United StatesEnvironmental ProtectionAgency
Air and Radiation EPA420-D-02-004October 2002
Diesel PMModel-To-MeasurementComparison
Printed on Recycled Paper
EPA420-D-02-004October 2002
Diesel PM Model-To-Measurement Comparison
Assessment and Standards DivisionOffice of Transportation and Air QualityU.S. Environmental Protection Agency
Prepared for EPA byICF Consulting
EPA Contract No. 68-C-01-164Work Assignment No. 0-5
NOTICE
This technical report does not necessarily represent final EPA decisions or positions.It is intended to present technical analysis of issues using data thatC are currently available.
The purpose in the release of such reports is to facilitate the exchange oftechnical information and to inform the public of technical developments which
may form the basis for a final EPA decision, position, or regulatory action.
Final Report
DIESEL PM MODEL-TO-MEASUREMENTCOMPARISON
EPA Contract 68-C-01-164Work Assignment No. 0-5
September 30, 2002
Prepared for:
James Richard Cook, Chad Bailey, Tesh RaoUSEPA Office of Transportation and Air Quality
2000 Traverwood DriveAnn Arbor, Michigan, 48105
Prepared by:
Jonathan Cohen, Christine Lee, Seshasai KanchiICF Consulting
101 Lucas Valley Road, Suite 230San Rafael, CA 94903
Rob KlausmeierdKC – de la Torre Klausmeier Consulting, Inc
1401 Foxtail Cove, Austin, TX 78704
ICF Consulting 2
Diesel PM Model-To-Measurement ComparisonEPA Contract 68-C-01-164, Work Assignment No. 0-5
CHAPTER 1. INTRODUCTION AND SUMMARY
The purpose of this project was to compare estimated diesel particulate matter (diesel PM or DPM)concentrations based on elemental carbon (EC) and black carbon (BC) data with modeled ambientconcentrations of DPM from the 1996 National-Scale Air Toxics Assessment (NSATA).
The NSATA used DPM inventory estimates from EPA’s final rule promulgating 2007 heavy dutyengine standards. Using the ASPEN dispersion model, NSATA developed estimates of 1996 annualaverage concentrations of DPM at census tracts nationwide. The goal of this project was to evaluatethe reasonableness of DPM estimates from dispersion models for this case by comparing theNSATA DPM concentration estimates with estimates based on measured EC and BCconcentrations.
EC measurements can be obtained from PM2.5 monitoring sites that sample PM2.5 using quartzfiber media. The EC is measured using thermo-optical analysis of the carbonaceous material. Manystudies have used thermal optimal transmission (TOT), the NIOSH method developed at Sunsetlaboratories. Some studies have used thermal optical reflectance (TOR), a method developed byDesert Research Institute. In addition, some sites measure ambient BC with an Aethalometer.EPA’s Office of Air Quality Planning and Standards is reviewing the measurement of ECthrough the Speciation Trends Network, and an Agency statement on the issue is forthcoming.For now, however, existing values developed using the TOT method are being used.
All these carbon concentration measurements can be used to estimate ambient DPM by usingconversion factors based on 1) source apportionment studies, 2) source-receptor model studies, and3) studies which examine the fraction of EC in DPM.
Our analysis was carried out as a series of steps that are detailed in this report:
1. A nationwide database was compiled containing elemental carbon (EC), organic carbon(OC), and black carbon (BC) concentration measurements from PM2.5 monitoring sites fromJanuary 1994 to December 2001. The database includes daily, annual, seasonal,weekday/weekend, and overall average concentrations and other summary statistics for eachmonitoring site.
2. Using results from several source apportionment studies, multiplicative conversion factorsto estimate diesel particulate matter (DPM) from EC, OC, and BC concentrations werecompiled. Average conversion factors were compiled together with lower and upper boundvalues.
3. Based on the results of steps 1 and 2, average, minimum, and maximum estimates of theoverall average DPM at each monitoring site were computed.
ICF Consulting 3
4. The monitored values of the DPM (derived in step 3) were statistically compared withmodeled values from the NSATA at the nearest census tract centroid and at the census tractcentroid with the maximum modeled value within 30 km. The goal of the comparison wasto determine if the modeled DPM concentrations in the NSATA agree reasonably well withestimates from monitored data.
Data Base
As described in Chapter 2, ICF developed a database of EC and BC measurements from PM2.5monitoring sites, consisting of EC and BC data collected in the time period from January 1, 1994to December 31, 2001. The database has been provided to EPA in a .DBF format, although ICF’sstatistical analyses were performed in SAS using a SAS database. The data sources withcurrently available data included 76 EPA PM speciation sampling sites, the Northern FrontRange Air Quality Study (NFRAQS), the Phoenix EPA PM Supersite, Interagency Monitoring ofProtected Visual Environments (IMPROVE), Clean Air Status and Trends Network(CASTNET), the California Multiple Air Toxics Exposure Study in the South Coast Air Basin(MATESii), and the 1995 Integrated Monitoring Study (IMS95). Data were also obtained fromone of EPA’s EMPACT grant recipients, Airbeat.
The EC and BC values “below the detection limit” were replaced by one half of the minimumdetection limit (MDL). Missing data were not used or substituted for, to avoid biasing theestimated standard deviations.
The database includes the following information, where available:
1. For each site measuring carbon on quartz fiber, the method by which EC and OCfractions (EC/OC) are determined
2. Latitude and longitude coordinates of the monitor3. Whether the monitor is in an urban or rural tract, based on NSATA assignments4. The minimum detection limit of the monitor and analytic method5. Monitor start and end dates6. Summary statistics of daily average EC, OC, or BC measurements for all data at the site
and also stratified by 1) year, 2) calendar quarter, 3) year and quarter, 4) weekday andweekend. The following summary statistics were obtained: mean, median, standarddeviation, geometric mean, geometric standard deviation, minimum, 10th, … 90th
percentile, maximum.7. The same set of summary statistics were obtained for ECOCX, an EC concentration value
developed from the EPA PM TOT speciation data to estimate the corresponding TORvalue, and for the various monitored DPM estimates computed by applying correctionfactors (described below) to the EC values.
8. At each PM2.5 monitoring site, the fractions of PM2.5 which are elemental carbon,organic carbon, sulfates, and nitrates.
ICF Consulting 4
Conversion Factors
As described in Chapter 3, ICF developed multiplicative “conversion factors” (CFs) forestimating ambient DPM based on the ambient EC, BC or OC measurements. For each site andcarbon type (EC, OC, or BC), a low-end, most likely (“average”), and high-end CF wasassigned, as discussed below. For EC sites, the estimated ambient DPM-high equals ambient ECmultiplied by the high-end CF for EC, the estimated ambient DPM-low equals ambient ECmultiplied by the low-end CF for EC, and the estimated ambient DPM-avg equals ambient ECmultiplied by the most likely CF for EC. BC and OC conversion factors were tabulated but notapplied to the concentration data. Separate sets of conversion factors were applied to EC datacollected by the TOR or TOT method.
The CFs were developed using existing source apportionment studies. Source apportionmentstudies for the West US included the Northern Front Range Air Quality Study (Denver,Colorado), the Los Angeles Study (various analyses based on data collected in 1982 at 4 urbanSouthern California sites), and the San Joaquin Valley Study. For the East US, information fromthe recent source apportionment study for sites in the South-East US was used. We contactedseveral experts and reviewed available literature to obtain information from these studies. Inparticular, James Schauer from University of Wisconsin-Madison provided very helpfulinformation. The conversion factors were developed by dividing the reported DPM concentrationby the reported total EC or OC concentration.
Since there are several source apportionment studies, each giving different estimates of the dieselcontribution at different receptors, and since there are several diesel exhaust source profiles inthe literature, several possible CF values could be applied for each site. For the TOR conversionfactors, we developed rural CFs for rural sites and urban CFs at urban sites. We could not obtainTOR data to match by region or season, since the available data were all collected in the winterand in the West US. For the TOT conversion factors we developed separate factors by quarter forthe East US and another set of factors for the West US. We could not obtain TOT data to matchby the urban or rural classification. The minimum, average, and maximum of the possible CFswill give the low-end, most likely, and high-end CF’s for that site.
Model to Monitor Comparisons
In Chapter 4, we will describe the DPM model-to-monitor comparisons. Using the CFs toconvert the monitored EC values to estimated DPM concentrations, we compared differencesbetween the monitored and modeled DPM values. For the modeled values, the NSATApredictions for 1996 using ASPEN (and CALPUFF, for the background) were used. Wecompared the monitored value to the NSATA prediction at the nearest ASPEN receptor site(census tract centroid). We also compared the monitored value to the maximum NSATAprediction within 30 km. For the monitored value we separately analyzed the site means ofDPM-high, DPM-low, and DPM-avg, as described above. The site means were computed byaveraging all the daily averages (from January 1, 1994 to December 31, 2001). There wereinsufficient data to restrict the monitored data to the modeling year 1996.
ICF Consulting 5
Separately for each model-to-monitor comparison, and for all locations, urban locations, andrural locations, we compared the modeled and monitored results using:
• Scatterplots of modeled against monitored values that include fitted regression lines• Tables summarizing the regression fits, average difference, and average percentage
difference for the modeled values against the monitored values• Tables summarizing the proportions of modeled values that were within 10%, 25%, 50%,
and 100% of the monitored values
Using the NSATA estimates of the percentage contributions of onroad and nonroad sources ateach matched receptor site, we also evaluated whether the regions dominated by either onroad ornonroad sources have better model-to-monitor comparisons. The same statistical comparisonswere applied to the subsets of sites dominated by non-road (at least 75 % of modeled DPM isnon-road) or on-road (at least 50 % of modeled DPM is on-road) emissions.
The inventory estimates from the NONROAD model have been revised since the NSATAassessment was conducted using the draft 2000 NONROAD model (also used to developinventories for EPA’s 2007 heavy duty standards). The EPA provided a single nationwidemultiplicative adjustment factor of 0.69 for the nonroad ambient DPM based on the ratio ofNONROAD year 1996 predictions from the 2002 and 2000 draft versions of the NONROADmodel . To determine if this 70 % adjustment lead to improved model-to-monitor comparisons,we applied the same statistical comparisons after adjusting the NSATA predictions using thenonroad adjustment factor (the onroad ambient DPM is unchanged).
Findings
The model-to-monitor comparisons for non-EPA TOR data (i.e. excluding the ECOCX estimatesof TOR from the EPA TOT data) were based on 15 monitoring sites. The model-to-monitorcomparisons for TOT data were based on 95 monitoring sites. The model-to-monitorcomparisons for TOR data including the EPA ECOCX values were based on 88 monitoring sites.
The regression model analyses were generally less useful because the R squared values were inmost cases less than 0.3 and the regressions tended to be over-influenced by the more extremevalues. Based on the regression results, the best model performance was for the DPM-minimummonitored value for TOT data but for the DPM-average value for TOR data. Results were verysimilar for the modeled values based on the 2000 and 2002 NONROAD model and were a littlebetter for the rural sites compared to the urban sites. For the Non-road-dominated subset of TOTsites, the regression model fitted better than the all sites regression, but the monitored valueswere significantly overpredicted. For the Non-road-dominated subset of TOR sites including theEPA ECOCX sites, the regression model fitted a bit worse than the all sites regression. Therewas not enough data to evaluate the On-road-dominated subset. The comparisons between themaximum modeled value within 30 km and the monitored values all showed that the monitoredvalue was significantly over-predicted.
A summary table of the differences between the nearest modeled values and the monitoredvalues is given on the next page. Based on the mean percentage difference and based on the
ICF Consulting 6
fraction of modeled values within 100 % of the monitored value, the best model performancewas consistently for the DPM-maximum value at the nearest census tract centroid using theestimates consistent with the 2002 NONROAD model. For the non-EPA TOR, for TOT, and forthe combination of TOR data from TOR sites and from EPA TOT converted to TOR (i.e., TORand ECOCX), the mean percentage differences were 26 %, 27 %, and –12 % and the fractions ofmodeled values within 100 % of the monitored value were 73 %, 80 %, and 92 %, respectively.These results compare favorably with the results of the model to monitor comparisons for otherpollutants in the NSATA assessment. For instance, ASPEN typically agrees with monitoring datawithin 30% half the time and within a factor of 2 most of the time. The best agreement is forbenzene where the results are within a factor of two for 89 percent of the cases and within 30%59 percent of the time. The median ratio of the benzene model to monitor comparisons was 0.92.Agreement for other HAPs varies, with median ratios of model to monitor values varyingbetween 0.65 for formaldehyde to 0.17 for lead.
We can conclude that the modeled diesel PM concentrations in NSATA agree reasonably wellwith monitor values, and the agreement is better than for other pollutants evaluated, except forbenzene.
ICF Consulting 7
Summary of differences between the nearest modeled concentration and the monitored values.
Fraction of ModeledValues Within
ModeledVariable1
MonitoredVariable2 N
MeanModeled
Value
MeanMonitored
Value
MeanDifference Mean
%Difference 10% 25% 50% 100%
concnear ECTOR 15 1.56 0.94 0.63 100 0.07 0.13 0.53 0.53concnear2 ECTOR 15 1.20 0.94 0.26 56 0.07 0.13 0.47 0.60concnear ECTORH 15 1.56 1.16 0.40 62 0.00 0.07 0.40 0.60concnear2 ECTORH 15 1.20 1.16 0.04 26 0.00 0.07 0.33 0.73concnear ECTORL 15 1.56 0.64 0.92 190 0.13 0.40 0.47 0.53concnear2 ECTORL 15 1.20 0.64 0.55 126 0.07 0.33 0.47 0.53concnear ECTOT 95 2.61 1.73 0.88 80 0.12 0.21 0.45 0.68concnear2 ECTOT 95 2.05 1.73 0.32 42 0.11 0.37 0.53 0.77concnear ECTOTH 95 2.61 2.10 0.52 61 0.11 0.22 0.46 0.74concnear2 ECTOTH 95 2.05 2.10 -0.05 27 0.11 0.35 0.53 0.80concnear ECTOTL 95 2.61 1.52 1.09 101 0.09 0.17 0.43 0.63concnear2 ECTOTL 95 2.05 1.52 0.52 58 0.09 0.32 0.52 0.72concnear TOR 88 2.31 1.70 0.61 47 0.10 0.30 0.59 0.78concnear2 TOR 88 1.81 1.70 0.11 15 0.17 0.30 0.59 0.85concnear TORH 88 2.31 2.23 0.08 13 0.11 0.26 0.60 0.84concnear2 TORH 88 1.81 2.23 -0.42 -12 0.08 0.22 0.52 0.92concnear TORL 88 2.31 1.19 1.12 110 0.10 0.26 0.41 0.65concnear2 TORL 88 1.81 1.19 0.62 65 0.14 0.31 0.52 0.74
Notes:1. Modeled variable:concnear Nearest modeled DPM concentration consistent with the draft 2000
NONROAD Modelconcnear2 Nearest modeled DPM concentration consistent with the draft 2002
NONROAD Model2. Monitored variable:ECTOR EC value multiplied by TOR average correction factor (missing for EC
measured using TOT).ECTORH EC value multiplied by TOR maximum correction factor (missing for EC
measured using TOT).ECTORL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOT).ECTOT EC value multiplied by TOT average correction factor (missing for EC
measured using TOR).ECTOTH EC value multiplied by TOT maximum correction factor (missing for EC
measured using TOR).ECTOTL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOR).TOR ECOCX value multiplied by TOT average correction factor for EPA data,
ECTOR for TOR data.TORH ECOCX value multiplied by TOT maximum correction factor for EPA
data, ECTOR for TOR data.TORL ECOCX value multiplied by TOR minimum correction factor for EPA
data, ECTOR for TOR data.
ICF Consulting 8
CHAPTER 2. DIESEL PARTICULATE MATTER EC/OC/BC DATABASE
EPA has been provided with three DbaseIII (DBF) files, comprising the EC/OC/BC(elemental/organic/black carbon) concentration database. The files were developed by compilingand processing data from the following studies (“source”):
• AIRBEAT• CASTNET• EPA (PM Speciation data)• IMPROVE (six selected sites )1
• IMS95• MATESII• NERL (Phoenix Supersite)• NFRAQS
All concentrations are reported as µg/m3.
The dailyavg.final.dbf file contains daily average concentrations by source, site_id, and date. Thevariables are listed in Table 2-1. Note that each daily average is possibly averaged acrossmultiple time periods during the day (e.g. some black carbon data was reported every fiveminutes) and/or multiple measuring instruments at the same location (e.g. EPA daily averagedata with multiple POCs on the same date). For MDLs, we either used values reported with thedatabase or used default values obtained from the literature. Raw values below the MDL werereplaced by one half of the MDL prior to computing the daily averages (this did not happen veryoften for the EC/OC/BC concentration data). For black carbon, MDLs were not reported in thedatabases and could not be obtained from the data suppliers. According to the “AethalometerBook” (Hansen, 2000), written by the company that makes the aethalometer instrument thatmeasures BC, the black carbon MDL depends upon the filter size, air flow rate, and averagingperiod. The filter size and air flow rate were not always available. Furthermore, since we areonly interested in daily averages, the MDLs of the five-minute or hourly values do not representthe precision of the daily averages. For these reasons, we did not use MDLs for the black carbondata, effectively assuming an MDL of zero. The method variable lists all methods used for thatday (“Aethalometer” applies to all black carbon data, other methods apply to EC and OC).
EPA’s Office of Air Quality Planning and Standards is reviewing the measurement of ECthrough the Speciation Trends Network, and an Agency statement on the issue is forthcoming.For now, however, existing values developed using the TOT method are being used.
Minimum detection limits (MDLs) were obtained from the data suppliers if possible. For theEPA PM speciation data, specific MDL’s for each of the various NIOSH methods were supplied;each daily measurement had an associated measurement method. The EPA MDLs were mostlyequal to or close to 0.146 µg/m3 for EC and OC. For Airbeat, the data supplier reported ECMDLs that were either 0.059 or 0.134 µg/m3 depending on the method used. For IMPROVE and
1 The data from the two Yellowstone Park sites YELL1 and YELL2 were treated as all coming from the YELL1 site.The Yellowstone Park monitoring site was moved a short distance in 1996.
ICF Consulting 9
NFRAQS, the data source contact was unable to give specific MDL values for the TOR method,because the TOR measurements of EC and OC are both sums of three components, each ofwhich has its own MDL. However, we were able to find a report by Chow and Watson (1998)that tabulated MDLs of 0.12 µg/m3 for EC and OC by TOR. We used these values for theIMPROVE and NFRAQS EC and OC data. For CASTNET, MATESII, and NERL, we wereunable to obtain specific MDL values for the NIOSH measurements, but Gary Lear (EPA)suggested a typical MDL value of 0.1 µg/m3 for EC and OC, which was used for these threestudies. For MATESII, the EC MDL was erroneously entered as 0.01 in the database; this has noaffect on the daily averages or other results since all the MATESII EC measurements were 0.47µg/m3 or greater.
Since some site-days had more than one measured EC or OC concentration, a daily average canbe treated as being below the MDL if any of the measurements for that day and carbon specieswere below the MDL. Using this definition, across the entire dataset, 9.9 % of the 13,993 ECdaily averages were below the MDL and 3.3 % of the 13,804 OC daily averages were below theMDL.
In addition to the carbon species EC, OC, and BC, the daily average file also includes dailyaverage concentrations for the various estimated DPM concentrations defined as follows:
ECOCX The EPA EC and OC data were collected using the NIOSH (SunsetLaboratory) method of thermal optical transmission (TOT). EPA alsocomputed a value ECOCX intended to approximate the equivalent ECconcentration based on thermal optical reflectance (TOR, as developedand applied by Desert Research Institute). This value is missing for non-EPA data.
ECTOR EC value multiplied by TOR average correction factor (missing for ECmeasured using TOT).
ECTORH EC value multiplied by TOR maximum correction factor (missing for ECmeasured using TOT).
ECTORL EC value multiplied by TOR minimum correction factor (missing for ECmeasured using TOT).
ECTOT EC value multiplied by TOT average correction factor (missing for ECmeasured using TOR).
ECTOTH EC value multiplied by TOT maximum correction factor (missing for ECmeasured using TOR).
ECTOTL EC value multiplied by TOR minimum correction factor (missing for ECmeasured using TOR).
ICF Consulting 10
EPATOR ECOCX value multiplied by TOR average correction factor (missing fornon-EPA data).
EPATORH ECOCX value multiplied by TOR maximum correction factor (missing fornon-EPA data).
EPATORL ECOCX value multiplied by TOR minimum correction factor (missing fornon-EPA data).
In the concentration summary file, EC, OC, BC, and the above DPM estimates are all referred toas “SPECIES”.
The TOR and TOT (NIOSH) minimum, maximum, and average conversion factors are reportedin Table 3-2 of Chapter 3 “Diesel Particulate Matter Conversion Factors.” The converted valuesestimate the diesel particulate matter (DPM) concentration. The applicable conversion factorsdepend upon the measurement method (TOR or TOT), whether the location is in the East orWest US, whether the site is urban or rural, and the calendar quarter. For this calculation, siteswith (signed) longitude less than −92° were treated as being in the West US. The Mississippiriver roughly lies along the −92° longitude line. Sites were defined as urban or rural based on theNSATA assignment for the nearest census tract centroid; this assignment is given by the variableurbannear in the site summary file.
The sitesummary.final.dbf file contains summary information about the individual sites(identified by the source and site_id variables). The variables are listed in Table 2-2. Informationincludes: the city or county; state; latitude and longitude; first and last measurement dates forEC, BC, or OC (within the time frame starting January 1, 1994); method; ratios of EC, OC,sulfate and nitrate to PM2.5; location, distance, location type (urban or rural) and modeledconcentration for the nearest modeled diesel PM2.5 concentration from NATA; location andmodeled concentration for the maximum modeled diesel PM2.5 concentration from NATAwithin 30 km, if any; and the dominant source. Note that the maximum modeled value within 30km is missing if there are no census tract centroids within 30 km. The method variable lists allmethods used for that site (“Aethalometer” applies to all black carbon data, other methods applyto EC and OC).
The first set of analyses used the 1996 NSATA model predictions consistent with the year 2000draft of the NONROAD model. The second set of analyses used 1996 NSATA model predictionsconsistent with the year 2002 draft of the NONROAD model: The earlier model’s ambient andbackground non-road components were both multiplied by 0.69 for every census tract. (since the1996 national modeled NONROAD DPM was reduced by 31 % for the 2002 draft model). Thevalues of the nearest modeled concentration and of the location and modeled value for themaximum modeled concentration within 30 km are each given separately for each version of theNONROAD model. The dominant source is defined for the 2000 NONROAD model only. If thetotal on-road modeled DPM is 50 % or greater of the total modeled DPM, the dominant source is“On-road.” If the total on-road modeled DPM is less than 25 % of the total modeled DPM, thedominant source is “Non-road.” Otherwise there is no dominant source.
ICF Consulting 11
The concsummary.final.dbf file contains summary statistics for the daily average concentrationdata by source and site, species (BC, EC, OC, ECOCX, ECTOR, ECTORL, ECTORH, ECTOT,ECTOTH, ECTOTL, EPATOR, EPATORL, EPATORH), year, calendar quarter, andweekday/weekend. The variables are listed in Table 2-3. Possible values of year are 1994, 1995,1996, 1997, 1998, 1999, 2000, 2001, and “All” (all years combined). Possible values of quarterare 1, 2, 3, 4, and “All” (all quarters, i.e., the entire year or years). Possible values of dayofweekare “Weekday” (Monday to Friday), “Weekend” (Saturday or Sunday), and “All.” For example,overall averages for a site are obtained by considering year = quarter = dayofweek = “All.”Summary statistics are available by year (including “All”) and/or by quarter (including “All”).For the weekend/weekday split, separate weekend and weekday summary statistics are reportedby year (including “All”) or by quarter (including “All”) but not for specific year and quartercombinations.
ICF Consulting 12
Table 2-1. Daily Average File (dailyavg.dbf)Variable DescriptionDate DateSource Data source (study)Site_id Site identifierCitycounty City or CountyState StateEC Elemental carbon daily averageOC Organic carbon daily averageBC Black carbon daily averageECOCX Estimated EC by TOR (EPA data)ECTOR Average estimated DPM (TOR data)ECTORH Maximum estimated DPM (TOR data)ECTORL Minimum estimated DPM (TOR data)ECTOT Average estimated DPM (TOT data)ECTOTH Maximum estimated DPM (TOT data)ECTOTL Minimum estimated DPM (TOT data)EPATOR Average estimated DPM (EPA ECOCX data)EPATORH Maximum estimated DPM (EPA ECOCX data)EPATORL Minimum estimated DPM (EPA ECOCX data)MinECMDL Minimum EC MDL for dateMaxECMDL Maximum EC MDL for dateMinOCMDL Minimum OC MDL for dateMaxOCMDL Maximum OC MDL for dateSulfate Sulfate daily averageNitrate Nitrate daily averagePM25 PM2.5 daily averageLatitude Latitude (degrees and fractions of a degree)Longitude Longitude (degrees and fractions of a degree)Method List of all measurement methods used on date,
separated by semicolons.
ICF Consulting 13
Table 2-2. Site Summary File (sitesummary.dbf)Variable DescriptionSource Data source (study)
Site_id Site identifierCitycounty City or CountyState StateLatitude Latitude (degrees and fractions of a
degree)Longitude Longitude (degrees and fractions of
a degree)MinECDate First date with non-missing ECMaxECDate Last date with non-missing ECMinBCDate First date with non-missing BCMaxBCDate Last date with non-missing BECMinOCDate First date with non-missing OCMaxOCDate Last date with non-missing OCMinECMDL Minimum EC MDL for siteMaxECMDL Maximum EC MDL for siteMinOCMDL Minimum OC MDL for siteMaxOCMDL Maximum OC MDL for siteEC_PM25 Mean EC divided by mean PM2.5
(for days when both were reported)OC_PM25 Mean OC divided by mean PM2.5
(for days when both were reported)Sulf_PM25 Mean sulfate divided by mean
PM2.5 (for days when both werereported)
Nitr_PM25 Mean nitrate divided by mean PM2.5
(for days when both were reported)Method List of all measurement methods
used at site, separated bysemicolons
Fipsmax FIPS code for census tract centroidwith maximum modeled DPMwithin 30 km (2000 NONROADmodel)
Tractmax Tract ID code for census tractcentroid with maximum modeledDPM within 30 km (2000NONROAD model)
Dist Distance (km) to nearest censustract centroid
ICF Consulting 14
Table 2-2. Site Summary File (sitesummary.dbf)Variable DescriptionMaxconc Maximum modeled DPM within 30
km (2000 NONROAD model)Concnear Modeled DPM at nearest census
tract centroid (2000 NONROADmodel)
Fipsnear FIPS code at nearest census tractcentroid
Tractnear Tract ID code at nearest census tractcentroid
Urbannear NSATA location type (U = urban,R = rural) at nearest census tractcentroid
Fipsmax2 FIPS code for census tract centroidwith maximum modeled DPMwithin 30 km (2002 NONROADmodel)
Tractmax2 Tract ID code for census tractcentroid with maximum modeledDPM within 30 km (2002NONROAD model)
Maxconc2 Maximum modeled DPM within 30km(2002 NONROAD model)
Concnear2 Modeled DPM at nearest censustract centroid (2002 NONROADmodel)
Dominant Dominant source: “Non-road,”“On-Road,” or “ ” (blank).Consistent with the 2000NONROAD model.
ICF Consulting 15
Table 2-3. Concentration Summary File (concsummary.dbf)Variable DescriptionSource Data source (study)
Site_id Site identifierYear Calendar year or “All”Quarter Calendar quarter or “All”DayofWeek “Weekday,” “Weekend,” or “All”Species EC, OC, BC, ECOCX, ECTOR,
ECTORL, ECTORH, ECTOT,ECTOTH, ECTOTL, EPATOR,EPATORL, or EPATORH
N Number of daysMean Arithmetic meanMedian MedianGeommean Geometric meanStddev Standard deviationMinimum MinimumMaximum MaximumPerc10 10th percentilePerc20 20th percentilePerc30 30th percentilePerc40 40th percentilePerc50 50th percentilePerc60 60th percentilePerc70 70th percentilePerc80 80th percentilePerc90 90th percentile
ICF Consulting 16
CHAPTER 3. DIESEL PARTICULATE MATTER CONVERSION FACTORS
This part of the effort was primarily carried out by the consulting firm dKC. We compiled“conversion factors” (CFs) for estimating ambient diesel particulate matter (DPM) based onelemental carbon (EC), organic carbon and PM2.5 measurements. We attempted to collectinformation on CFs for black carbon (BC), but found none in the literature. See below for adiscussion of the recommended treatment of the black carbon data. The CFs were collected fromexisting source apportionment and source-receptor model studies.
We received assistance from the following researchers to identify data sources: James Schauer(University of Wisconsin), Philip Hopke (Clarkston University), Alan Gertler (Desert ResearchInstitute), and Steve Cadle (GM Research). Overall, we identified and compiled data on CFsfrom the following sources:
1. Zheng, Cass, Schauer et al (2002).Apportionments of PM2.5 (mass) and organic carbon in PM2.5 for 8 SE US sites:4 urban, 3 rural and 1 suburban. Season: All 4 individually. Dr. Schauer alsoprovided an elemental carbon breakdown by season (but not by site).
2. Ramadan, Song, and Hopke (2000).Apportionments of PM (mass) in Phoenix, AZ. Season: Annual Average.
3. Schauer et al (1996).Apportionments of primary fine organic aerosol and fine particulate massconcentrations for 4 urban sites in Southern California. Season: Annual Average.
4. Schauer and Cass (2000).Apportionments of primary fine organic aerosol and fine particulate massconcentrations for 3 sites in the Central Valley of California: 2 urban and 1 rural.Season: Winter
5. Watson, Fujita, Chow, Zielinska et al (1998).NFRAQS. This was the most comprehensive and current analysis of sources ofambient PM. Two techniques were used to apportion ambient PM: ConventionalCMB and Extended Species CMB. Extended Species CMB breaks down gasolinevehicle emissions into 3 categories: cold start, hot transient, and high PM emitter(e.g. a vehicle with visible smoke). PM was apportioned into total carbon, organiccarbon, elemental carbon and PM2.5. A total of 9 sites were evaluated: 3 urban,4 rural, one suburban, and one to characterize regional transport. Two sites wereused for the Extended Species CMB analysis: one rural and one urban. For theExtended Species CMB analysis, a temporal apportionment was done. Season:Winter
6. Air Improvement Resources (1997).Summary and analysis of available data on contribution of gasoline poweredvehicles to ambient levels of fine particulate matter. Most of the data was covered
ICF Consulting 17
in Reference #3. Included projections of sources of fine carbonaceous PM. for 4Southern California sites. Season: Annual Average.
7. Cass (1997).Summary and analysis of available data on contribution of motor vehicles toambient levels of fine particulate matter. Most of the data was covered inReference #3. Included projections of sources of elemental carbon for 3 SouthernCalifornia sites. Season: Annual Average.
The numbering of these source apportionment study references is arbitrary but is retained herefor consistency with the attached Excel spreadsheet.
Dr. Schauer noted that several important source apportionment studies are currently underway,and results should start being available in the next 6 months.
We compiled data on conversion factors (CFs) into an Excel spreadsheet provided to EPA. Thespreadsheet contains a page for CFs from each reference. The following information wascompiled for each data point:
• Year data were collected
• Site evaluated
• State data were collected
• Type of site: urban, rural, suburban (in some cases more specific types were used,e.g. rural down valley).
• Season data were collected
• Ambient Measurement Technique: Thermal optical transmission (TOT, alsoreferred to as the NIOSH method, as developed by Sunset laboratories) or thermaloptical reflectance (TOR, as developed and applied by Desert Research Institute,which was also used for the IMPROVE database and the Northern Front Rangestudy).
• PM measurement parameter (organic carbon, elemental carbon, PM2.5)
• Concentration apportioned to diesel powered engines
• Concentration apportioned to gasoline engine exhaust. (One of the data sources,Northern Front Range Air Quality Study, had a breakdown of gasoline poweredvehicle emissions into LDHV Cold Start, LDGV hot stabilized (warmed-up vehicleemissions, and LDGV high PM emitter)
• Total concentration
ICF Consulting 18
• % of PM from diesel, for each measurement parameter This value was calculatedfor each site (in some cases, by season) as the ratio of the CMB estimated dieselfine particulate matter to the site average total fine particulate matterconcentration. Both values were either reported in the source apportionment studyreport or were obtained directly from the researcher.
• Multiplicative conversion factor to convert total organic carbon (OC) or totalelemental carbon (EC) concentration to diesel PM2.5 concentration. Conversionfactors (CFs) were calculated by dividing the diesel PM2.5 concentration reportedby the study by the total organic carbon or elemental carbon concentrationreported by the study.
• Z factor: % of diesel PM2.5 that is OC or EC. This was calculated by dividing thereported % of OC or EC that is from diesels by the CF calculated above. The % ofdiesel PM2.5 that is OC or EC varied significantly for data based on the twomeasurement techniques: TOR or NIOSH.
The Z factor equals the OC or EC diesel PM2.5 concentration divided by the total diesel PM2.5
concentration. This factor can be compared with the measured OC or EC fraction in theassociated diesel PM2.5 source profile. For most of the studies, the diesel source profiles were noteasily obtained. For reference 5 (NFRAQS), the calculated Z factor for EC is compared with thesource profile Z factor in the attached spreadsheet. The calculated and source profile Z factorswere quite close except for the Chatfield and Highlands sites. For consistency with the treatmentof other studies, for all the NFRAQS sites (including Chatfield and Highlands) we used themultiplicative conversion factor defined above and did not correct for any differences betweenthe calculated and source profile Z factors.
The originally proposed approach for this project was to compute the conversion factors as thepercentages of diesel in PM2.5 OC or EC (from CMB) divided by the source profile percentage ofOC or EC in diesel PM2.5 (i.e., divided by the source profile Z factor). This alternative approachgives almost the same results, as can be shown in the reference 5 worksheets, which demonstratethat the two methods give almost the same conversion factors for the NFRAQS sites, except forthe Chatfield and Highlands sites. The method used here does not require the CMB study toprovide a source apportionment of OC or EC (just the source apportionment for PM2.5) and doesnot need the diesel source profile.
The spreadsheet also contains sheets that compile available data on the following:
• Organic Carbon (OC) conversion factors (conversion factors to convert total OCto diesel PM2.5 concentration).
• Elemental Carbon (EC) conversion factors (conversion factors to convert total ECto diesel PM2.5 concentration).
• Fraction of fine particulate mass attributed to diesels.
ICF Consulting 19
Table 3-1 presents the minimum, maximum, and average diesel fraction of PM2.5 as a functionof:
• Urban or rural• Season• East or West US
The reported minimum, maximum, and average values in Table 3-1 are the minima, maxima, andarithmetic means of the “% of PM from diesel” values across all sites (and seasons, whereapplicable) in the given site subset.
Table 3-2 presents the minimum, maximum, and average EC conversion factors as a function of:
• Measurement technique• East or West US• Season• Urban or rural
The reported minimum, maximum, and average values in Table 3-2 are the minima, maxima, andarithmetic means of the EC conversion factors across all sites (and seasons, where applicable) inthe given site subset. For the NIOSH (same as TOT) data collected in the East, the minimum,maximum, and average conversion factors are all equal. This is because these values were basedonly on the Zheng, Cass, Schauer, et al (2002) study. For this project, Dr, Schauer provided ECsummary data from this study averaged over sites, by season. Hence only one value is availablefor NIOSH data for each season in the Eastern US.
Table 3-3 presents the minimum, maximum, and average OC conversion factors as a function of:
• Urban or rural• Season• East or West US
The reported minimum, maximum, and average values in Table 3-3 are the minima, maxima, andarithmetic means of the OC conversion factors across all sites (and seasons, where applicable) inthe given site subset.
Black Carbon
Black carbon is measured on an “aethalometer,” a measuring instrument developed by MageeScientific. The following summary is taken from the “Aethalometer Book,” by Hansen (2000).
“The Aethalometer is an instrument that provides a real-time readout of the concentrationof ‘Black’ or ‘Elemental’ carbon aerosol particles. (BC or EC). These particles (“soot”)are emitted from all types of combustion, most notably from diesel exhaust. ‘BC’ isdefined by blackness, an optical measurement. The Aethalometer uses an optical
ICF Consulting 20
measurement, and gives a continuous readout. The ‘EC’ definition is more common. It isbased on a thermal-chemical measurement, an analysis of material collected on a filtersample for several hours and then sent to a laboratory. Research at Harvard showed thatthe Aethalometer BC measurement is directly related and equivalent to the filter-basedEC measurement. In fact, an option in the software allows it to read out in EC units.”
More details are given in the full document (Hansen, 2000) and in various references, includingAllen et al (1999), Chow et al (1993), Hansen and Mc Murry (1990), and Liousse et al (1991).
On this basis, and because none of the source apportionment studies that we found used blackcarbon measurements, we recommend using the same conversion factors to convert BC and ECconcentrations to diesel PM2.5.
Recommendations
The final columns in Tables 3-2 and 3-3 give our recommendations for which minimum,maximum, and average EC and OC conversion factors should be applied to the database. ForBC, the available data are more limited and we did not find any source apportionment studiesbased on BC measurements. OC is not as useful a surrogate for diesel PM as EC because dieselPM source profiles tend to contain much more EC than OC, on average, and because the dieselfraction in EC is typically estimated to be much higher than the diesel fraction in OC. Forexample, the NFRAQS study (Watson, Fujita, Chow, Zielinska, et al, 1998), determined that ECcontains about 60 % diesel and OC contains about 8 % diesel. Therefore, EPA determined thatonly EC data be used for the model to monitor comparisons.
For EC, as shown in Table 3-2, available CF data based on the NIOSH (TOT) method in the Eastmainly allows a breakdown by season. There is not enough seasonal data to stratify by locationtype. The seasonal stratification in the East is based only on reference 1, which had data forJanuary, April, July, August, and October only. Thus for the data in the East, the seasonalstratification is equivalent to a quarterly stratification: Winter = Quarter 1, Spring = Quarter 2,Summer = Quarter 3, Fall = Quarter 4. For the East US, we recommend using the EC conversionfactors for each daily mean according to the calendar quarter (equivalently, the season). For theWest US, the available data were collected at urban sites in Los Angeles and the season was notreported. Thus we suggest applying the same factors for all EC data collected in the West,regardless of location type. For observations based on the TOR method, we suggest thatconversion factors be based on location type (urban or rural), regardless of season. This is due toan absence of TOR data from non-Winter observations. For BC, the approximate equivalencebetween EC and BC suggests using the same conversion factors as for EC.
For OC, separate conversion factors for TOR and TOT data were not computed, although theywould be preferred due to the wide differences in the two measurement methods. For OC, asshown in Table 3-3, the data in the East is stratified by Urban or Rural location and by season,but the data in the West is only available for the winter season. The seasonal stratification in theEast is based only on reference 1, which had data for January, April, July, August, and Octoberonly. Thus for the data in the East, the seasonal stratification is equivalent to a quarterlystratification: Winter = Quarter 1, Spring = Quarter 2, Summer = Quarter 3, Fall = Quarter 4. For
ICF Consulting 21
the East US, we recommend using the OC conversion factors for each daily mean according tothe calendar quarter (equivalently, the season), and whether the site is urban or rural. For theWest US at rural sites, only winter data are available for OC conversion factors, but at rural sitesin the winter, the CF distributions for East and West are quite similar (the factors in the West area little lower). On this basis, assuming the same applies to all seasons, we recommend applyingthe OC CF’s for the seasonal/quarterly totals to the daily averages at West US rural sites in thefirst three quarters. For West US rural sites in the winter, i.e., Quarter 1, we recommend usingthe corresponding OC CF distribution. For urban sites, the winter CF distributions are verydifferent between the East and West sites—the averages differ by a factor of about two—so thisapproach is not recommended. Instead, for West US urban sites, we recommend using the“Urban All” OC CF distribution since the uncertainty range is conservatively wide, the mean isclose to the mean for West US urban sites, and the minimum is the same as the minimum forWest US urban sites in the winter.
ICF Consulting 22
Table 3-1Summary of Percent of Fine PM Apportioned to Diesels
% Contribution From DieselsLocationtype
Season EastorWest Minimum* Maximum* Average*
Fall East 8.9% 10.9% 9.8%Spring East 11.4% 15.2% 12.9%Summer East 7.5% 10.9% 8.7%Winter East 10.0% 13.5% 12.1%
West 2.7% 11.4% 5.9%
Rural
Winter Total 2.7% 13.5% 7.8%RuralTotal 2.7% 15.2% 9.1%
All West 12.7% 35.7% 21.2%Fall East 7.5% 32.0% 20.7%Spring East 10.1% 29.9% 19.8%Summer East 9.6% 25.5% 14.8%Winter East 17.0% 24.1% 21.2%
West 5.3% 12.7% 9.4%
Urban
Winter Total 5.3% 24.1% 13.3%UrbanTotal 5.3% 35.7% 16.6%GrandTotal 2.7% 35.7% 13.9%
Notes:* Minimum, maximum, or average value across all sites of the % contribution from diesel, whichis defined as the ratio of the CMB estimate of diesel PM2.5 divided by the total PM2.5
ICF
Con
sult
ing
23
Tab
le 3
-2Su
mm
ary
of C
alcu
late
d E
lem
enta
l Car
bon
(EC
) Con
vers
ion
Fac
tors
(Con
vers
ion
fact
ors
to c
onve
rt t
otal
EC
to
dies
el P
M2.
5 co
ncen
trat
ion)
Rec
omm
ende
dC
onve
rsio
nF
acto
rs
Am
bien
tM
easu
rem
ent
Tec
hniq
ue:T
OR
or N
IOSH
Eas
t or
Wes
tSe
ason
Loc
atio
nT
ype
Gen
eral
MIN
*M
AX
*A
VE
RA
GE
*E
AST
WE
ST
Eas
tFa
ll (Q
4)M
ixed
2.3
2.3
2.3
XE
ast
Spri
ng (
Q2)
Mix
ed2.
42.
42.
4X
Eas
tSu
mm
er(Q
3)M
ixed
2.1
2.1
2.1
X
Eas
tW
inte
r (Q
1)M
ixed
2.2
2.2
2.2
X
NIO
SH
Wes
tU
nkno
wn
Urb
an1.
22.
41.
6X
NIO
SH T
otal
1.2
2.4
2.0
Win
ter
Rur
al0.
61.
00.
8X
XU
rban
0.5
1.0
0.7
XX
TO
R
Win
ter
Tot
al0.
51.
00.
8T
OR
Tot
al0.
51.
00.
8G
rand
Tot
al0.
52.
41.
3N
otes
:*
Min
imum
, max
imum
, or
aver
age
valu
e ac
ross
all
site
s of
the
estim
ated
con
vers
ion
fact
ors.
ICF
Con
sult
ing
24
Tab
le 3
Sum
mar
y of
Cal
cula
ted
OC
Con
vers
ion
Fac
tors
(Con
vers
ion
fact
ors
to c
onve
rt t
otal
OC
to
dies
el P
M2.
5 co
ncen
trat
ion)
Cal
cula
ted
OC
Con
vers
ion
Fac
tor
Rec
omm
ende
dC
onve
rsio
n F
acto
rs L
ocat
ion
Typ
eG
ener
al
Seas
onE
ast
or Wes
tM
inim
um*
Max
imum
*M
axim
um*
Eas
tW
est
Fall
Eas
t0.
50.
50.
5X
Fall
Tot
al0.
50.
50.
5X
Spri
ngE
ast
0.5
1.3
0.8
XSp
ring
Tot
al0.
51.
30.
8X
Sum
mer
Eas
t0.
51.
91.
4X
Sum
mer
Tot
al0.
51.
91.
4X
Win
ter
Eas
t0.
40.
50.
5X
Wes
t0.
20.
50.
4X
Rur
al
Win
ter
Tot
al0.
20.
50.
4R
ural
All
0.2
1.9
0.6
All
Wes
t0.
61.
30.
9A
ll T
otal
0.6
1.3
0.9
Fall
Eas
t0.
91.
31.
0X
Fall
Tot
al0.
91.
31.
0Sp
ring
Eas
t0.
92.
01.
3X
Spri
ng T
otal
0.9
2.0
1.3
Sum
mer
Eas
t1.
01.
91.
5X
Sum
mer
Tot
al1.
01.
91.
5W
inte
rE
ast
0.4
0.8
0.6
XW
est
0.2
0.5
0.4
Urb
an
Win
ter
Tot
al0.
20.
80.
5U
rban
All
0.2
2.0
0.9
XA
ll0.
22.
00.
8N
otes
:*
Min
imum
, max
imum
, or
aver
age
valu
e ac
ross
all
site
s of
the
estim
ated
con
vers
ion
fact
ors.
ICF Consulting 25
CHAPTER 4. MODEL-TO-MONITOR COMPARISONS
Using the CFs to convert the monitored EC values to estimated DPM concentrations, wecompared differences between the monitored and modeled DPM values. For the modeled values,the NSATA predictions for 1996 using ASPEN (and CALPUFF, for the background) were used.We compared the monitored value to the NSATA prediction at the nearest ASPEN receptor site(census tract centroid). These comparisons were made for the original NSATA model predictionsconsistent with the draft 2000 NONROAD model (“concnear”) and for the revised NSATAmodel predictions consistent with the draft 2002 NONROAD model (“concnear2”) with non-road model predictions reduced to 69 % of their original value. We also compared the monitoredvalue to the maximum NSATA prediction within 30 km. These comparisons were also made forthe original NSATA model predictions consistent with the draft 2000 NONROAD model(“maxconc”) and for the revised NSATA model predictions consistent with the draft 2002NONROAD model (“maxconc2”)
For the monitored value we separately analyzed the site means of DPM-high, DPM-low, andDPM-avg, as described above. The site means were computed by averaging all the dailyaverages (from January 1, 1994 onward), since there were insufficient data to restrict themonitored data to the modeling year 1996. The first set of comparisons used EC data collectedby the TOR method only (average, minimum, and maximum DPM values ECTOR, ECTORL,and ECTORH respectively). The second set of comparisons used EC data collected by the TOT(NIOSH) method only (average, minimum, and maximum DPM values ECTOT, ECTOTL, andECTOTH respectively). The third set of comparisons combined the EC data collected using theTOR method with the EPA ECOCX value (based on estimated TOR values). This set of average,minimum, and maximum DPM values are denoted by TOR, TORL, and TORH, respectively.
Each of these model-to-monitor comparisons were applied to the subsets of all locations, urbanlocations, and rural locations. Additionally, but only for the predictions consistent with the draft2000 NONROAD model, we considered the subsets of sites with modeled DPM dominated byNon-road (at least 75 % non-road) or by On-road (at least 50 % on-road).
We compared the modeled and monitored results using:
• Scatterplots of modeled against monitored values that include fitted regression lines• Tables summarizing the regression fits, average difference, and average percentage
difference for the modeled values against the monitored values• Tables summarizing the proportions of modeled values that were within 10%, 25%, 50%,
and 100% of the monitored values
Table 4-1 summarizes each of the regressions. The modeled values are regressed against theaverage, maximum, and minimum “DPM monitored” values for each given data subset. The“DPM monitored” value is the EC value multiplied by the applicable conversion factor toconvert it to the estimated DPM. Scatterplots are shown for a few representative cases. In eachscatterplot, the fitted regression lines of the modeled values against the DPM monitored values.Each plot shows the modeled values plotted against the average, maximum, and minimum DPMmonitored values for a given data subset. The three regression lines are shown together with the
ICF Consulting 26
Y=X line. The Y=X line has zero intercept and slope 1 and represents the ideal case wheremodeled and monitored values agree precisely. For easier comparison, the regressions have beennumbered in Table 4-1, and those numbers are included in the Figure footnotes.
In some cases, all three regression lines intersect and have the same intercept (but differentslopes). This occurs in cases where the same set of three conversion factor values apply at eachmonitored value in the data subset. In some cases there are some data points where all threeDPM monitored values are identical, or almost identical. This is attributable to the fact that forNIOSH data in the East, the minimum, maximum, and average estimated CFs were identical foreach season, and the seasonal values were almost identical, as shown in Table 3-2.
Table 4-1 presents the regression lines in tabular format. For each case is presented: theintercept, its standard error, the slope, its standard error, and the R squared goodness-of-fitstatistic. Ideally the intercept should be close to zero and the slope should be close to 1. Rsquared values close to 1 indicate a good fit of the simple linear model and values close to zeroindicate a poor fit.
Tables 4-2 to 4-4, respectively, show the monitored DPM values and nearest modeled value(consistent with the draft 2000 NONROAD model) for the TOR sites, for the TOT sites, and forthe combination of the TOR sites and the EPA sites with the ECOCX data. These data were usedfor the “concnear” regressions (except that the TOR outlier value mentioned below was excludedfrom the regressions).
Table 4-5 presents the mean modeled and monitored values, the mean difference, the meanpercentage difference, and the fractions of modeled values within 10 %, 25 %, 50 %, or 100 % ofthe monitored values. Ideally the mean differences and mean percentage differences should beclose to zero and the fractions of modeled values within a small percentage of the modeledvalues should be close to one.
Review of preliminary scatterplots showed clearly that for the TOR data, there was one site withan extremely high modeled value that was greater than 12 ug/m3 compared to monitored DPMvalues close to 1 ug/m3. Other TOR modeled to monitor ratios were much lower. This obviousoutlier value from the IMPROVE Washington DC site (WASH1) was removed from all theseanalyses although it was retained in the database.
Results
TOR Regressions excluding EPA data
The TOR comparisons excluding the EPA ECOCX data are based on only 15 monitoring sitesand therefore are relatively less representative. As shown in Figure 4-1 (regressions 1, 19, and37), for all sites combined, the regression line of the nearest modeled concentration against theDPM average is closest to the Y=X line but, as shown in Table 1, regression 1, the R squaredvalue is 0.22, indicating a poor regression fit. The model tends to overpredict as shown by thefact that more of the values are above the Y=X line. The same analysis for rural sites only(regressions 2. 20, and 38) shows a better fit for the regression but a greater tendency to
ICF Consulting 27
overpredict, with slopes close to 2. There are too few urban TOR locations to properly evaluatethis subset. There are also too few Non-road dominated locations (and zero on-road dominatedlocations) to compare the modeled and monitored data. Comparing Figure 4-2 to Figure 4-1, theregression model using the revised NONROAD model appears to fit slightly worse but thenumerical differences are relatively small for most locations. The comparisons between themaximum modeled value within 30 km and the monitored values show that the monitored valueis significantly over-predicted (e.g., see Figure 4-3).
TOT Regressions
The TOT comparisons are primarily based on the EPA PM speciation data producing a total of95 monitored values for each of the DPM-minimum, DPM-maximum, and DPM-average. Figure4 uses all these data. The best results are for the DPM-minimum case, with a R squared value of0.17, an intercept of 1.52 and a slope of 0.72. Figure 4-4 shows that there are some monitoredvalues that are significantly over-predicted and some monitored values that are significantlyunder-predicted. The results for the urban and rural subsets are similar although the modelperformance appears to be a little better for the rural sites. For the Non-road dominated sites, theregression model fits better than the all sites regression, but the monitored values aresignificantly overpredicted (the slope is 1.82 for the DPM-average, regression 58). There are notenough data points to properly evaluate the subset of On-road dominated sites. ComparingFigure 4-4 to Figure 4-5, the regression model using the revised NONROAD model appears to fitslightly better but the numerical differences are relatively small for most locations; the DPM-minimum predictions give the best model performance. Similarly to TOR, the comparisonsbetween the maximum modeled value within 30 km and the monitored values show that themonitored value is significantly over-predicted.
TOR Regressions including EPA data
The TOR comparisons including the EPA ECOCX data are based on 88 monitoring sites. Of thethree monitored DPM values, the best model performance is obtained for the DPM-averagevalue, regression 127, with a slope of 0.91 and an intercept of 0.77 although the R squared valueis only 0.12. (Figure 4-6). The regression results for the rural and urban subsets are very similar,but the regression fit is a little better for the rural subset. For the 18 Non-road dominated sites,the model performance is a bit worse than the performance for all sites. There are insufficientlymany sites to evaluate the subset of on-road dominated sites. The numerical differences betweenthe 2000 and 2002 models are relatively small for most locations and the model performance isvery similar. (Using the revised model the R squared values are slightly higher, the intercept iscloser to zero, but the slope is further from 1). The comparisons between the maximum modeledvalue within 30 km and the monitored values show that the monitored value is significantly over-predicted.
Differences and percentage differences
The following table extracted from Table 4-5 summarizes the differences and percentagedifferences between the nearest modeled value and the monitored values. The ECTOR outliervalue for the IMPROVE Washington DC site is excluded from these calculations.
ICF Consulting 28
Summary of differences between the nearest modeled concentration and the monitored values.
Fraction of ModeledValues Within
ModeledVariable1
MonitoredVariable2 N
MeanModeled
Value
MeanMonitored
Value
MeanDifference Mean
%Difference 10% 25% 50% 100%
concnear ECTOR 15 1.56 0.94 0.63 100 0.07 0.13 0.53 0.53concnear2 ECTOR 15 1.20 0.94 0.26 56 0.07 0.13 0.47 0.60concnear ECTORH 15 1.56 1.16 0.40 62 0.00 0.07 0.40 0.60concnear2 ECTORH 15 1.20 1.16 0.04 26 0.00 0.07 0.33 0.73concnear ECTORL 15 1.56 0.64 0.92 190 0.13 0.40 0.47 0.53concnear2 ECTORL 15 1.20 0.64 0.55 126 0.07 0.33 0.47 0.53concnear ECTOT 95 2.61 1.73 0.88 80 0.12 0.21 0.45 0.68concnear2 ECTOT 95 2.05 1.73 0.32 42 0.11 0.37 0.53 0.77concnear ECTOTH 95 2.61 2.10 0.52 61 0.11 0.22 0.46 0.74concnear2 ECTOTH 95 2.05 2.10 -0.05 27 0.11 0.35 0.53 0.80concnear ECTOTL 95 2.61 1.52 1.09 101 0.09 0.17 0.43 0.63concnear2 ECTOTL 95 2.05 1.52 0.52 58 0.09 0.32 0.52 0.72concnear TOR 88 2.31 1.70 0.61 47 0.10 0.30 0.59 0.78concnear2 TOR 88 1.81 1.70 0.11 15 0.17 0.30 0.59 0.85concnear TORH 88 2.31 2.23 0.08 13 0.11 0.26 0.60 0.84concnear2 TORH 88 1.81 2.23 -0.42 -12 0.08 0.22 0.52 0.92concnear TORL 88 2.31 1.19 1.12 110 0.10 0.26 0.41 0.65concnear2 TORL 88 1.81 1.19 0.62 65 0.14 0.31 0.52 0.74
Notes:3. Modeled variable:concnear Nearest modeled DPM concentration consistent with the draft 2000
NONROAD Modelconcnear2 Nearest modeled DPM concentration consistent with the draft 2002
NONROAD Model4. Monitored variable:ECTOR EC value multiplied by TOR average correction factor (missing for EC
measured using TOT).ECTORH EC value multiplied by TOR maximum correction factor (missing for EC
measured using TOT).ECTORL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOT).ECTOT EC value multiplied by TOT average correction factor (missing for EC
measured using TOR).ECTOTH EC value multiplied by TOT maximum correction factor (missing for EC
measured using TOR).ECTOTL EC value multiplied by TOR minimum correction factor (missing for EC
measured using TOR).TOR ECOCX value multiplied by TOT average correction factor for EPA data,
ECTOR for TOR data.TORH ECOCX value multiplied by TOT maximum correction factor for EPA
data, ECTOR for TOR data..TORL ECOCX value multiplied by TOR minimum correction factor for EPA
data, ECTOR for TOR data.
ICF Consulting 29
Tables 4-2 to 4-4 show the monitored and modeled (nearest tract, DPM modeled valuesconsistent with the draft 2000 NONROAD model) DPM values. In most cases the modeledvalues are within at most a factor of 2 of the monitored values.
Table 4-5 summarizes the differences and percentage differences between all the sets of modeledand monitored values. The ECTOR outlier value for the IMPROVE Washington DC site isexcluded from these calculations.
For the TOR comparisons excluding the EPA ECOCX data, the best model performance basedon the mean percentage difference and based on the fraction of modeled values within 100 % ofthe monitored value is for the DPM-maximum value consistent with the 2002 NONROADmodel. For all 15 sites the mean percentage difference is 26 % and the fraction of modeledvalues within 100 % of the monitored value is 73 %. This fraction would have been 69 %including the outlier. For the 10 rural sites the mean percentage difference is 20 % and thefraction of modeled values within 100 % of the monitored value is 70 %. For the 5 urban sitesthe mean percentage difference is 39 % and the fraction of modeled values within 100 % of themonitored value is 80 %. Interestingly, this finding is different to the regression analyses whichfound the model performance to be slightly worse with the modeled value consistent with therevised NONROAD model and better using the DPM-average monitored values.
For the TOT comparisons, the best model performance based on the mean percentage differenceand based on the fraction of modeled values within 100 % of the monitored value is also for theDPM-maximum value consistent with the 2002 NONROAD model. (The results for the subset ofOn-road dominated sites are even better, but are based on only 6 monitors) For all 95 sites themean percentage difference is 27 % and the fraction of modeled values within 100 % of themonitored value is 80 %. For the 30 rural sites the mean percentage difference is 16 % and thefraction of modeled values within 100 % of the monitored value is 80 %. For the 65 urban sitesthe mean percentage difference is 32 % and the fraction of modeled values within 100 % of themonitored value is 80 %.
For the TOR comparisons including the EPA ECOCX data, the best model performance basedon the mean percentage difference and based on the fraction of modeled values within 100 % ofthe monitored value is also for the DPM-maximum value using the 2002 NONROAD model. Forall 88 sites the mean percentage difference is -12 % and the fraction of modeled values within100 % of the monitored value is 92 %. For the 30 rural sites the mean percentage difference is 6% and the fraction of modeled values within 100 % of the monitored value is 87 %. For the 58urban sites the mean percentage difference is -14 % and the fraction of modeled values within100 % of the monitored value is 95 %.
Discussion
The model performance evaluation based on the regression models leads to different conclusionsthan the model performance evaluation based on the differences. One primary reason is that theregression analyses are more influenced by the extreme values. For most purposes the analysis ofthe differences is more useful, especially since all of the regression models fitted relativelypoorly (except for those with only 2 data points). The best model performance based on the mean
ICF Consulting 30
percentage difference and based on the fraction of modeled values within 100 % of themonitored value is for the DPM-maximum value consistent with the 2002 NONROAD model.The corresponding fractions of modeled values within 100 % of the monitored value are 73 %for all TOR sites excluding the EPA ECOCX data, 80 % for all TOT sites, and 92 % for all TORsites including the EPA ECOCX data. As discussed in Chapter 1, this performance comparesfavorably with the model to monitor results for the other pollutants assessed in the NSATA,except for benzene.
ICF
Con
sult
ing
31
Tab
le 4
-1. R
egre
ssio
n m
odel
s fo
r m
odel
ed a
gain
st m
onito
red
DP
M c
once
ntra
tions
.
Nu
mb
erM
od
eled
Var
iab
le1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
NIn
terc
ept
Inte
rcep
tS
tan
dar
d E
rro
rS
lop
eS
lop
eS
tan
dar
d E
rro
rR S
qu
ared
1co
ncne
arE
CT
OR
All
All
150.
800.
490.
810.
420.
222
conc
near
EC
TO
RA
llR
ural
100.
040.
611.
570.
590.
473
conc
near
EC
TO
RA
llU
rban
51.
780.
740.
070.
550.
014
conc
near
EC
TO
RN
on-r
oad
All
41.
992.
220.
391.
200.
055
conc
near
EC
TO
RN
on-r
oad
Rur
al2
-2.3
7.
3.20
.1.
006
conc
near
EC
TO
RN
on-r
oad
Urb
an2
3.71
.-0
.64
.1.
007
conc
near
2E
CT
OR
All
All
150.
670.
360.
560.
310.
208
conc
near
2E
CT
OR
All
Rur
al10
0.14
0.45
1.07
0.43
0.44
9co
ncne
ar2
EC
TO
RA
llU
rban
51.
390.
560.
040.
410.
0010
max
conc
EC
TO
RA
llA
ll14
3.40
1.33
1.41
1.12
0.12
11m
axco
ncE
CT
OR
All
Rur
al9
2.24
1.95
2.16
1.79
0.17
12m
axco
ncE
CT
OR
All
Urb
an5
4.91
2.16
0.75
1.60
0.07
13m
axco
ncE
CT
OR
Non
-roa
dA
ll4
4.45
2.07
1.32
1.12
0.41
14m
axco
ncE
CT
OR
Non
-roa
dR
ural
27.
60.
0.00
..
15m
axco
ncE
CT
OR
Non
-roa
dU
rban
22.
23.
1.98
.1.
0016
max
conc
2E
CT
OR
All
All
142.
490.
940.
980.
790.
1117
max
conc
2E
CT
OR
All
Rur
al9
1.66
1.38
1.50
1.26
0.17
18m
axco
nc2
EC
TO
RA
llU
rban
53.
571.
490.
511.
100.
0719
conc
near
EC
TO
RH
All
All
150.
890.
480.
580.
330.
1920
conc
near
EC
TO
RH
All
Rur
al10
0.04
0.61
1.37
0.51
0.47
21co
ncne
arE
CT
OR
HA
llU
rban
51.
780.
740.
050.
390.
0122
conc
near
EC
TO
RH
Non
-roa
dA
ll4
2.28
2.12
0.17
0.89
0.02
23co
ncne
arE
CT
OR
HN
on-r
oad
Rur
al2
-2.3
7.
2.80
.1.
0024
conc
near
EC
TO
RH
Non
-roa
dU
rban
23.
71.
-0.4
6.
1.00
25co
ncne
ar2
EC
TO
RH
All
All
150.
730.
350.
400.
240.
1826
conc
near
2E
CT
OR
HA
llR
ural
100.
140.
450.
940.
380.
4427
conc
near
2E
CT
OR
HA
llU
rban
51.
390.
560.
030.
300.
0028
max
conc
EC
TO
RH
All
All
143.
451.
271.
100.
840.
1229
max
conc
EC
TO
RH
All
Rur
al9
2.24
1.95
1.89
1.56
0.17
30m
axco
ncE
CT
OR
HA
llU
rban
54.
912.
160.
541.
150.
0731
max
conc
EC
TO
RH
Non
-roa
dA
ll4
4.90
2.12
0.82
0.89
0.30
32m
axco
ncE
CT
OR
HN
on-r
oad
Rur
al2
7.60
.0.
00.
.33
max
conc
EC
TO
RH
Non
-roa
dU
rban
22.
23.
1.42
.1.
00
ICF
Con
sult
ing
32
Tab
le 4
-1. R
egre
ssio
n m
odel
s fo
r m
odel
ed a
gain
st m
onito
red
DP
M c
once
ntra
tions
.
Nu
mb
erM
od
eled
Var
iab
le1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
NIn
terc
ept
Inte
rcep
tS
tan
dar
d E
rro
rS
lop
eS
lop
eS
tan
dar
d E
rro
rR S
qu
ared
34m
axco
nc2
EC
TO
RH
All
All
142.
520.
890.
770.
590.
1235
max
conc
2E
CT
OR
HA
llR
ural
91.
661.
381.
311.
100.
1736
max
conc
2E
CT
OR
HA
llU
rban
53.
571.
490.
370.
790.
0737
conc
near
EC
TO
RL
All
All
150.
820.
491.
150.
610.
2138
conc
near
EC
TO
RL
All
Rur
al10
0.04
0.61
2.33
0.87
0.47
39co
ncne
arE
CT
OR
LA
llU
rban
51.
780.
740.
100.
770.
0140
conc
near
EC
TO
RL
Non
-roa
dA
ll4
2.07
2.20
0.49
1.71
0.04
41co
ncne
arE
CT
OR
LN
on-r
oad
Rur
al2
-2.3
7.
4.75
.1.
0042
conc
near
EC
TO
RL
Non
-roa
dU
rban
23.
71.
-0.9
0.
1.00
43co
ncne
ar2
EC
TO
RL
All
All
150.
690.
350.
790.
450.
1944
conc
near
2E
CT
OR
LA
llR
ural
100.
140.
451.
590.
640.
4445
conc
near
2E
CT
OR
LA
llU
rban
51.
390.
560.
060.
580.
0046
max
conc
EC
TO
RL
All
All
143.
401.
322.
051.
600.
1247
max
conc
EC
TO
RL
All
Rur
al9
2.24
1.95
3.20
2.65
0.17
48m
axco
ncE
CT
OR
LA
llU
rban
54.
912.
161.
052.
250.
0749
max
conc
EC
TO
RL
Non
-roa
dA
ll4
4.57
2.09
1.80
1.63
0.38
50m
axco
ncE
CT
OR
LN
on-r
oad
Rur
al2
7.60
.0.
00.
.51
max
conc
EC
TO
RL
Non
-roa
dU
rban
22.
23.
2.79
.1.
0052
max
conc
2E
CT
OR
LA
llA
ll14
2.49
0.93
1.42
1.12
0.12
53m
axco
nc2
EC
TO
RL
All
Rur
al9
1.66
1.38
2.23
1.87
0.17
54m
axco
nc2
EC
TO
RL
All
Urb
an5
3.57
1.49
0.72
1.56
0.07
55co
ncne
arE
CT
OT
All
All
951.
740.
320.
510.
140.
1256
conc
near
EC
TO
TA
llR
ural
301.
260.
430.
540.
230.
1657
conc
near
EC
TO
TA
llU
rban
652.
050.
430.
450.
180.
0958
conc
near
EC
TO
TN
on-r
oad
All
191.
601.
311.
820.
590.
3659
conc
near
EC
TO
TN
on-r
oad
Rur
al7
0.69
2.18
2.16
1.38
0.33
60co
ncne
arE
CT
OT
Non
-roa
dU
rban
122.
311.
981.
610.
790.
2961
conc
near
EC
TO
TO
n-ro
adA
ll6
1.48
0.45
0.21
0.18
0.25
62co
ncne
arE
CT
OT
On-
road
Rur
al2
1.43
.0.
47.
1.00
63co
ncne
arE
CT
OT
On-
road
Urb
an4
1.24
0.74
0.25
0.26
0.32
64co
ncne
ar2
EC
TO
TA
llA
ll95
1.37
0.23
0.39
0.10
0.13
65co
ncne
ar2
EC
TO
TA
llR
ural
300.
960.
310.
430.
170.
1966
conc
near
2E
CT
OT
All
Urb
an65
1.63
0.31
0.34
0.13
0.10
ICF
Con
sult
ing
33
Tab
le 4
-1. R
egre
ssio
n m
odel
s fo
r m
odel
ed a
gain
st m
onito
red
DP
M c
once
ntra
tions
.
Nu
mb
erM
od
eled
Var
iab
le1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
NIn
terc
ept
Inte
rcep
tS
tan
dar
d E
rro
rS
lop
eS
lop
eS
tan
dar
d E
rro
rR S
qu
ared
67m
axco
ncE
CT
OT
All
All
954.
674.
847.
912.
150.
1368
max
conc
EC
TO
TA
llR
ural
30-0
.35
2.22
5.49
1.22
0.42
69m
axco
ncE
CT
OT
All
Urb
an65
8.85
7.13
7.73
2.95
0.10
70m
axco
ncE
CT
OT
Non
-roa
dA
ll19
-4.2
426
.43
27.1
111
.98
0.23
71m
axco
ncE
CT
OT
Non
-roa
dR
ural
7-4
.39
7.71
9.05
4.87
0.41
72m
axco
ncE
CT
OT
Non
-roa
dU
rban
1227
.73
40.3
919
.83
16.1
70.
1373
max
conc
EC
TO
TO
n-ro
adA
ll6
3.39
1.46
0.74
0.59
0.27
8374
max
conc
EC
TO
TO
n-ro
adR
ural
22.
11.
2.06
.1.
0000
75m
axco
ncE
CT
OT
On-
road
Urb
an4
3.17
2.51
0.75
0.88
0.26
6076
max
conc
2E
CT
OT
All
All
953.
453.
375.
521.
500.
1271
77m
axco
nc2
EC
TO
TA
llR
ural
30-0
.09
1.54
3.88
0.84
0.43
1778
max
conc
2E
CT
OT
All
Urb
an65
6.37
4.96
5.38
2.05
0.09
8179
conc
near
EC
TO
TH
All
All
952.
010.
300.
290.
100.
0792
80co
ncne
arE
CT
OT
HA
llR
ural
301.
480.
400.
300.
160.
1058
81co
ncne
arE
CT
OT
HA
llU
rban
652.
320.
400.
260.
130.
0598
82co
ncne
arE
CT
OT
HN
on-r
oad
All
192.
071.
361.
350.
530.
2792
83co
ncne
arE
CT
OT
HN
on-r
oad
Rur
al7
2.64
2.85
0.65
1.42
0.03
9884
conc
near
EC
TO
TH
Non
-roa
dU
rban
122.
651.
811.
300.
630.
2986
85co
ncne
arE
CT
OT
HO
n-ro
adA
ll6
1.56
0.44
0.12
0.12
0.20
0486
conc
near
EC
TO
TH
On-
road
Rur
al2
1.59
.0.
26.
1.00
0087
conc
near
EC
TO
TH
On-
road
Urb
an4
1.36
0.74
0.15
0.18
0.25
3788
conc
near
2E
CT
OT
HA
llA
ll95
1.57
0.22
0.23
0.07
0.09
0389
conc
near
2E
CT
OT
HA
llR
ural
301.
140.
290.
240.
120.
1300
90co
ncne
ar2
EC
TO
TH
All
Urb
an65
1.83
0.29
0.20
0.09
0.06
6691
max
conc
EC
TO
TH
All
All
959.
114.
554.
411.
560.
0793
92m
axco
ncE
CT
OT
HA
llR
ural
301.
062.
113.
540.
840.
3864
93m
axco
ncE
CT
OT
HA
llU
rban
6514
.11
6.62
4.15
2.14
0.05
6499
max
conc
EC
TO
TH
Non
-roa
dA
ll19
19.3
428
.55
12.9
411
.05
0.07
4695
max
conc
EC
TO
TH
Non
-roa
dR
ural
71.
7510
.41
3.80
5.19
0.09
7096
max
conc
EC
TO
TH
Non
-roa
dU
rban
1253
.90
39.1
47.
3113
.65
0.02
7997
max
conc
EC
TO
TH
On-
road
All
63.
661.
410.
440.
400.
2261
98m
axco
ncE
CT
OT
HO
n-ro
adR
ural
22.
80.
1.14
.1.
0000
99m
axco
ncE
CT
OT
HO
n-ro
adU
rban
43.
532.
480.
430.
610.
2002
ICF
Con
sult
ing
34
Tab
le 4
-1. R
egre
ssio
n m
odel
s fo
r m
odel
ed a
gain
st m
onito
red
DP
M c
once
ntra
tions
.
Nu
mb
erM
od
eled
Var
iab
le1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
NIn
terc
ept
Inte
rcep
tS
tan
dar
d E
rro
rS
lop
eS
lop
eS
tan
dar
d E
rro
rR S
qu
ared
100
max
conc
2E
CT
OT
HA
llA
ll95
6.54
3.17
3.08
1.09
0.07
9610
1m
axco
nc2
EC
TO
TH
All
Rur
al30
0.90
1.46
2.50
0.58
0.39
5210
2m
axco
nc2
EC
TO
TH
All
Urb
an65
10.0
34.
612.
891.
490.
0563
103
conc
near
EC
TO
TL
All
All
951.
520.
330.
720.
170.
1573
104
conc
near
EC
TO
TL
All
Rur
al30
1.01
0.44
0.85
0.30
0.22
2710
5co
ncne
arE
CT
OT
LA
llU
rban
651.
850.
450.
620.
220.
1171
106
conc
near
EC
TO
TL
Non
-roa
dA
ll19
1.92
1.23
1.83
0.61
0.34
8510
7co
ncne
arE
CT
OT
LN
on-r
oad
Rur
al7
0.59
1.50
2.65
1.09
0.54
0010
8co
ncne
arE
CT
OT
LN
on-r
oad
Urb
an12
2.74
2.03
1.52
0.87
0.23
3310
9co
ncne
arE
CT
OT
LO
n-ro
adA
ll6
1.39
0.47
0.33
0.25
0.30
6111
0co
ncne
arE
CT
OT
LO
n-ro
adR
ural
21.
14.
0.86
.1.
0000
111
conc
near
EC
TO
TL
On-
road
Urb
an4
1.11
0.74
0.40
0.34
0.40
9511
2co
ncne
ar2
EC
TO
TL
All
All
951.
210.
240.
550.
130.
1719
113
conc
near
2E
CT
OT
LA
llR
ural
300.
770.
320.
670.
210.
2580
114
conc
near
2E
CT
OT
LA
llU
rban
651.
480.
330.
470.
160.
1250
115
max
conc
EC
TO
TL
All
All
951.
024.
9911
.38
2.61
0.16
9611
6m
axco
ncE
CT
OT
LA
llR
ural
30-1
.52
2.37
7.43
1.60
0.43
5011
7m
axco
ncE
CT
OT
LA
llU
rban
654.
547.
3811
.25
3.55
0.13
7311
8m
axco
ncE
CT
OT
LN
on-r
oad
All
19-1
0.13
23.0
533
.26
11.3
40.
3360
119
max
conc
EC
TO
TL
Non
-roa
dR
ural
7-3
.85
5.39
10.3
63.
920.
5827
120
max
conc
EC
TO
TL
Non
-roa
dU
rban
1214
.93
37.6
427
.43
16.1
40.
2242
121
max
conc
EC
TO
TL
On-
road
All
63.
061.
491.
150.
800.
3415
122
max
conc
EC
TO
TL
On-
road
Rur
al2
0.83
.3.
76.
1.00
0012
3m
axco
ncE
CT
OT
LO
n-ro
adU
rban
42.
732.
511.
201.
160.
3499
124
max
conc
2E
CT
OT
LA
llA
ll95
0.91
3.47
7.94
1.82
0.17
0112
5m
axco
nc2
EC
TO
TL
All
Rur
al30
-0.9
31.
645.
251.
110.
4462
126
max
conc
2E
CT
OT
LA
llU
rban
653.
375.
147.
832.
470.
1372
127
conc
near
TO
RA
llA
ll88
0.77
0.48
0.91
0.27
0.11
7712
8co
ncne
arT
OR
All
Rur
al30
0.34
0.66
1.06
0.37
0.22
8612
9co
ncne
arT
OR
All
Urb
an58
1.26
0.73
0.68
0.40
0.04
8313
0co
ncne
arT
OR
Non
-roa
dA
ll18
2.04
1.65
1.06
0.83
0.09
2013
1co
ncne
arT
OR
Non
-roa
dR
ural
90.
602.
481.
571.
240.
1861
132
conc
near
TO
RN
on-r
oad
Urb
an9
2.93
2.39
0.82
1.22
0.06
06
ICF
Con
sult
ing
35
Tab
le 4
-1. R
egre
ssio
n m
odel
s fo
r m
odel
ed a
gain
st m
onito
red
DP
M c
once
ntra
tions
.
Nu
mb
erM
od
eled
Var
iab
le1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
NIn
terc
ept
Inte
rcep
tS
tan
dar
d E
rro
rS
lop
eS
lop
eS
tan
dar
d E
rro
rR S
qu
ared
133
conc
near
TO
RO
n-ro
adA
ll5
1.82
1.67
-0.0
30.
840.
0006
134
conc
near
TO
RO
n-ro
adR
ural
21.
20.
0.40
.1.
0000
135
conc
near
TO
RO
n-ro
adU
rban
38.
231.
20-3
.52
0.63
0.96
8813
6co
ncne
ar2
TO
RA
llA
ll88
0.63
0.36
0.69
0.20
0.12
2813
7co
ncne
ar2
TO
RA
llR
ural
300.
290.
470.
800.
260.
2455
138
conc
near
2T
OR
All
Urb
an58
1.02
0.55
0.52
0.30
0.04
9713
9m
axco
ncT
OR
All
All
87-3
.38
9.21
10.4
65.
060.
0478
140
max
conc
TO
RA
llR
ural
290.
282.
853.
581.
570.
1616
141
max
conc
TO
RA
llU
rban
58-9
.75
15.6
816
.33
8.62
0.06
0314
2m
axco
ncT
OR
Non
-roa
dA
ll18
-0.9
538
.73
20.7
419
.53
0.06
5914
3m
axco
ncT
OR
Non
-roa
dR
ural
90.
938.
624.
034.
300.
1118
144
max
conc
TO
RN
on-r
oad
Urb
an9
-0.1
658
.79
36.9
429
.99
0.17
8114
5m
axco
ncT
OR
On-
road
All
54.
405.
46-0
.02
2.74
0.00
0014
6m
axco
ncT
OR
On-
road
Rur
al2
1.10
.1.
75.
1.00
0014
7m
axco
ncT
OR
On-
road
Urb
an3
26.7
93.
66-1
1.99
1.93
0.97
4814
8m
axco
nc2
TO
RA
llA
ll87
-2.2
76.
417.
353.
530.
0486
149
max
conc
2T
OR
All
Rur
al29
0.20
1.99
2.60
1.09
0.17
2915
0m
axco
nc2
TO
RA
llU
rban
58-6
.62
10.9
211
.39
6.00
0.06
0415
1co
ncne
arT
OR
HA
llA
ll88
0.79
0.48
0.68
0.20
0.11
8315
2co
ncne
arT
OR
HA
llR
ural
300.
340.
660.
920.
320.
2286
153
conc
near
TO
RH
All
Urb
an58
1.26
0.73
0.49
0.29
0.04
8315
4co
ncne
arT
OR
HN
on-r
oad
All
181.
941.
570.
880.
620.
1103
155
conc
near
TO
RH
Non
-roa
dR
ural
90.
602.
481.
371.
080.
1861
156
conc
near
TO
RH
Non
-roa
dU
rban
92.
932.
390.
590.
880.
0606
157
conc
near
TO
RH
On-
road
All
52.
541.
71-0
.31
0.67
0.06
6715
8co
ncne
arT
OR
HO
n-ro
adR
ural
21.
20.
0.35
.1.
0000
159
conc
near
TO
RH
On-
road
Urb
an3
8.23
1.20
-2.5
30.
450.
9688
160
conc
near
2T
OR
HA
llA
ll88
0.63
0.36
0.53
0.15
0.12
7216
1co
ncne
ar2
TO
RH
All
Rur
al30
0.29
0.47
0.70
0.23
0.24
5516
2co
ncne
ar2
TO
RH
All
Urb
an58
1.02
0.55
0.37
0.22
0.04
9716
3m
axco
ncT
OR
HA
llA
ll87
-7.1
98.
949.
673.
740.
0728
164
max
conc
TO
RH
All
Rur
al29
0.28
2.85
3.13
1.37
0.16
1616
5m
axco
ncT
OR
HA
llU
rban
58-9
.75
15.6
811
.73
6.19
0.06
03
ICF
Con
sult
ing
36
Tab
le 4
-1. R
egre
ssio
n m
odel
s fo
r m
odel
ed a
gain
st m
onito
red
DP
M c
once
ntra
tions
.
Nu
mb
erM
od
eled
Var
iab
le1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
NIn
terc
ept
Inte
rcep
tS
tan
dar
d E
rro
rS
lop
eS
lop
eS
tan
dar
d E
rro
rR S
qu
ared
166
max
conc
TO
RH
Non
-roa
dA
ll18
-20.
8835
.28
24.8
614
.00
0.16
4816
7m
axco
ncT
OR
HN
on-r
oad
Rur
al9
0.93
8.62
3.52
3.75
0.11
1816
8m
axco
ncT
OR
HN
on-r
oad
Urb
an9
-0.1
658
.79
26.5
521
.56
0.17
8116
9m
axco
ncT
OR
HO
n-ro
adA
ll5
5.68
5.73
-0.5
22.
240.
0179
170
max
conc
TO
RH
On-
road
Rur
al2
1.10
.1.
53.
1.00
0017
1m
axco
ncT
OR
HO
n-ro
adU
rban
326
.79
3.66
-8.6
21.
390.
9748
172
max
conc
2T
OR
HA
llA
ll87
-4.9
16.
236.
782.
610.
0737
173
max
conc
2T
OR
HA
llR
ural
290.
201.
992.
270.
960.
1729
174
max
conc
2T
OR
HA
llU
rban
58-6
.62
10.9
28.
184.
310.
0604
175
conc
near
TO
RL
All
All
880.
750.
481.
310.
380.
1194
176
conc
near
TO
RL
All
Rur
al30
0.34
0.66
1.57
0.54
0.22
8617
7co
ncne
arT
OR
LA
llU
rban
581.
260.
730.
950.
560.
0483
178
conc
near
TO
RL
Non
-roa
dA
ll18
1.98
1.64
1.57
1.19
0.09
8217
9co
ncne
arT
OR
LN
on-r
oad
Rur
al9
0.60
2.48
2.32
1.83
0.18
6118
0co
ncne
arT
OR
LN
on-r
oad
Urb
an9
2.93
2.39
1.15
1.72
0.06
0618
1co
ncne
arT
OR
LO
n-ro
adA
ll5
2.01
1.73
-0.1
91.
250.
0073
182
conc
near
TO
RL
On-
road
Rur
al2
1.20
.0.
59.
1.00
0018
3co
ncne
arT
OR
LO
n-ro
adU
rban
38.
231.
20-4
.96
0.89
0.96
8818
4co
ncne
ar2
TO
RL
All
All
880.
620.
361.
010.
290.
1256
185
conc
near
2T
OR
LA
llR
ural
300.
290.
471.
180.
390.
2455
186
conc
near
2T
OR
LA
llU
rban
581.
020.
550.
730.
430.
0497
187
max
conc
TO
RL
All
All
87-4
.70
9.19
16.0
87.
240.
0549
188
max
conc
TO
RL
All
Rur
al29
0.28
2.85
5.30
2.32
0.16
1618
9m
axco
ncT
OR
LA
llU
rban
58-9
.75
15.6
822
.99
12.1
30.
0603
190
max
conc
TO
RL
Non
-roa
dA
ll18
-7.2
038
.11
34.8
327
.75
0.08
9619
1m
axco
ncT
OR
LN
on-r
oad
Rur
al9
0.93
8.62
5.98
6.37
0.11
1819
2m
axco
ncT
OR
LN
on-r
oad
Urb
an9
-0.1
658
.79
52.0
142
.24
0.17
8119
3m
axco
ncT
OR
LO
n-ro
adA
ll5
4.73
5.66
-0.2
74.
090.
0014
194
max
conc
TO
RL
On-
road
Rur
al2
1.10
.2.
59.
1.00
0019
5m
axco
ncT
OR
LO
n-ro
adU
rban
326
.79
3.66
-16.
882.
720.
9748
196
max
conc
2T
OR
LA
llA
ll87
-3.1
96.
4011
.29
5.04
0.05
5719
7m
axco
nc2
TO
RL
All
Rur
al29
0.20
1.99
3.86
1.62
0.17
2919
8m
axco
nc2
TO
RL
All
Urb
an58
-6.6
210
.92
16.0
48.
450.
0604
ICF
Con
sult
ing
37
Not
es f
or T
able
4-1
:1.
M
odel
ed v
aria
ble:
conc
near
Nea
rest
mod
eled
DPM
con
cent
ratio
n co
nsis
tent
with
the
draf
t 200
0 N
ON
RO
AD
Mod
elco
ncne
ar2
Nea
rest
mod
eled
DPM
con
cent
ratio
n co
nsis
tent
with
the
draf
t 200
2 N
ON
RO
AD
Mod
elm
axco
ncN
earb
y (w
ithin
30
km)
max
imum
mod
eled
DPM
con
cent
ratio
n co
nsis
tent
with
the
draf
t 200
0 N
ON
RO
AD
Mod
elm
axco
nc2
Nea
rby
(with
in 3
0 km
) m
axim
um m
odel
ed D
PM c
once
ntra
tion
cons
iste
nt w
ith th
e dr
aft 2
002
NO
NR
OA
DM
odel
2.
Mon
itore
d va
riab
le:
EC
TO
RE
C v
alue
mul
tipl
ied
by T
OR
ave
rage
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
RH
EC
val
ue m
ultip
lied
by T
OR
max
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
RL
EC
val
ue m
ultip
lied
by T
OR
min
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
TE
C v
alue
mul
tipl
ied
by T
OT
ave
rage
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
EC
TO
TH
EC
val
ue m
ultip
lied
by T
OT
max
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
EC
TO
TL
EC
val
ue m
ultip
lied
by T
OR
min
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
TO
RE
CO
CX
val
ue m
ultip
lied
by T
OT
ave
rage
cor
rect
ion
fact
or f
or E
PA d
ata,
EC
TO
R f
or T
OR
dat
a.T
OR
HE
CO
CX
val
ue m
ultip
lied
by T
OT
max
imum
cor
rect
ion
fact
or f
or E
PA d
ata,
EC
TO
R f
or T
OR
dat
a..
TO
RL
EC
OC
X v
alue
mul
tiplie
d by
TO
R m
inim
um c
orre
ctio
n fa
ctor
for
EPA
dat
a, E
CT
OR
for
TO
R d
ata.
3.
Subs
et:
All
All
sit
esN
on-r
oad
Site
s do
min
ated
by
Non
-roa
d so
urce
(at
leas
t 75
% o
f m
odel
ed D
PM c
onsi
sten
t with
the
draf
t 200
0 N
ON
RO
AD
Mod
el)
On-
road
Site
s do
min
ated
by
On-
road
sou
rce
(at l
east
50
% o
f m
odel
ed D
PM c
onsi
sten
t with
the
draf
t 200
0 N
ON
RO
AD
Mod
el)
ICF
Con
sult
ing
38
Tab
le 4
-2. N
eare
st m
odel
ed c
once
ntra
tion
and
EC
TO
R, E
CT
OR
L, a
nd E
CT
OR
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
EC
TO
R4
EC
TO
RL
4E
CT
OR
H4
Oc
ea
nvi
lleN
J2.
032
R0.
4357
550.
2940
030.
4987
56Se
att
leW
A3.
2264
UN
on
-ro
ad
0.75
129
0.53
3525
1.04
5273
She
na
nd
oa
h N
atio
na
lP
ark
VA
0.96
12R
0.30
8284
0.20
7999
0.35
2855
Lake
Ta
ho
eN
V0.
6111
R0.
9424
380.
6358
621.
0786
94W
ash
ing
ton
DC
DC
12.2
63R
No
n-
roa
d1.
0127
30.
6832
881.
1591
49
Ye
llow
sto
ne
Na
tion
al
Pa
rkW
Y0.
0548
R0.
1066
370.
0719
480.
1220
55
De
nve
rC
O0.
7626
RN
on
-ro
ad
0.97
8226
0.66
0008
1.11
9656
De
nve
rC
O1.
9723
UN
on
-ro
ad
2.71
1257
1.92
5386
3.77
2184
De
nve
rC
O1.
5939
U0.
4525
860.
3214
020.
6296
85D
en
ver
CO
0.73
04R
1.33
0274
0.89
7534
1.52
2603
De
nve
rC
O0.
7953
U0.
8205
710.
5827
241.
1416
64D
en
ver
CO
2.59
18R
0.81
5005
0.54
9883
0.93
2837
De
nve
rC
O1.
6963
U0.
5937
280.
4216
330.
8260
56D
en
ver
CO
1.39
13R
1.06
5238
0.71
8715
1.21
9248
De
nve
rC
O0.
4174
R0.
5609
070.
3784
430.
6420
02D
en
ver
CO
4.58
09R
No
n-
roa
d2.
1697
91.
4639
552.
4834
95
Not
es f
or T
able
4-2
:1.
C
ity
/ cou
nty
and
Stat
e m
ay a
ppea
r m
ultip
le ti
mes
if th
ere
are
seve
ral d
iffe
rent
mon
itori
ng s
ites
at th
at g
ener
al lo
catio
n.2.
M
odel
ed v
aria
ble
= “
conc
near
”N
eare
st m
odel
ed D
PM c
once
ntra
tion
cons
iste
nt w
ith th
e dr
aft 2
000
NO
NR
OA
D M
odel
3.
Subs
et:
Non
-roa
dSi
tes
dom
inat
ed b
y N
on-r
oad
sour
ce (
at le
ast 7
5 %
of
mod
eled
DPM
con
sist
ent w
ith th
e dr
aft 2
000
NO
NR
OA
DM
odel
)O
n-ro
adSi
tes
dom
inat
ed b
y O
n-ro
ad s
ourc
e (a
t lea
st 5
0 %
of
mod
eled
DPM
con
sist
ent w
ith th
e dr
aft 2
000
NO
NR
OA
DM
odel
)4.
M
onito
red
vari
able
:
ICF
Con
sult
ing
39
EC
TO
RE
C v
alue
mul
tipl
ied
by T
OR
ave
rage
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
RH
EC
val
ue m
ultip
lied
by T
OR
max
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
RL
EC
val
ue m
ultip
lied
by T
OR
min
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
5.
Out
lier
valu
e no
t use
d fo
r st
atis
tical
com
pari
sons
ICF
Con
sult
ing
40
Tab
le 4
-3. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
T, T
OT
L, a
nd T
OT
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
T4
TO
TL
4T
OT
H4
Ch
ico
pe
eM
A1.
7019
U0.
6837
160.
6837
160.
6837
16H
arr
iso
nM
A4.
0016
UN
on
-ro
ad
1.88
9732
1.88
9732
1.88
9732
Ad
am
sP
A1.
501
R0.
8383
760.
8383
760.
8383
76C
ha
mp
aig
nIL
1.20
5R
0.73
6247
0.73
6247
0.73
6247
Trig
gK
Y0.
8134
R0.
6744
210.
6744
210.
6744
21To
mp
kin
sN
Y0.
7298
R0.
5104
660.
5104
660.
5104
66W
ash
ing
ton
IN0.
8839
R0.
7095
420.
7095
420.
7095
42M
erc
er
PA
0.94
14R
0.98
0116
0.98
0116
0.98
0116
No
ble
OH
0.97
18R
1.20
8977
1.20
8977
1.20
8977
Win
n P
aris
hLA
0.35
88R
0.44
4273
0.32
9798
0.64
5967
Jeff
ers
on
Ala
ba
ma
1.99
58U
3.01
4976
3.01
4976
3.01
4976
Ma
rico
pa
Ariz
on
a1.
5009
U1.
3527
11.
0041
591.
9668
24Fr
esn
oC
alif
orn
ia0.
6956
UO
n-r
oa
d1.
4865
941.
1035
462.
1614
9K
ern
Ca
lifo
rnia
0.71
67U
1.72
0329
1.27
7054
2.50
1338
Riv
ers
ide
Ca
lifo
rnia
2.31
67R
2.19
6692
1.63
0673
3.19
3963
Sac
ram
en
toC
alif
orn
ia1.
2622
U1.
1891
640.
8827
541.
7290
3Sa
n D
ieg
oC
alif
orn
ia1.
6952
UO
n-r
oa
d1.
3420
480.
9962
441.
9513
21Sa
nta
Cla
raC
alif
orn
ia2.
5023
U1.
7174
21.
2748
952.
4971
08V
en
tura
Ca
lifo
rnia
2.25
41R
On
-ro
ad
1.74
411.
2947
2.53
59A
da
ms
Co
lora
do
1.74
3R
No
n-
roa
d1.
9298
11.
4325
582.
8059
2
Ke
nt
De
law
are
2.30
66U
0.95
8036
0.95
8036
0.95
8036
Ne
w C
ast
leD
ela
wa
re3.
903
U1.
9802
881.
9802
881.
9802
88D
istric
t o
f C
olu
mb
iaD
istric
t o
fC
olu
mb
ia3.
3894
U1.
6544
081.
6544
081.
6544
08
Da
de
Flo
rida
2.47
64U
4.23
7155
4.23
7155
4.23
7155
Hill
sbo
rou
gh
Flo
rida
1.44
1U
1.04
8591
1.04
8591
1.04
8591
De
Ka
lbG
eo
rgia
3.31
68R
2.37
7515
2.37
7515
2.37
7515
Co
ok
Illin
ois
10.2
885
UN
on
-ro
ad
2.23
2918
2.23
2918
2.23
2918
Co
ok
Illin
ois
2.24
96U
1.82
7492
1.82
7492
1.82
7492
ICF
Con
sult
ing
41
Tab
le 4
-3. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
T, T
OT
L, a
nd T
OT
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
T4
TO
TL
4T
OT
H4
Co
ok
Illin
ois
2.34
21U
1.43
8204
1.43
8204
1.43
8204
Ma
rion
Ind
ian
a2.
3757
U1.
3648
531.
3648
531.
3648
53Li
nn
Iow
a1.
0643
R0.
7156
30.
7156
30.
7156
3P
olk
Iow
a1.
0352
R0.
5885
650.
4369
10.
8557
67Sc
ott
Iow
a1.
4764
U0.
7646
040.
7646
040.
7646
04W
yan
do
tte
Ka
nsa
s1.
7453
U1.
1027
30.
8185
911.
6033
57Ea
st B
ato
n R
ou
ge
Pa
rish
Lou
isia
na
6.74
31R
No
n-
roa
d1.
4202
141.
4202
141.
4202
14
Balti
mo
reM
ary
lan
d2.
9017
U1.
5484
651.
5484
651.
5484
65Ba
ltim
ore
Ma
ryla
nd
3.75
13U
1.10
4987
1.10
4987
1.10
4987
Ha
mp
de
nM
ass
ac
hu
sett
s1.
7019
U0.
5865
250.
5865
250.
5865
25Su
ffo
lkM
ass
ac
hu
sett
s4.
6792
U1.
3472
311.
3472
311.
3472
31O
akl
an
dM
ich
iga
n1.
9572
U1.
6752
11.
6752
11.
6752
1W
ayn
eM
ich
iga
n2.
0105
U1.
5521
971.
5521
971.
5521
97W
ayn
eM
ich
iga
n2.
0734
U1.
2488
261.
2488
261.
2488
26H
en
ne
pin
Min
ne
sota
2.24
69U
0.49
3075
0.36
6025
0.71
6925
He
nn
ep
inM
inn
eso
ta2.
3247
U0.
8561
370.
6355
371.
2448
13H
arr
iso
nM
issi
ssip
pi
0.98
87U
0.94
3544
0.94
3544
0.94
3544
Jeff
ers
on
Mis
sou
ri1.
2994
R1.
2184
581.
2184
581.
2184
58St
. Lo
uis
Mis
sou
ri2.
1799
U1.
5822
431.
5822
431.
5822
43M
isso
ula
Mo
nta
na
0.51
05U
0.73
4287
0.54
5084
1.06
7644
Do
ug
las
Ne
bra
ska
1.35
8U
0.67
3697
0.50
0106
0.97
9547
Wa
sho
eN
eva
da
3.98
17R
No
n-
roa
d1.
6047
141.
1912
292.
3332
34
Ca
md
en
Ne
w J
ers
ey
4.18
74U
1.40
286
1.40
286
1.40
286
Mid
dle
sex
Ne
w J
ers
ey
3.38
77U
1.40
4459
1.40
4459
1.40
4459
Mo
rris
Ne
w J
ers
ey
2.83
48R
0.82
8597
0.82
8597
0.82
8597
Un
ion
Ne
w J
ers
ey
5.29
35U
No
n-
roa
d4.
1983
294.
1983
294.
1983
29
Bro
nx
Ne
w Y
ork
8.00
06U
No
n-
roa
d1.
7750
31.
7750
31.
7750
3
Bro
nx
Ne
w Y
ork
7.14
04U
No
n-
roa
d2.
6067
242.
6067
242.
6067
24
Bro
nx
Ne
w Y
ork
8.59
15U
No
n-
2.39
4239
2.39
4239
2.39
4239
ICF
Con
sult
ing
42
Tab
le 4
-3. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
T, T
OT
L, a
nd T
OT
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
T4
TO
TL
4T
OT
H4
roa
dEs
sex
Ne
w Y
ork
0.22
18R
0.38
3065
0.38
3065
0.38
3065
Mo
nro
eN
ew
Yo
rk1.
7872
RO
n-r
oa
d0.
7513
360.
7513
360.
7513
36Q
ue
en
sN
ew
Yo
rk6.
6119
UN
on
-ro
ad
1.62
4842
1.62
4842
1.62
4842
Ste
ub
en
Ne
w Y
ork
0.59
36R
0.44
4674
0.44
4674
0.44
4674
Me
ckl
en
bu
rgN
ort
h C
aro
lina
2.27
06U
1.35
2154
1.35
2154
1.35
2154
Burle
igh
No
rth
Da
kota
0.53
87U
No
n-
roa
d0.
3444
820.
2557
20.
5008
72
Ca
ssN
ort
h D
ako
ta0.
6603
RN
on
-ro
ad
0.59
1956
0.43
9428
0.86
0697
Cu
yah
og
aO
hio
6.95
79R
No
n-
roa
d2.
4340
122.
4340
122.
4340
12
Tuls
aO
kla
ho
ma
1.48
89U
0.76
7198
0.56
9515
1.11
5497
Mu
ltno
ma
hO
reg
on
2.08
5U
1.14
655
0.85
112
1.66
707
Ad
am
sP
en
nsy
lva
nia
1.50
1R
0.57
1485
0.57
1485
0.57
1485
Alle
gh
en
yP
en
nsy
lva
nia
2.77
74U
1.83
2849
1.83
2849
1.83
2849
Alle
gh
en
yP
en
nsy
lva
nia
1.96
99U
1.36
5899
1.36
5899
1.36
5899
Ph
ilad
elp
hia
Pe
nn
sylv
an
ia4.
9986
U1.
7690
761.
7690
761.
7690
76W
ash
ing
ton
Pe
nn
sylv
an
ia1.
1656
R0.
8565
760.
8565
760.
8565
76W
est
mo
rela
nd
Pe
nn
sylv
an
ia1.
6285
U1.
5885
811.
5885
811.
5885
81C
ha
rlest
on
Sou
th C
aro
lina
0.89
3U
0.85
3549
0.85
3549
0.85
3549
She
lby
Ten
ne
sse
e3.
0845
UN
on
-ro
ad
2.99
937
2.99
937
2.99
937
Da
llas
Texa
s3.
3398
RN
on
-ro
ad
0.84
3332
0.62
6032
1.22
6195
El P
aso
Texa
s0.
8152
U0.
8507
710.
6315
541.
2370
1H
arr
isTe
xas
2.39
03U
0.48
2679
0.35
8308
0.70
1809
Salt
Lake
Uta
h1.
5764
U1.
2924
070.
9593
941.
8791
44U
tah
Uta
h0.
606
R0.
8868
360.
6583
261.
2894
49C
hitt
en
de
nV
erm
on
t1.
855
U0.
7999
920.
7999
920.
7999
92R
ich
mo
nd
Virg
inia
1.54
32U
1.14
5547
1.14
5547
1.14
5547
Kin
gW
ash
ing
ton
3.65
13R
No
n-
roa
d1.
4930
961.
1083
722.
1709
44
ICF
Con
sult
ing
43
Tab
le 4
-3. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
T, T
OT
L, a
nd T
OT
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
T4
TO
TL
4T
OT
H4
Kin
gW
ash
ing
ton
3.22
64U
No
n-
roa
d1.
0087
210.
7488
051.
4666
68
Milw
au
kee
Wis
co
nsi
n2.
3524
UO
n-r
oa
d1.
3064
851.
3064
851.
3064
85P
ue
rto
Ric
o1.
1811
U5.
0364
145.
0364
145.
0364
14A
na
he
imC
A3.
5344
U3.
7534
112.
7862
745.
4574
13Bu
rba
nk
CA
2.60
58U
On
-ro
ad
5.17
8159
3.84
3909
7.52
8979
Fon
tan
aC
A2.
0988
R5.
0476
553.
7470
327.
3392
29H
un
ting
ton
Pa
rkC
A4.
6059
U7.
3902
855.
4860
410
.745
38C
en
tra
l LA
CA
3.05
31R
5.75
4731
4.27
1917
8.36
7309
Lon
g B
ea
ch
CA
11.7
44U
No
n-
roa
d4.
1345
513.
0692
076.
0115
87
Pic
o R
ive
raC
A2.
4915
U7.
0975
025.
2686
9810
.319
68R
ub
ido
ux
CA
2.55
44U
5.59
7411
4.15
5133
8.13
8567
Ph
oe
nix
AZ
2.56
82U
No
n-
roa
d1.
7796
641.
3211
2.58
761
Not
es f
or T
able
4-3
1.
Cit
y / c
ount
y an
d St
ate
may
app
ear
mul
tiple
tim
es if
ther
e ar
e se
vera
l dif
fere
nt m
onito
ring
site
s at
that
gen
eral
loca
tion.
2.
Mod
eled
var
iabl
e =
“co
ncne
ar”
Nea
rest
mod
eled
DPM
con
cent
ratio
n co
nsis
tent
with
the
draf
t 200
0 N
ON
RO
AD
Mod
el3.
Su
bset
:N
on-r
oad
Site
s do
min
ated
by
Non
-roa
d so
urce
(at
leas
t 75
% o
f m
odel
ed D
PM c
onsi
sten
t with
the
draf
t 200
0 N
ON
RO
AD
Mod
el)
On-
road
Site
s do
min
ated
by
On-
road
sou
rce
(at l
east
50
% o
f m
odel
ed D
PM c
onsi
sten
t with
the
draf
t 200
0 N
ON
RO
AD
Mod
el)
4.
Mon
itore
d va
riab
le:
EC
TO
TE
C v
alue
mul
tipl
ied
by T
OT
ave
rage
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
EC
TO
TH
EC
val
ue m
ultip
lied
by T
OT
max
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
EC
TO
TL
EC
val
ue m
ultip
lied
by T
OR
min
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
5.
Out
lier
valu
e no
t use
d fo
r st
atis
tical
com
pari
sons
ICF
Con
sult
ing
44
Tab
le 4
-4. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
R, T
OR
L, a
nd T
OR
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
R4
TO
RL
4T
OR
H4
Jeffe
rson
Ala
bam
a1.
9958
U2.
8728
32.
0401
263.
9969
81M
aric
opa
Ariz
ona
1.50
09U
1.65
6433
1.17
6308
2.30
4603
Fre
sno
Cal
iforn
ia0.
6956
UO
n-ro
ad2.
1124
931.
5001
762.
9391
2K
ern
Cal
iforn
ia0.
7167
U2.
2417
411.
5919
613.
1189
44R
iver
side
Cal
iforn
ia2.
3167
R3.
2206
932.
1729
983.
6863
36S
acra
men
toC
alifo
rnia
1.26
22U
1.80
5432
1.28
2118
2.51
1905
San
Die
goC
alifo
rnia
1.69
52U
On-
road
1.90
3438
1.35
1717
2.64
8261
San
ta C
lara
Cal
iforn
ia2.
5023
U1.
8496
251.
3135
022.
5733
92V
entu
raC
alifo
rnia
2.25
41R
On-
road
2.62
6535
1.77
212
3.00
6275
Ada
ms
Col
orad
o1.
743
RN
on-r
oad
2.50
8408
1.69
242
2.87
1069
Ken
tD
elaw
are
2.30
66U
1.82
3765
1.29
5138
2.53
7412
New
Cas
tleD
elaw
are
3.90
3U
2.25
5988
1.60
2079
3.13
8766
Dis
tric
t of C
olum
bia
Dis
tric
t of C
olum
bia
3.38
94U
1.98
1439
1.40
7109
2.75
6785
Dad
eF
lorid
a2.
4764
U2.
2731
141.
6142
413.
1625
94H
illsb
orou
ghF
lorid
a1.
441
U1.
5642
981.
1108
782.
1764
14D
eKal
bG
eorg
ia3.
3168
R2.
8126
881.
8977
173.
2193
41C
ook
Illin
ois
2.24
96U
1.83
4895
1.30
3041
2.55
2898
Coo
kIll
inoi
s2.
3421
U1.
2646
690.
8980
981.
7595
4M
ario
nIn
dian
a2.
3757
U1.
8532
1.31
6041
2.57
8366
Linn
Iow
a1.
0643
R1.
3312
750.
8982
11.
5237
48P
olk
Iow
a1.
0352
R1.
5766
181.
0637
421.
8045
63S
cott
Iow
a1.
4764
U1.
2535
910.
8902
311.
7441
27W
yand
otte
Kan
sas
1.74
53U
2.23
7625
1.58
9038
3.11
3217
Eas
t Bat
on R
ouge
Par
ish
Loui
sian
a6.
7431
RN
on-r
oad
1.65
8044
1.11
868
1.89
7761
Bal
timor
eM
aryl
and
2.90
17U
1.88
6666
1.33
9806
2.62
4927
Bal
timor
eM
aryl
and
3.75
13U
1.90
302
1.35
142
2.64
768
Ham
pden
Mas
sach
uset
ts1.
7019
U0.
8989
090.
6383
561.
2506
56S
uffo
lkM
assa
chus
etts
4.67
92U
1.87
3434
1.33
041
2.60
6518
Oak
land
Mic
higa
n1.
9572
U1.
8078
171.
2838
122.
5152
24W
ayne
Mic
higa
n2.
0105
U1.
7921
321.
2726
732.
4934
01W
ayne
Mic
higa
n2.
0734
U2.
0179
591.
4330
432.
8075
95H
enne
pin
Min
neso
ta2.
3247
U1.
6034
131.
1386
552.
2308
35
ICF
Con
sult
ing
45
Tab
le 4
-4. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
R, T
OR
L, a
nd T
OR
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
R4
TO
RL
4T
OR
H4
Har
rison
Mis
siss
ippi
0.98
87U
1.57
3738
1.11
7582
2.18
9548
Jeffe
rson
Mis
sour
i1.
2994
R2.
2740
061.
5342
692.
6027
78S
t. Lo
uis
Mis
sour
i2.
1799
U1.
8688
361.
3271
442.
6001
19M
isso
ula
Mon
tana
0.51
05U
1.68
2974
1.19
5156
2.34
1529
Dou
glas
Neb
rask
a1.
358
U1.
4315
21.
0165
871.
9916
8W
asho
eN
evad
a3.
9817
RN
on-r
oad
2.80
9452
1.89
5534
3.21
5638
Cam
den
New
Jer
sey
4.18
74U
1.91
7837
1.36
1942
2.66
8295
Mid
dles
exN
ew J
erse
y3.
3877
U1.
7338
851.
2313
12.
4123
62M
orris
New
Jer
sey
2.83
48R
1.79
8739
1.21
3607
2.05
8798
Uni
onN
ew J
erse
y5.
2935
UN
on-r
oad
2.86
6781
2.03
583
3.98
8565
Bro
nxN
ew Y
ork
7.14
04U
Non
-roa
d2.
2124
771.
5711
793.
0782
28B
ronx
New
Yor
k8.
5915
UN
on-r
oad
1.83
2909
1.30
1631
2.55
0134
Ess
exN
ew Y
ork
0.22
18R
1.10
2289
0.74
3713
1.26
1656
Mon
roe
New
Yor
k1.
7872
RO
n-ro
ad1.
4592
650.
9845
641.
6702
43Q
ueen
sN
ew Y
ork
6.61
19U
Non
-roa
d1.
5821
661.
1235
672.
2012
75S
teub
enN
ew Y
ork
0.59
36R
1.22
3072
0.82
5205
1.39
9902
Mec
klen
burg
Nor
th C
arol
ina
2.27
06U
1.98
954
1.41
2862
2.76
8056
Bur
leig
hN
orth
Dak
ota
0.53
87U
Non
-roa
d0.
9671
50.
6868
171.
3456
Cas
sN
orth
Dak
ota
0.66
03R
Non
-roa
d1.
5129
521.
0207
871.
7316
92C
uyah
oga
Ohi
o6.
9579
RN
on-r
oad
2.63
7039
1.77
9207
3.01
8298
Tul
saO
klah
oma
1.48
89U
1.85
8217
1.31
9603
2.58
5345
Mul
tnom
ahO
rego
n2.
085
U1.
8004
1.27
8545
2.50
4904
Ada
ms
Pen
nsyl
vani
a1.
501
R1.
9873
281.
3408
482.
2746
53A
llegh
eny
Pen
nsyl
vani
a2.
7774
U2.
1119
181.
4997
682.
9383
2A
llegh
eny
Pen
nsyl
vani
a1.
9699
U1.
9462
731.
3821
362.
7078
58P
hila
delp
hia
Pen
nsyl
vani
a4.
9986
U1.
8391
061.
3060
322.
5587
56W
ashi
ngto
nP
enns
ylva
nia
1.16
56R
2.08
5923
1.40
737
2.38
7502
Wes
tmor
elan
dP
enns
ylva
nia
1.62
85U
1.96
3036
1.39
404
2.73
118
Cha
rlest
onS
outh
Car
olin
a0.
893
U1.
6143
351.
1464
122.
2460
32S
helb
yT
enne
ssee
3.08
45U
Non
-roa
d2.
4157
631.
7155
423.
3610
61D
alla
sT
exas
3.33
98R
Non
-roa
d1.
4419
50.
9728
821.
6504
25E
l Pas
oT
exas
0.81
52U
1.03
7519
0.73
6789
1.44
3505
Har
risT
exas
2.39
03U
0.88
649
0.62
9536
1.23
3377
ICF
Con
sult
ing
46
Tab
le 4
-4. N
eare
st m
odel
ed c
once
ntra
tion
and
TO
R, T
OR
L, a
nd T
OR
H m
onito
red
valu
es fo
r ea
ch s
ite.
Cit
y / C
ou
nty
1S
tate
1N
eare
stM
od
eled
Co
nce
ntr
atio
n2
Urb
an o
rR
ura
lS
ub
set3
TO
R4
TO
RL
4T
OR
H4
Sal
t Lak
eU
tah
1.57
64U
1.67
9742
1.19
286
2.33
7033
Uta
hU
tah
0.60
6R
1.79
0894
1.20
8314
2.04
9819
Chi
ttend
enV
erm
ont
1.85
5U
1.39
7072
0.99
2123
1.94
3752
Ric
hmon
dV
irgin
ia1.
5432
U2.
0931
341.
4864
292.
9121
87K
ing
Was
hing
ton
3.65
13R
Non
-roa
d1.
5180
211.
0242
071.
7374
94K
ing
Was
hing
ton
3.22
64U
Non
-roa
d0.
9589
050.
6809
621.
3341
29M
ilwau
kee
Wis
cons
in2.
3524
UO
n-ro
ad1.
6469
251.
1695
552.
2913
74P
uert
o R
ico
1.18
11U
2.45
4399
1.74
2979
3.41
4816
Oce
anvi
lleN
J2.
032
R0.
4357
550.
2940
030.
4987
56S
eattl
eW
A3.
2264
UN
on-r
oad
0.75
129
0.53
3525
1.04
5273
She
nand
oah
Nat
iona
l Par
kV
A0.
9612
R0.
3082
840.
2079
990.
3528
55La
ke T
ahoe
NV
0.61
11R
0.94
2438
0.63
5862
1.07
8694
Was
hing
ton
DC
DC
12.2
635
RN
on-r
oad
1.01
273
0.68
3288
1.15
9149
Yel
low
ston
e N
atio
nal P
ark
WY
0.05
48R
0.10
6637
0.07
1948
0.12
2055
Den
ver
CO
0.76
26R
Non
-roa
d0.
9782
260.
6600
081.
1196
56D
enve
rC
O1.
9723
UN
on-r
oad
2.71
1257
1.92
5386
3.77
2184
Den
ver
CO
1.59
39U
0.45
2586
0.32
1402
0.62
9685
Den
ver
CO
0.73
04R
1.33
0274
0.89
7534
1.52
2603
Den
ver
CO
0.79
53U
0.82
0571
0.58
2724
1.14
1664
Den
ver
CO
2.59
18R
0.81
5005
0.54
9883
0.93
2837
Den
ver
CO
1.69
63U
0.59
3728
0.42
1633
0.82
6056
Den
ver
CO
1.39
13R
1.06
5238
0.71
8715
1.21
9248
Den
ver
CO
0.41
74R
0.56
0907
0.37
8443
0.64
2002
Den
ver
CO
4.58
09R
Non
-roa
d2.
1697
91.
4639
552.
4834
95
Not
es f
or T
able
4-4
:1.
C
ity
/ cou
nty
and
Stat
e m
ay a
ppea
r m
ultip
le ti
mes
if th
ere
are
seve
ral d
iffe
rent
mon
itori
ng s
ites
at th
at g
ener
al lo
catio
n.2.
M
odel
ed v
aria
ble
= “
conc
near
”N
eare
st m
odel
ed D
PM c
once
ntra
tion
cons
iste
nt w
ith th
e dr
aft 2
000
NO
NR
OA
D M
odel
3.
Subs
et:
Non
-roa
dSi
tes
dom
inat
ed b
y N
on-r
oad
sour
ce (
at le
ast 7
5 %
of
mod
eled
DPM
con
sist
ent w
ith th
e dr
aft 2
000
NO
NR
OA
DM
odel
)O
n-ro
adSi
tes
dom
inat
ed b
y O
n-ro
ad s
ourc
e (a
t lea
st 5
0 %
of
mod
eled
DPM
con
sist
ent w
ith th
e dr
aft 2
000
NO
NR
OA
DM
odel
)
ICF
Con
sult
ing
47
4.
Mon
itore
d va
riab
le:
TO
RE
CO
CX
val
ue m
ultip
lied
by T
OT
ave
rage
cor
rect
ion
fact
or f
or E
PA d
ata,
EC
TO
R f
or T
OR
dat
a.T
OR
HE
CO
CX
val
ue m
ultip
lied
by T
OT
max
imum
cor
rect
ion
fact
or f
or E
PA d
ata,
EC
TO
R f
or T
OR
dat
a..
TO
RL
EC
OC
X v
alue
mul
tiplie
d by
TO
R m
inim
um c
orre
ctio
n fa
ctor
for
EPA
dat
a, E
CT
OR
for
TO
R d
ata.
5.
Out
lier
valu
e no
t use
d fo
r st
atis
tical
com
pari
sons
ICF
Con
sult
ing
48
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%co
ncne
arE
CT
OR
All
All
151.
560.
940.
6310
00.
070.
130.
530.
53co
ncne
arE
CT
OR
All
Rur
al10
1.41
0.87
0.54
760.
000.
100.
600.
60co
ncne
arE
CT
OR
All
Urb
an5
1.86
1.07
0.79
147
0.20
0.20
0.40
0.40
conc
near
EC
TO
RN
on-r
oad
All
42.
641.
650.
9898
0.00
0.25
0.50
0.50
conc
near
EC
TO
RN
on-r
oad
Rur
al2
2.67
1.57
1.10
450.
000.
500.
500.
50co
ncne
arE
CT
OR
Non
-roa
dU
rban
22.
601.
730.
8715
10.
000.
000.
500.
50co
ncne
ar2
EC
TO
RA
llA
ll15
1.20
0.94
0.26
560.
070.
130.
470.
60co
ncne
ar2
EC
TO
RA
llR
ural
101.
070.
870.
2038
0.10
0.10
0.50
0.70
conc
near
2E
CT
OR
All
Urb
an5
1.44
1.07
0.37
930.
000.
200.
400.
40m
axco
ncE
CT
OR
All
All
144.
801.
003.
8151
60.
070.
070.
070.
21m
axco
ncE
CT
OR
All
Rur
al9
4.30
0.96
3.35
416
0.11
0.11
0.11
0.33
max
conc
EC
TO
RA
llU
rban
55.
711.
074.
6469
60.
000.
000.
000.
00m
axco
ncE
CT
OR
Non
-roa
dA
ll4
6.63
1.65
4.98
376
0.00
0.00
0.00
0.00
max
conc
EC
TO
RN
on-r
oad
Rur
al2
7.60
1.57
6.03
464
0.00
0.00
0.00
0.00
max
conc
EC
TO
RN
on-r
oad
Urb
an2
5.66
1.73
3.93
288
0.00
0.00
0.00
0.00
max
conc
2E
CT
OR
All
All
143.
461.
002.
4734
60.
000.
070.
210.
36m
axco
nc2
EC
TO
RA
llR
ural
93.
100.
962.
1527
50.
000.
110.
330.
33m
axco
nc2
EC
TO
RA
llU
rban
54.
111.
073.
0547
30.
000.
000.
000.
40co
ncne
arE
CT
OR
HA
llA
ll15
1.56
1.16
0.40
620.
000.
070.
400.
60co
ncne
arE
CT
OR
HA
llR
ural
101.
411.
000.
4254
0.00
0.10
0.40
0.70
conc
near
EC
TO
RH
All
Urb
an5
1.86
1.48
0.37
780.
000.
000.
400.
40co
ncne
arE
CT
OR
HN
on-r
oad
All
42.
642.
110.
5353
0.00
0.00
0.50
0.75
conc
near
EC
TO
RH
Non
-roa
dR
ural
22.
671.
800.
8726
0.00
0.00
0.50
1.00
conc
near
EC
TO
RH
Non
-roa
dU
rban
22.
602.
410.
1980
0.00
0.00
0.50
0.50
conc
near
2E
CT
OR
HA
llA
ll15
1.20
1.16
0.04
260.
000.
070.
330.
73co
ncne
ar2
EC
TO
RH
All
Rur
al10
1.07
1.00
0.08
200.
000.
100.
400.
70co
ncne
ar2
EC
TO
RH
All
Urb
an5
1.44
1.48
-0.0
439
0.00
0.00
0.20
0.80
max
conc
EC
TO
RH
All
All
144.
801.
233.
5739
40.
000.
070.
140.
29m
axco
ncE
CT
OR
HA
llR
ural
94.
301.
093.
2135
10.
000.
110.
220.
33m
axco
ncE
CT
OR
HA
llU
rban
55.
711.
484.
2247
20.
000.
000.
000.
20m
axco
ncE
CT
OR
HN
on-r
oad
All
46.
632.
114.
5328
60.
000.
000.
000.
00
ICF
Con
sult
ing
49
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%m
axco
ncE
CT
OR
HN
on-r
oad
Rur
al2
7.60
1.80
5.80
392
0.00
0.00
0.00
0.00
max
conc
EC
TO
RH
Non
-roa
dU
rban
25.
662.
413.
2517
90.
000.
000.
000.
00m
axco
nc2
EC
TO
RH
All
All
143.
461.
232.
2325
80.
070.
140.
360.
36m
axco
nc2
EC
TO
RH
All
Rur
al9
3.10
1.09
2.01
228
0.11
0.22
0.33
0.33
max
conc
2E
CT
OR
HA
llU
rban
54.
111.
482.
6331
20.
000.
000.
400.
40co
ncne
arE
CT
OR
LA
llA
ll15
1.56
0.64
0.92
190
0.13
0.40
0.47
0.53
conc
near
EC
TO
RL
All
Rur
al10
1.41
0.59
0.83
161
0.10
0.50
0.50
0.60
conc
near
EC
TO
RL
All
Urb
an5
1.86
0.76
1.10
248
0.20
0.20
0.40
0.40
conc
near
EC
TO
RL
Non
-roa
dA
ll4
2.64
1.15
1.49
184
0.25
0.50
0.50
0.50
conc
near
EC
TO
RL
Non
-roa
dR
ural
22.
671.
061.
6111
40.
000.
500.
500.
50co
ncne
arE
CT
OR
LN
on-r
oad
Urb
an2
2.60
1.23
1.37
254
0.50
0.50
0.50
0.50
conc
near
2E
CT
OR
LA
llA
ll15
1.20
0.64
0.55
126
0.07
0.33
0.47
0.53
conc
near
2E
CT
OR
LA
llR
ural
101.
070.
590.
4910
40.
000.
300.
500.
60co
ncne
ar2
EC
TO
RL
All
Urb
an5
1.44
0.76
0.68
171
0.20
0.40
0.40
0.40
max
conc
EC
TO
RL
All
All
144.
800.
694.
1279
20.
000.
000.
070.
07m
axco
ncE
CT
OR
LA
llR
ural
94.
300.
653.
6666
50.
000.
000.
110.
11m
axco
ncE
CT
OR
LA
llU
rban
55.
710.
764.
9510
210.
000.
000.
000.
00m
axco
ncE
CT
OR
LN
on-r
oad
All
46.
631.
155.
4859
10.
000.
000.
000.
00m
axco
ncE
CT
OR
LN
on-r
oad
Rur
al2
7.60
1.06
6.54
735
0.00
0.00
0.00
0.00
max
conc
EC
TO
RL
Non
-roa
dU
rban
25.
661.
234.
4344
60.
000.
000.
000.
00m
axco
nc2
EC
TO
RL
All
All
143.
460.
692.
7854
50.
070.
070.
070.
14m
axco
nc2
EC
TO
RL
All
Rur
al9
3.10
0.65
2.46
456
0.11
0.11
0.11
0.22
max
conc
2E
CT
OR
LA
llU
rban
54.
110.
763.
3570
70.
000.
000.
000.
00co
ncne
arE
CT
OT
All
All
952.
611.
730.
8880
0.12
0.21
0.45
0.68
conc
near
EC
TO
TA
llR
ural
301.
991.
360.
6366
0.13
0.30
0.60
0.73
conc
near
EC
TO
TA
llU
rban
652.
901.
901.
0086
0.11
0.17
0.38
0.66
conc
near
EC
TO
TN
on-r
oad
All
195.
171.
963.
2017
10.
110.
160.
260.
32co
ncne
arE
CT
OT
Non
-roa
dR
ural
73.
871.
472.
3916
40.
140.
290.
290.
29co
ncne
arE
CT
OT
Non
-roa
dU
rban
125.
922.
253.
6817
50.
080.
080.
250.
33co
ncne
arE
CT
OT
On-
road
All
61.
901.
97-0
.07
280.
000.
000.
500.
83co
ncne
arE
CT
OT
On-
road
Rur
al2
2.02
1.25
0.77
840.
000.
000.
500.
50co
ncne
arE
CT
OT
On-
road
Urb
an4
1.84
2.33
-0.4
91
0.00
0.00
0.50
1.00
ICF
Con
sult
ing
50
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%co
ncne
ar2
EC
TO
TA
llA
ll95
2.05
1.73
0.32
420.
110.
370.
530.
77co
ncne
ar2
EC
TO
TA
llR
ural
301.
551.
360.
1930
0.10
0.40
0.63
0.80
conc
near
2E
CT
OT
All
Urb
an65
2.28
1.90
0.37
470.
110.
350.
480.
75m
axco
ncE
CT
OT
All
All
9518
.36
1.73
16.6
391
80.
020.
040.
080.
12m
axco
ncE
CT
OT
All
Rur
al30
7.11
1.36
5.76
423
0.03
0.03
0.10
0.13
max
conc
EC
TO
TA
llU
rban
6523
.55
1.90
21.6
511
460.
020.
050.
080.
11m
axco
ncE
CT
OT
Non
-roa
dA
ll19
48.9
81.
9647
.02
2142
0.00
0.00
0.00
0.05
max
conc
EC
TO
TN
on-r
oad
Rur
al7
8.95
1.47
7.48
489
0.00
0.00
0.00
0.14
max
conc
EC
TO
TN
on-r
oad
Urb
an12
72.3
42.
2570
.09
3106
0.00
0.00
0.00
0.00
max
conc
EC
TO
TO
n-ro
adA
ll6
4.84
1.97
2.87
214
0.00
0.17
0.33
0.33
max
conc
EC
TO
TO
n-ro
adR
ural
24.
681.
253.
4330
70.
000.
000.
000.
00m
axco
ncE
CT
OT
On-
road
Urb
an4
4.92
2.33
2.59
168
0.00
0.25
0.50
0.50
max
conc
2E
CT
OT
All
All
9512
.99
1.73
11.2
662
60.
020.
090.
120.
21m
axco
nc2
EC
TO
TA
llR
ural
305.
171.
363.
8128
40.
000.
130.
130.
30m
axco
nc2
EC
TO
TA
llU
rban
6516
.60
1.90
14.7
078
30.
030.
080.
110.
17co
ncne
arE
CT
OT
HA
llA
ll95
2.61
2.10
0.52
610.
110.
220.
460.
74co
ncne
arE
CT
OT
HA
llR
ural
301.
991.
710.
2848
0.07
0.27
0.57
0.80
conc
near
EC
TO
TH
All
Urb
an65
2.90
2.28
0.63
670.
120.
200.
420.
71co
ncne
arE
CT
OT
HN
on-r
oad
All
195.
172.
292.
8813
80.
160.
210.
320.
47co
ncne
arE
CT
OT
HN
on-r
oad
Rur
al7
3.87
1.89
1.98
116
0.00
0.14
0.29
0.57
conc
near
EC
TO
TH
Non
-roa
dU
rban
125.
922.
523.
4015
10.
250.
250.
330.
42co
ncne
arE
CT
OT
HO
n-ro
adA
ll6
1.90
2.71
-0.8
110
0.00
0.33
0.33
0.83
conc
near
EC
TO
TH
On-
road
Rur
al2
2.02
1.64
0.38
630.
000.
500.
500.
50co
ncne
arE
CT
OT
HO
n-ro
adU
rban
41.
843.
24-1
.40
-17
0.00
0.25
0.25
1.00
conc
near
2E
CT
OT
HA
llA
ll95
2.05
2.10
-0.0
527
0.11
0.35
0.53
0.80
conc
near
2E
CT
OT
HA
llR
ural
301.
551.
71-0
.16
160.
100.
400.
600.
80co
ncne
ar2
EC
TO
TH
All
Urb
an65
2.28
2.28
0.00
320.
110.
320.
490.
80m
axco
ncE
CT
OT
HA
llA
ll95
18.3
62.
1016
.26
832
0.03
0.05
0.12
0.19
max
conc
EC
TO
TH
All
Rur
al30
7.11
1.71
5.41
369
0.00
0.03
0.17
0.27
max
conc
EC
TO
TH
All
Urb
an65
23.5
52.
2821
.27
1046
0.05
0.06
0.09
0.15
max
conc
EC
TO
TH
Non
-roa
dA
ll19
48.9
82.
2946
.69
2051
0.00
0.05
0.05
0.16
max
conc
EC
TO
TH
Non
-roa
dR
ural
78.
951.
897.
0637
60.
000.
140.
140.
43
ICF
Con
sult
ing
51
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%m
axco
ncE
CT
OT
HN
on-r
oad
Urb
an12
72.3
42.
5269
.81
3027
0.00
0.00
0.00
0.00
max
conc
EC
TO
TH
On-
road
All
64.
842.
712.
1316
90.
170.
170.
330.
33m
axco
ncE
CT
OT
HO
n-ro
adR
ural
24.
681.
643.
0325
60.
000.
000.
000.
00m
axco
ncE
CT
OT
HO
n-ro
adU
rban
44.
923.
241.
6812
50.
250.
250.
500.
50m
axco
nc2
EC
TO
TH
All
All
9512
.99
2.10
10.9
056
40.
020.
080.
180.
27m
axco
nc2
EC
TO
TH
All
Rur
al30
5.17
1.71
3.46
245
0.03
0.17
0.27
0.37
max
conc
2E
CT
OT
HA
llU
rban
6516
.60
2.28
14.3
371
20.
020.
050.
140.
23co
ncne
arE
CT
OT
LA
llA
ll95
2.61
1.52
1.09
101
0.09
0.17
0.43
0.63
conc
near
EC
TO
TL
All
Rur
al30
1.99
1.16
0.83
860.
130.
270.
570.
70co
ncne
arE
CT
OT
LA
llU
rban
652.
901.
691.
2110
80.
080.
120.
370.
60co
ncne
arE
CT
OT
LN
on-r
oad
All
195.
171.
783.
3920
70.
050.
110.
160.
26co
ncne
arE
CT
OT
LN
on-r
oad
Rur
al7
3.87
1.24
2.63
219
0.00
0.14
0.14
0.29
conc
near
EC
TO
TL
Non
-roa
dU
rban
125.
922.
093.
8320
10.
080.
080.
170.
25co
ncne
arE
CT
OT
LO
n-ro
adA
ll6
1.90
1.55
0.35
490.
000.
000.
330.
83co
ncne
arE
CT
OT
LO
n-ro
adR
ural
22.
021.
021.
0010
60.
000.
000.
000.
50co
ncne
arE
CT
OT
LO
n-ro
adU
rban
41.
841.
810.
0220
0.00
0.00
0.50
1.00
conc
near
2E
CT
OT
LA
llA
ll95
2.05
1.52
0.52
580.
090.
320.
520.
72co
ncne
ar2
EC
TO
TL
All
Rur
al30
1.55
1.16
0.39
460.
100.
430.
630.
73co
ncne
ar2
EC
TO
TL
All
Urb
an65
2.28
1.69
0.59
640.
090.
260.
460.
71m
axco
ncE
CT
OT
LA
llA
ll95
18.3
61.
5216
.84
1014
0.02
0.02
0.05
0.11
max
conc
EC
TO
TL
All
Rur
al30
7.11
1.16
5.95
484
0.00
0.00
0.10
0.10
max
conc
EC
TO
TL
All
Urb
an65
23.5
51.
6921
.86
1258
0.03
0.03
0.03
0.11
max
conc
EC
TO
TL
Non
-roa
dA
ll19
48.9
81.
7847
.21
2244
0.00
0.00
0.00
0.00
max
conc
EC
TO
TL
Non
-roa
dR
ural
78.
951.
247.
7161
50.
000.
000.
000.
00m
axco
ncE
CT
OT
LN
on-r
oad
Urb
an12
72.3
42.
0970
.25
3194
0.00
0.00
0.00
0.00
max
conc
EC
TO
TL
On-
road
All
64.
841.
553.
2926
50.
170.
170.
170.
33m
axco
ncE
CT
OT
LO
n-ro
adR
ural
24.
681.
023.
6536
30.
000.
000.
000.
00m
axco
ncE
CT
OT
LO
n-ro
adU
rban
44.
921.
813.
1021
60.
250.
250.
250.
50m
axco
nc2
EC
TO
TL
All
All
9512
.99
1.52
11.4
769
40.
000.
070.
110.
16m
axco
nc2
EC
TO
TL
All
Rur
al30
5.17
1.16
4.01
329
0.00
0.10
0.10
0.17
max
conc
2E
CT
OT
LA
llU
rban
6516
.60
1.69
14.9
186
30.
000.
060.
110.
15co
ncne
arT
OR
All
All
882.
311.
700.
6147
0.10
0.30
0.59
0.78
ICF
Con
sult
ing
52
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%co
ncne
arT
OR
All
Rur
al30
2.04
1.60
0.44
380.
000.
200.
570.
73co
ncne
arT
OR
All
Urb
an58
2.44
1.75
0.70
510.
160.
340.
600.
81co
ncne
arT
OR
Non
-roa
dA
ll18
4.01
1.86
2.14
128
0.00
0.06
0.33
0.44
conc
near
TO
RN
on-r
oad
Rur
al9
3.60
1.91
1.69
870.
000.
110.
330.
44co
ncne
arT
OR
Non
-roa
dU
rban
94.
411.
812.
6016
80.
000.
000.
330.
44co
ncne
arT
OR
On-
road
All
51.
761.
95-0
.19
-50.
000.
600.
801.
00co
ncne
arT
OR
On-
road
Rur
al2
2.02
2.04
-0.0
24
0.00
1.00
1.00
1.00
conc
near
TO
RO
n-ro
adU
rban
31.
581.
89-0
.31
-12
0.00
0.33
0.67
1.00
conc
near
2T
OR
All
All
881.
811.
700.
1115
0.17
0.30
0.59
0.85
conc
near
2T
OR
All
Rur
al30
1.57
1.60
-0.0
37
0.13
0.17
0.50
0.87
conc
near
2T
OR
All
Urb
an58
1.93
1.75
0.19
190.
190.
360.
640.
84m
axco
ncT
OR
All
All
8714
.57
1.72
12.8
572
10.
050.
070.
170.
25m
axco
ncT
OR
All
Rur
al29
6.20
1.65
4.55
296
0.07
0.10
0.24
0.38
max
conc
TO
RA
llU
rban
5818
.75
1.75
17.0
193
30.
030.
050.
140.
19m
axco
ncT
OR
Non
-roa
dA
ll18
37.6
91.
8635
.83
1802
0.00
0.00
0.11
0.11
max
conc
TO
RN
on-r
oad
Rur
al9
8.65
1.91
6.74
365
0.00
0.00
0.22
0.22
max
conc
TO
RN
on-r
oad
Urb
an9
66.7
31.
8164
.92
3239
0.00
0.00
0.00
0.00
max
conc
TO
RO
n-ro
adA
ll5
4.36
1.95
2.41
134
0.00
0.00
0.20
0.20
max
conc
TO
RO
n-ro
adR
ural
24.
682.
042.
6313
40.
000.
000.
000.
00m
axco
ncT
OR
On-
road
Urb
an3
4.16
1.89
2.27
135
0.00
0.00
0.33
0.33
max
conc
2T
OR
All
All
8710
.34
1.72
8.63
484
0.06
0.16
0.23
0.40
max
conc
2T
OR
All
Rur
al29
4.51
1.65
2.85
189
0.07
0.28
0.38
0.55
max
conc
2T
OR
All
Urb
an58
13.2
61.
7511
.51
632
0.05
0.10
0.16
0.33
conc
near
TO
RH
All
All
882.
312.
230.
0813
0.11
0.26
0.60
0.84
conc
near
TO
RH
All
Rur
al30
2.04
1.83
0.21
210.
070.
130.
470.
77co
ncne
arT
OR
HA
llU
rban
582.
442.
430.
018
0.14
0.33
0.67
0.88
conc
near
TO
RH
Non
-roa
dA
ll18
4.01
2.36
1.65
780.
060.
110.
330.
50co
ncne
arT
OR
HN
on-r
oad
Rur
al9
3.60
2.19
1.41
640.
000.
110.
330.
56co
ncne
arT
OR
HN
on-r
oad
Urb
an9
4.41
2.52
1.89
930.
110.
110.
330.
44co
ncne
arT
OR
HO
n-ro
adA
ll5
1.76
2.51
-0.7
5-2
60.
400.
400.
801.
00co
ncne
arT
OR
HO
n-ro
adR
ural
22.
022.
34-0
.32
-90.
500.
501.
001.
00co
ncne
arT
OR
HO
n-ro
adU
rban
31.
582.
63-1
.05
-37
0.33
0.33
0.67
1.00
ICF
Con
sult
ing
53
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%co
ncne
ar2
TO
RH
All
All
881.
812.
23-0
.42
-12
0.08
0.22
0.52
0.92
conc
near
2T
OR
HA
llR
ural
301.
571.
83-0
.27
-60.
070.
170.
400.
87co
ncne
ar2
TO
RH
All
Urb
an58
1.93
2.43
-0.5
0-1
40.
090.
240.
590.
95m
axco
ncT
OR
HA
llA
ll87
14.5
72.
2512
.32
510
0.06
0.14
0.22
0.36
max
conc
TO
RH
All
Rur
al29
6.20
1.89
4.31
246
0.07
0.17
0.34
0.45
max
conc
TO
RH
All
Urb
an58
18.7
52.
4316
.32
643
0.05
0.12
0.16
0.31
max
conc
TO
RH
Non
-roa
dA
ll18
37.6
92.
3635
.33
1303
0.00
0.06
0.11
0.17
max
conc
TO
RH
Non
-roa
dR
ural
98.
652.
196.
4630
60.
000.
110.
220.
22m
axco
ncT
OR
HN
on-r
oad
Urb
an9
66.7
32.
5264
.21
2300
0.00
0.00
0.00
0.11
max
conc
TO
RH
On-
road
All
54.
362.
511.
8583
0.00
0.00
0.00
0.60
max
conc
TO
RH
On-
road
Rur
al2
4.68
2.34
2.34
104
0.00
0.00
0.00
0.50
max
conc
TO
RH
On-
road
Urb
an3
4.16
2.63
1.53
690.
000.
000.
000.
67m
axco
nc2
TO
RH
All
All
8710
.34
2.25
8.09
335
0.08
0.18
0.32
0.51
max
conc
2T
OR
HA
llR
ural
294.
511.
892.
6115
20.
170.
280.
410.
62m
axco
nc2
TO
RH
All
Urb
an58
13.2
62.
4310
.83
426
0.03
0.14
0.28
0.45
conc
near
TO
RL
All
All
882.
311.
191.
1211
00.
100.
260.
410.
65co
ncne
arT
OR
LA
llR
ural
302.
041.
080.
9610
50.
130.
400.
530.
67co
ncne
arT
OR
LA
llU
rban
582.
441.
241.
2011
20.
090.
190.
340.
64co
ncne
arT
OR
LN
on-r
oad
All
184.
011.
292.
7222
80.
110.
220.
280.
33co
ncne
arT
OR
LN
on-r
oad
Rur
al9
3.60
1.29
2.31
178
0.11
0.22
0.33
0.33
conc
near
TO
RL
Non
-roa
dU
rban
94.
411.
293.
1227
80.
110.
220.
220.
33co
ncne
arT
OR
LO
n-ro
adA
ll5
1.76
1.36
0.40
360.
000.
000.
400.
80co
ncne
arT
OR
LO
n-ro
adR
ural
22.
021.
380.
6454
0.00
0.00
0.50
1.00
conc
near
TO
RL
On-
road
Urb
an3
1.58
1.34
0.24
240.
000.
000.
330.
67co
ncne
ar2
TO
RL
All
All
881.
811.
190.
6265
0.14
0.31
0.52
0.74
conc
near
2T
OR
LA
llR
ural
301.
571.
080.
4959
0.10
0.30
0.50
0.73
conc
near
2T
OR
LA
llU
rban
581.
931.
240.
6968
0.16
0.31
0.53
0.74
max
conc
TO
RL
All
All
8714
.57
1.20
13.3
710
660.
010.
030.
070.
13m
axco
ncT
OR
LA
llR
ural
296.
201.
125.
0848
70.
030.
030.
070.
14m
axco
ncT
OR
LA
llU
rban
5818
.75
1.24
17.5
113
550.
000.
030.
070.
12m
axco
ncT
OR
LN
on-r
oad
All
1837
.69
1.29
36.4
025
960.
060.
060.
060.
06m
axco
ncT
OR
LN
on-r
oad
Rur
al9
8.65
1.29
7.36
589
0.11
0.11
0.11
0.11
ICF
Con
sult
ing
54
Tab
le 4
-5. S
umm
ary
of d
iffer
ence
s an
d pe
rcen
tage
diff
eren
ces
betw
een
mod
eled
aga
inst
mon
itore
d D
PM
con
cent
ratio
ns.
Fra
ctio
n o
f M
od
eled
Val
ues
Wit
hin
Mo
del
ed V
aria
ble
1M
on
ito
red
Var
iab
le2S
ub
set3
Lo
cati
on
N
Mea
nM
od
eled
Val
ue
Mea
nM
on
ito
red
Val
ue
Mea
nD
iffe
ren
ceM
ean
% D
iffe
ren
ce10
%25
%50
%10
0%m
axco
ncT
OR
LN
on-r
oad
Urb
an9
66.7
31.
2965
.44
4602
0.00
0.00
0.00
0.00
max
conc
TO
RL
On-
road
All
54.
361.
363.
0123
70.
000.
200.
200.
20m
axco
ncT
OR
LO
n-ro
adR
ural
24.
681.
383.
3024
60.
000.
000.
000.
00m
axco
ncT
OR
LO
n-ro
adU
rban
34.
161.
342.
8123
10.
000.
330.
330.
33m
axco
nc2
TO
RL
All
All
8710
.34
1.20
9.14
730
0.02
0.03
0.11
0.24
max
conc
2T
OR
LA
llR
ural
294.
511.
123.
3932
80.
030.
030.
140.
34m
axco
nc2
TO
RL
All
Urb
an58
13.2
61.
2412
.02
931
0.02
0.03
0.10
0.19
Not
es o
n ne
xt p
age.
ICF
Con
sult
ing
55
Not
es f
or T
able
4-5
:
5.
Mod
eled
var
iabl
e:co
ncne
arN
eare
st m
odel
ed D
PM c
once
ntra
tion
cons
iste
nt w
ith th
e dr
aft 2
000
NO
NR
OA
D M
odel
conc
near
2N
eare
st m
odel
ed D
PM c
once
ntra
tion
cons
iste
nt w
ith th
e dr
aft 2
002
NO
NR
OA
D M
odel
max
conc
Nea
rby
(with
in 3
0 km
) m
axim
um m
odel
ed D
PM c
once
ntra
tion
cons
iste
nt w
ith th
e dr
aft 2
000
NO
NR
OA
DM
odel
max
conc
2N
earb
y (w
ithin
30
km)
max
imum
mod
eled
DPM
con
cent
ratio
n co
nsis
tent
with
the
draf
t 200
2 N
ON
RO
AD
Mod
el
6.
Mon
itore
d va
riab
le:
EC
TO
RE
C v
alue
mul
tipl
ied
by T
OR
ave
rage
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
RH
EC
val
ue m
ultip
lied
by T
OR
max
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
RL
EC
val
ue m
ultip
lied
by T
OR
min
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
T).
EC
TO
TE
C v
alue
mul
tipl
ied
by T
OT
ave
rage
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
EC
TO
TH
EC
val
ue m
ultip
lied
by T
OT
max
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
EC
TO
TL
EC
val
ue m
ultip
lied
by T
OR
min
imum
cor
rect
ion
fact
or (
mis
sing
for
EC
mea
sure
d us
ing
TO
R).
TO
RE
CO
CX
val
ue m
ultip
lied
by T
OT
ave
rage
cor
rect
ion
fact
or f
or E
PA d
ata,
EC
TO
R f
or T
OR
dat
a.T
OR
HE
CO
CX
val
ue m
ultip
lied
by T
OT
max
imum
cor
rect
ion
fact
or f
or E
PA d
ata,
EC
TO
R f
or T
OR
dat
a..
TO
RL
EC
OC
X v
alue
mul
tiplie
d by
TO
R m
inim
um c
orre
ctio
n fa
ctor
for
EPA
dat
a, E
CT
OR
for
TO
R d
ata.
7.
Subs
et:
All
All
sit
esN
on-r
oad
Site
s do
min
ated
by
Non
-roa
d so
urce
(at
leas
t 75
% o
f m
odel
ed D
PM c
onsi
sten
t with
the
draf
t 200
0 N
ON
RO
AD
Mod
elO
n-ro
adSi
tes
dom
inat
ed b
y O
n-ro
ad s
ourc
e (a
t lea
st 5
0 %
of
mod
eled
DPM
con
sist
ent w
ith th
e dr
aft 2
000
NO
NR
OA
DM
odel
ICF
Con
sult
ing
56
ICF
Con
sult
ing
57
ICF
Con
sult
ing
58
ICF
Con
sult
ing
59
ICF
Con
sult
ing
60
ICF
Con
sult
ing
61
ICF Consulting 62
REFERENCES
Air Improvement Resources. (1997). Contribution of Gasoline Powered Vehicles to AmbientLevels of Fine Particulate Matter. CRC Project A-18.
Allen, G. A., Lawrence, J., and Koutrakis, P. (1999). Field Validation of a Semi-ContinuousMethod for Aerosol Black Carbon (Aethalometer) and temporal Patterns of Summertime HourlyBlack Carbon Measurements in Southwestern P. A. Atmospheric Environment. 33 (5), pp. 817-823.
Cass, G. R. (1997). Contribution of Vehicle Emissions to Ambient Carbonaceous ParticulateMatter: A Review and Synthesis of the Available Data in the South Coast Air Basin. CRC ProjectA-18.
Chow, J. C. and Watson, J. G. (1998). Guideline on Speciated Particulate Monitoring. DesertResearch Institute.
Chow, J. C., Watson, J. G., Pritchett, L. C., Pierson, W. R., Frazier, C. A., and Purcell, R. G.(1993). The DRI Thermal/Optical Reflectance Carbon Analysis System: Description, Evaluation,and Applications in U.S. Air Quality Studies. Atmospheric Environment. Vol 27A, No. 8, pp.1185 –1201.
Hansen, A. D. A. (2000). The Aethalometer. Magee Scientific Company, Berkeley, California,USA.
Hansen, A. D. A, and McMurry, P. H. (1990). An intercomparison of measurements of aerosolelemental carbon during the 1986 carbonaceous species method comparison study. J. Air &Waste Manage. Assoc. 40, pp. 394-395.
Liousse, C., Cashier, H., and Jennings, S. G. (1993). Optical and Thermal Measurements ofBlack Carbon Aerosol Content in Different Environments: Variation of the Specific AttenuationCross Section, Sigma (σ).Atmospheric Environment. Vol 27A, No. 8, pp. 1203 –1211.
Ramadan, Z., Song, X-H, and Hopke, P. K. (2000). Identification of Sources of Phoenix Aerosolby Positive Matrix Factorization. J. Air & Waste Manage. Assoc. 50, pp. 1308-1320.
Schauer, J. J., and Cass, G. R. (2000). Source Apportionment of Wintertime Gas-Phase andParticle Phase Air Pollutants Using Organic Compounds as Tracers. Environmenal Science andTechnology. Vol 34, No. 9, pp. 1821 –1832.
Schauer, J. J., Rogge, W. F., Hildemann, L. M., Mazurek, M. A., Cass, G. R., and Simoneit, B.R. T. (1996). Source Apportionment of Airborne PM Using Organic Compounds as Tracers.Atmospheric Environment. Vol 30, No. 22, pp. 3837 –3855.
ICF Consulting 63
Watson, J. G., Fujita, E., Chow, J. G., Zielinska, B., Richards, L. W., Neff, W., and Dietrich, D.(1998). Northern Front Range Air Quality Study Final Report. Desert Research Institute. 6580-685-8750.1F2.
Zheng, M., Cass, G. R., Schauer, J. J., and Edgerton, E. S. (2002). Source Apportionment ofPM2.5 in the Southeastern United States Using Solvent-Extractable Organic Compounds asTracers. Environmental Science and Technology. To appear.