldesignated monitors, and finally averagingfor 3 consecutive years. The steps can besummarized as follows:

(1) Average 24-hour measurements toobtain quarterly means at each monitor.

(2) Average quarterly means to obtainannual means at each monitor.

(3) Average across designated monitoringsites to obtain an annual spatial mean for anarea (this can be one site in which case thespatial mean is equal to the annual mean).

(4) Average 3 years of annual spatial meansto obtain a 3-year average of spatiallyaveraged annual means.

(b) In the case of spatial averaging, 3 yearsof spatial averages are required todemonstrate that the standard has been met.Designated sites with less than 3 years of datashall be included in spatial averages for thoseyears that data completeness requirementsare met. For the annual PM2.5 standard, a yearmeets data completeness requirements whenat least 75 percent of the scheduled samplingdays for each quarter have valid data.However, years with high concentrations andmore than a minimal amount of data (at least11 samples in each quarter) shall not beignored just because they are comprised ofquarters with less than complete data. Thus,in computing annual spatially averagedmeans, years containing quarters with at least11 samples but less than 75 percent datacompleteness shall be included in thecomputation if the resulting spatiallyaveraged annual mean concentration(rounded according to the conventions ofsection 2.3 of this appendix) is greater thanthe level of the standard.

(c) Situations may arise in which there arecompelling reasons to retain years containingquarters which do not meet the datacompleteness requirement of 75 percent orthe minimum number of 11 samples. The useof less than complete data is subject to theapproval of the appropriate RegionalAdministrator.

(d) The equations for calculating the 3-yearaverage annual mean of the PM2.5 standardare given in section 2.5 of this appendix.

2.2 24-Hour PM2.5 Standard.(a) The 24-hour PM2.5 standard is met

when the 3-year average of the 98th percentilevalues at each monitoring site is less than orequal to 65 µg/m3. This comparison shall bebased on 3 consecutive, complete years of airquality data. A year meets data completenessrequirements when at least 75 percent of thescheduled sampling days for each quarterhave valid data. However, years with highconcentrations shall not be ignored justbecause they are comprised of quarters withless than complete data. Thus, in computingthe 3-year average 98th percentile value, yearscontaining quarters with less than 75 percentdata completeness shall be included in thecomputation if the annual 98th percentilevalue (rounded according to the conventionsof section 2.3 of this appendix) is greater thanthe level of the standard.

(b) Situations may arise in which there arecompelling reasons to retain years containingquarters which do not meet the datacompleteness requirement. The use of lessthan complete data is subject to the approvalof the appropriate Regional Administrator.

(c) The equations for calculating the 3-yearaverage of the annual 98th percentile valuesis given in section 2.6 of this appendix.

2.3 Rounding Conventions. For thepurposes of comparing calculated values tothe applicable level of the standard, it isnecessary to round the final results of thecalculations described in sections 2.5 and 2.6of this appendix. For the annual PM2.5

standard, the 3-year average of the spatiallyaveraged annual means shall be rounded tothe nearest 0.1 µg/m3 (decimals 0.05 andgreater are rounded up to the next 0.1, andany decimal lower than 0.05 is roundeddown to the nearest 0.1). For the 24-hourPM2.5 standard, the 3-year average of theannual 98th percentile values shall berounded to the nearest 1 µg/m3 (decimals 0.5and greater are rounded up to nearest wholenumber, and any decimal lower than 0.5 isrounded down to the nearest whole number).

2.4 Monitoring Considerations.(a) Section 58.13 of this chapter specifies

the required minimum frequency of samplingfor PM2.5. Exceptions to the specifiedsampling frequencies, such as a reducedfrequency during a season of expected lowconcentrations, are subject to the approval ofthe appropriate Regional Administrator.Section 58.14 of 40 CFR part 58 and section2.8 of Appendix D of 40 CFR part 58, specifywhich monitors are eligible for makingcomparisons with the PM standards. Indetermining a spatial mean using two ormore monitoring sites operating in a givenyear, the annual mean for an individual sitemay be included in the spatial mean if andonly if the mean for that site meets thecriterion specified in § 2.8 of Appendix D of40 CFR part 58. In the event data from anotherwise eligible site is excluded from beingaveraged with data from other sites on thebasis of this criterion, then the 3-year meanfrom that site shall be compared directly tothe annual standard.

(b) For the annual PM2.5 standard, whendesignated monitors are located at the samesite and are reporting PM2.5 values for thesame time periods, and when spatialaveraging has been chosen, their

concentrations shall be averaged before anarea-wide spatial average is calculated. Suchmonitors will then be considered as onemonitor.

2.5 Equations for the Annual PM2.5

Standard.(a) An annual mean value for PM2.5 is

determined by first averaging the daily valuesof a calendar quarter:

Equation 1

xn

xq y sq

i q y si

nq

, , , , ,==∑1

1

where:

x̄q,y,s = the mean for quarter q of year y forsite s;

nq = the number of monitored values in thequarter; and

xi,q,y,s = the ith value in quarter q for year yfor site s.

(b) The following equation is then to beused for calculation of the annual mean:

Equation 2

x xy s q y sq

, , ,==

∑1

4 1

4

where:

x̄y,s = the annual mean concentration for yeary (y = 1, 2, or 3) and for site s; and

x̄q,y,s = the mean for quarter q of year y forsite s.

(c) (1) The spatially averaged annual meanfor year y is computed by first calculating theannual mean for each site designated to beincluded in a spatial average, x̄y,s, and thencomputing the average of these values acrosssites:

Equation 3

xn

xys

y ss

ns

==∑1

1,

where:

x̄y = the spatially averaged mean for year y;

x̄y,s = the annual mean for year y and site s;and

ns = the number of sites designated to beaveraged.

(2) In the event that an area designated forspatial averaging has two or more sites at thesame location producing data for the sametime periods, the sites are averaged togetherbefore using Equation 3 by:

Equation 4

xn

xy s*c

y ss

nc

, ,==∑1

1

where:

x̄y,s* = the annual mean for year y for the sitesat the same location (which will now beconsidered one site);


nc = the number of sites at the same locationdesignated to be included in the spatialaverage; and

x̄y,s = the annual mean for year y and site s.

(d) The 3-year average of the spatiallyaveraged annual means is calculated by usingthe following equation:

Equation 5

x xyy

==

∑1

3 1

3

where:

x̄ = the 3-year average of the spatiallyaveraged annual means; and

x̄y = the spatially averaged annual mean foryear y.

Example 1—Area Designated for SpatialAveraging That Meets the Primary AnnualPM2.5 Standard.

a. In an area designated for spatialaveraging, four designated monitors recordeddata in at least 1 year of a particular 3-yearperiod. Using Equations 1 and 2, the annualmeans for PM2.5 at each site are calculated foreach year. The following table can be createdfrom the results. Data completenesspercentages for the quarter with the fewestnumber of samples are also shown.

Table 1.—Results from Equations 1 and 2

Site #1 Site #2 Site #3 Site #4 Spatial mean

Year 1 .............................. Annual mean (µg/m3) ....................... 12.7 ...................... ...................... ...................... 12.7% data completeness ....................... 80 0 0 0 ......................

Year 2 .............................. Annual mean (µg/m3) ....................... 12.6 17.5 15.2 ...................... 15.05% data completeness ....................... 90 63 38 0 ......................

Year 3 .............................. Annual mean (µg/m3) ....................... 12.5 18.5 14.1 16.9 15.50% data completeness ....................... 90 80 85 50 ......................

3-year mean ..................... ........................................................... ...................... ...................... ...................... ...................... 14.42

b. The data from these sites are averagedin the order described in section 2.1 of thisappendix. Note that the annual mean fromsite #3 in year 2 and the annual mean fromsite #4 in year 3 do not meet the 75 percentdata completeness criteria. Assuming the 38percent data completeness represents aquarter with fewer than 11 samples, site #3in year 2 does not meet the minimum datacompleteness requirement of 11 samples ineach quarter. The site is therefore excluded

from the calculation of the spatial mean foryear 2. However, since the spatial mean foryear 3 is above the level of the standard andthe minimum data requirement of 11 samplesin each quarter has been met, the annualmean from site #4 in year 3 is included inthe calculation of the spatial mean for year3 and in the calculation of the 3-year average.The 3-year average is rounded to 14.4 µg/m3,indicating that this area meets the annualPM2.5 standard.

Example 2—Area With Two Monitors at theSame Location That Meets the PrimaryAnnual PM2.5 Standard.

a. In an area designated for spatialaveraging, six designated monitors, with twomonitors at the same location (#5 and #6),recorded data in a particular 3-year period.Using Equations 1 and 2, the annual meansfor PM2.5 are calculated for each year. Thefollowing table can be created from theresults.

Table 2.—Results From Equations 1 and 2

Annual mean (µg/m3) Site #1 Site #2 Site #3 Site #4 Site #5 Site #6 Average of#5 and #6

Spatialmean

Year 1 .................................... 12.9 9.9 12.6 11.1 14.5 14.6 14.55 12.21Year 2 .................................... 14.5 13.3 12.2 10.9 16.1 16.0 16.05 13.39Year 3 .................................... 14.4 12.4 11.5 9.7 12.3 12.1 12.20 12.043-Year mean .......................... .................... .................... .................... .................... .................... .................... .................. 12.55

b. The annual means for sites #5 and #6 areaveraged together using Equation 4 before thespatial average is calculated using Equation3 since they are in the same location. The 3-year mean is rounded to 12.6 µg/m3,indicating that this area meets the annualPM2.5 standard.

Example 3—Area With a Single Monitor ThatMeets the Primary Annual PM2.5 Standard.

a. Given data from a single monitor in anarea, the calculations are as follows. UsingEquations 1 and 2, the annual means forPM2.5 are calculated for each year. If the

annual means are 10.28, 17.38, and 12.25 µg/m3, then the 3-year mean is:

x g m= × =(1 / 3) (10.28 +17.38 +12.25) 13.303 µ / .3

b. This value is rounded to 13.3, indicatingthat this area meets the annual PM2.5

standard.2.6 Equations for the 24-Hour PM2.5

Standard.(a) When the data for a particular site and

year meet the data completenessrequirements in section 2.2 of this appendix,calculation of the 98th percentile isaccomplished by the following steps. All thedaily values from a particular site and yearcomprise a series of values (x1, x2, x3, ..., xn),

that can be sorted into a series where eachnumber is equal to or larger than thepreceding number (x[1], x[2], x[3], ..., x[n]). Inthis case, x[1] is the smallest number and x[n]

is the largest value. The 98th percentile isfound from the sorted series of daily valueswhich is ordered from the lowest to thehighest number. Compute (0.98) × (n) as thenumber ‘‘i.d’’, where ‘‘i’’ is the integer partof the result and ‘‘d’’ is the decimal part ofthe result. The 98th percentile value for yeary, P0.98, y, is given by Equation 6:

Equation 6

P Xy i0 98 1. , = +[ ]where:P0.98,y = 98th percentile for year y;

x[i∂1] = the (i+1)th number in the orderedseries of numbers; and

i = the integer part of the product of 0.98 andn.


(b) The 3-year average 98th percentile isthen calculated by averaging the annual 98th

percentiles:

Equation 7

P

P yy

0 98

0 981

3

3.

. ,

= =∑

(c) The 3-year average 98th percentile isrounded according to the conventions insection 2.3 of this appendix before acomparison with the standard is made.

Example 4—Ambient Monitoring Site WithEvery-Day Sampling That Meets the Primary24-Hour PM2.5 Standard.

a. In each year of a particular 3 year period,varying numbers of daily PM2.5 values (e.g.,

281, 304, and 296) out of a possible 365values were recorded at a particular site withthe following ranked values (in µg/m3):

Table 3.—Ordered Monitoring Data For 3 Years

Year 1 Year 2 Year 3

j rank Xj value j rank Xj value j rank Xj value

275 57.9 296 54.3 290 66.0276 59.0 297 57.1 291 68.4277 62.2 298 63.0 292 69.8

b. Using Equation 6, the 98th percentilevalues for each year are calculated as follows:

0 98 281 1 276 59 00 98 1 2763. . /. , 275.38× = ⇒ + = ⇒ = =[ ]i P X g mµ

0 98 304 1 298 63 00 98 2 2983. . /. , 297.92× = ⇒ + = ⇒ = =[ ]i P X g mµ

0 98 296 1 291 680 98 3 2913. .4 /. , 290.07× = ⇒ + = ⇒ = =[ ]i P X g mµ

c. 1. Using Equation 7, the 3-year average98th percentile is calculated as follows:

P g m g m0 983 359 0 63 0 68

363.

. . .4.46 / /= + + = µ µ, which rounds to 63 .

2. Therefore, this site meets the 24-hourPM2.5 standard.3.0 Comparisons with the PM10 Standards.

3.1 Annual PM10 Standard.(a) The annual PM10 standard is met when

the 3-year average of the annual mean PM10

concentrations at each monitoring site is lessthan or equal to 50 µg/m3. The 3-year averageof the annual means is determined byaveraging quarterly means to obtain annualmean PM10 concentrations for 3 consecutive,complete years at each monitoring site. Thesteps can be summarized as follows:

(1) Average 24-hour measurements toobtain a quarterly mean.

(2) Average quarterly means to obtain anannual mean.

(3) Average annual means to obtain a 3-year mean.

(b) For the annual PM10 standard, a yearmeets data completeness requirements whenat least 75 percent of the scheduled samplingdays for each quarter have valid data.However, years with high concentrations andmore than a minimal amount of data (at least11 samples in each quarter) shall not beignored just because they are comprised of

quarters with less than complete data. Thus,in computing the 3-year average annual meanconcentration, years containing quarters withat least 11 samples but less than 75 percentdata completeness shall be included in thecomputation if the annual meanconcentration (rounded according to theconventions of section 2.3 of this appendix)is greater than the level of the standard.

(c) Situations may arise in which there arecompelling reasons to retain years containingquarters which do not meet the datacompleteness requirement of 75 percent orthe minimum number of 11 samples. The useof less than complete data is subject to theapproval of the appropriate RegionalAdministrator.

(d) The equations for calculating the 3-yearaverage annual mean of the PM10 standardare given in section 3.5 of this appendix.

3.2 24-Hour PM10 Standard.(a) The 24-hour PM10 standard is met when

the 3-year average of the annual 99th

percentile values at each monitoring site isless than or equal to 150 µg/m3. Thiscomparison shall be based on 3 consecutive,complete years of air quality data. A year

meets data completeness requirements whenat least 75 percent of the scheduled samplingdays for each quarter have valid data.However, years with high concentrationsshall not be ignored just because they arecomprised of quarters with less thancomplete data. Thus, in computing the 3-yearaverage of the annual 99th percentile values,years containing quarters with less than 75percent data completeness shall be includedin the computation if the annual 99th

percentile value (rounded according to theconventions of section 2.3 of this appendix)is greater than the level of the standard.

(b) Situations may arise in which there arecompelling reasons to retain years containingquarters which do not meet the datacompleteness requirement. The use of lessthan complete data is subject to the approvalof the appropriate Regional Administrator.

(c) The equation for calculating the 3-yearaverage of the annual 99th percentile valuesis given in section 2.6 of this appendix.

3.3 Rounding Conventions. For the annualPM10 standard, the 3-year average of theannual PM10 means shall be rounded to thenearest 1 µg/m3 (decimals 0.5 and greater are


rounded up to the next whole number, andany decimal less than 0.5 is rounded downto the nearest whole number). For the 24-hour PM10 standard, the 3-year average of theannual 99th percentile values of PM10 shall berounded to the nearest 10 µg/m3 (155 µg/m3

and greater would be rounded to 160 µg/m3

and 154 µg/m3 and less would be rounded to150 µg/m3).

3.4 Monitoring Considerations. Section58.13 of this chapter specifies the requiredminimum frequency of sampling for PM10.Exceptions to the specified samplingfrequencies, such as a reduced frequencyduring a season of expected lowconcentrations, are subject to the approval ofthe appropriate Regional Administrator. Formaking comparisons with the PM10 NAAQS,all sites meeting applicable requirements inpart 58 of this chapter would be used.

3.5 Equations for the Annual PM10

Standard.(a) An annual arithmetic mean value for

PM10 is determined by first averaging the 24-hour values of a calendar quarter using thefollowing equation:

Equation 8

xn

xq yq

i q yi

nq

, , ,==∑1

1

where:x̄q,y = the mean for quarter q of year y;

nq = the number of monitored values in thequarter; and

xi,q,y = the ith value in quarter q for year y.

(b) The following equation is then to beused for calculation of the annual mean:

Equation 9

x xy q yq

==

∑1

4 1

4

,

where:

x̄y = the annual mean concentration for yeary, (y=1, 2, or 3); and

xq,y = the mean for a quarter q of year y.

(c) The 3-year average of the annual meansis calculated by using the following equation:

Equation 10

x xyy

==

∑1

3 1

3

where:

x̄ = the 3-year average of the annual means;and

x̄y = the annual mean for calendar year y.

Example 5—Ambient Monitoring Site ThatDoes Not Meet the Annual PM10 Standard.

a. Given data from a PM10 monitor andusing Equations 8 and 9, the annual meansfor PM10 are calculated for each year. If theannual means are 52.42, 82.17, and 63.23 µg/m3, then the 3-year average annual mean is:

x g m= × + + =(1 / 3) (52.42 82.17 63.23) 65.94, which is rounded to 66 µ / .3

b. Therefore, this site does not meet theannual PM10 standard.

3.6 Equation for the 24-Hour PM10

Standard.(a) When the data for a particular site and

year meet the data completenessrequirements in section 3.2 of this appendix,calculation of the 99th percentile isaccomplished by the following steps. All thedaily values from a particular site and yearcomprise a series of values (x1, x2, x3, ..., xn)that can be sorted into a series where eachnumber is equal to or larger than thepreceding number (x[1], x[2], x[3], ..., x[n]). Inthis case, x[1] is the smallest number and x[n]is the largest value. The 99th percentile isfound from the sorted series of daily valueswhich is ordered from the lowest to thehighest number. Compute (0.99) × (n) as thenumber ‘‘i.d’’, where ‘‘i’’ is the integer part

of the result and ‘‘d’’ is the decimal part ofthe result. The 99th percentile value for yeary, P0.99,y, is given by Equation 11:

Equation 11

P Xy i0 99 1. , = +[ ]where:

P0.99,y = the 99th percentile for year y;

x[i∂1] = the (i+1)th number in the orderedseries of numbers; and

i = the integer part of the product of 0.99 andn.

(b) The 3-year average 99th percentile valueis then calculated by averaging the annual99th percentiles:

Equation 12

P

P yy

0 99

0 991

3

3.

. ,

= =∑

(c) The 3-year average 99th percentile isrounded according to the conventions insection 3.3 of this appendix before acomparison with the standard is made.

Example 6—Ambient Monitoring Site WithSampling Every Sixth Day That Meets thePrimary 24-Hour PM10 Standard.

a. In each year of a particular 3 year period,varying numbers of PM10 daily values (e.g.,110, 98, and 100) out of a possible 121 dailyvalues were recorded at a particular site withthe following ranked values (in µg/m3):

Table 4.—Ordered Monitoring Data For 3 Years

Year 1 Year 2 Year 3

j rank Xj value j rank Xj value j rank Xj value

108 120 96 143 98 140109 128 97 148 99 144110 130 98 150 100 147

b. Using Equation 11, the 99th percentilevalues for each year are calculated as follows:

0 99 1 109 1280 99 1 1093. /. , 110 = 108.9 × ⇒ + = ⇒ = =[ ]i P X g mµ

0 99 1 98 1500 99 2 983. /. , 98 = 97.02 × ⇒ + = ⇒ = =[ ]i P X g mµ


0 99 1 100 1470 99 3 1003. /. , 100 = 99× ⇒ + = ⇒ = =[ ]i P X g mµ

c. 1. Using Equation 12, the 3-year average99th percentile is calculated as follows:

128 50 147

3141 7 1403 3+ + = . / / rounds to .µ µg m g m

2. Therefore, this site meets the 24-hourPM10 standard.

[FR Doc. 97–18577 Filed 7–17–97; 8:45 am]BILLING CODE 6560–50–F

Date post:	17-Feb-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times