Optical Measurements Optical Measurements of CSN and FRM Teflon Filtersof CSN and FRM Teflon Filtersto Estimate Elemental Carbonto Estimate Elemental Carbon

to Support Health Studies, PM to Support Health Studies, PM NAAQS Implementation and ClimateNAAQS Implementation and Climate

Neil FrankNeil FrankJoann RiceJoann RiceJoe Joe TikvartTikvart

US EPA / OAQPSUS EPA / OAQPS

For Presentation at NAMC in Nashville, TN November 4, 2009

2

FindingsFindingsMany Benefits of Optical MeasurementsMany Benefits of Optical Measurements

Inexpensive utilization of Inexpensive utilization of archivedarchived PM2.5 filtersPM2.5 filtersCan predict EC by measuring the darkness of the filtersCan predict EC by measuring the darkness of the filtersBetter and more consistent relationship with new Better and more consistent relationship with new IMPROVE carbon methodIMPROVE carbon methodOpportunity to cost effectively expand our current Opportunity to cost effectively expand our current ““ECEC”” monitoring network formonitoring network for

Health researchersHealth researchersClimateClimate--related purposesrelated purposesEmission and modeling efforts Emission and modeling efforts QA of speciation carbonQA of speciation carbon

3

BackgroundBackground

Optical measurements of filters used for decadesOptical measurements of filters used for decadesIMPROVE has been doing optical measurements for a long time (forIMPROVE has been doing optical measurements for a long time (for QC QC of thermal EC), using HIPSof thermal EC), using HIPS““British SmokeBritish Smoke”” with with ReflectometerReflectometer widely used in Europewidely used in Europe

Why is EPA now interested?Why is EPA now interested?EPA is transitioning the OCEPA is transitioning the OC--EC protocol from the NIOSHEC protocol from the NIOSH--type method type method to IMPROVE protocol (started May 2007)to IMPROVE protocol (started May 2007)Health community expressed Health community expressed

concern about possibleconcern about possiblediscontinuity in EPAdiscontinuity in EPA’’s EC s EC time series and recommended time series and recommended optical measurements optical measurements to estimate EC.to estimate EC.

(EPA (EPA ““Health Research Health Research WorkshopWorkshop”” –– April 2008)April 2008) Mostly May-June data

New EC ~1.2*Old EC(network median)

4

Other MotivationsOther MotivationsLack of temporal and spatial detail in EPA Lack of temporal and spatial detail in EPA speciation networks speciation networks

EC has sharp gradients in urban areasEC has sharp gradients in urban areasEveryday EC and more spatial resolution neededEveryday EC and more spatial resolution needed

Support Health Studies and PM2.5 NAAQS Support Health Studies and PM2.5 NAAQS ImplementationImplementation

There are 940 PMThere are 940 PM2.52.5 FRM samplers which can be FRM samplers which can be used to estimate ECused to estimate EC

5

Optical measurements of FRM teflon filters

can help increase the “EC”spatial resolution

NYC Tri-State area

Elizabeth NJS.Bronx NY

40 FRM sites forpotential expansion of “EC” measurements

Existing Speciation DataIncluding EC

The Tri-State Area has 7 sites sampling every three days. More sites and daily sampling would be better!

and, provide daily EC at PM2.5 sites which sample every day!

6

More motivations:More motivations:to better understand to better understand ““black carbonblack carbon”” for climatefor climate

Last week, President Signed Appropriation Bill Last week, President Signed Appropriation Bill (HR2996)(HR2996)

"Not later than 18 months "Not later than 18 months ……. the Administrator. the Administrator…… shall carry shall carry out and submit to Congress the results of a out and submit to Congress the results of a study on domestic study on domestic and international black carbon emissionsand international black carbon emissions that shall include an that shall include an inventory of the major sources of black carbon, an assessment ofinventory of the major sources of black carbon, an assessment ofthe impacts of black carbon on global and regional climate, an the impacts of black carbon on global and regional climate, an assessment of potential metrics and approaches for quantifying assessment of potential metrics and approaches for quantifying the climatic effects of black carbon emissions the climatic effects of black carbon emissions ……......““

EPA may need to define EPA may need to define ““BCBC”” and must understand and must understand differences between optical and thermal methodsdifferences between optical and thermal methods

7

Optical Measurements highly correlate with Optical Measurements highly correlate with ““ECEC””

““SmokeSmoke StainStainReflectometerReflectometer”” recently recently used with Teflon Filtersused with Teflon FiltersCyrysCyrys et. al., 2003et. al., 2003

““SiteSite--specific calibration specific calibration with EC is necessary.with EC is necessary.””

Thurston et al.Thurston et al.““South Bronx South Bronx Environmental Health Environmental Health Policy StudyPolicy Study”” (2007)(2007)

Sometimes EC Sometimes EC calibrated optical calibrated optical measurements reported measurements reported as as ““BCBC””

Thurston(summer-time

Cyrys

8

Recent OAQPS WorkRecent OAQPS WorkDRI study of Teflon filters (2007)*DRI study of Teflon filters (2007)*

To produce PM2.5 composition at 20 FRM To produce PM2.5 composition at 20 FRM sites without speciation datasites without speciation data““ECEC”” using densitometer at 3 FRMusing densitometer at 3 FRM--CSN sitesCSN sites

* 2007 Study is posted on the PM2.5 NAAQS Designations Web Sitehttp://www.epa.gov/ttn/naaqs/pm/docs/available_new_speciation_data_pm2.5_naa.pdf

Comparison filter carbon measurements at the Fresno SupersiteJohn G. Watson and Judith C. Chow (2002)

Linear up to 5 ug/m3 EC

9

EC vs FRM Babs (Densitometeslope FAC

Milwaukee 0.0441 11.3sacramento 0.0475 10.5toledo 0.0485 10.3

Is the 550nm densitometer detecting brown carbon?

Milwau

kee (

r=0.9)

and T

oledo

(r=0

.8)

Sacram

ento

(r=0.9

)

• 35 high 2004-06 FRM concentrations days• EC is measured with the EPA NIOSH-type protocol

PILOT densitometer study (2007)

Similar filter absorption coefficients, but unexplained offset in slope

10

Study of Magee (OT21) Study of Magee (OT21) TransmissometerTransmissometerTeflon filters from 7 CSN sites (6.7 LPM)Teflon filters from 7 CSN sites (6.7 LPM)

Representing different emission influencesRepresenting different emission influences4 sites ( )which switched TOA protocol in May 20074 sites ( )which switched TOA protocol in May 2007

To permit comparisons to new EC_TOR and old EC_CSNTo permit comparisons to new EC_TOR and old EC_CSN1in6 day filters were selected per site from 2007 sampling1in6 day filters were selected per site from 2007 samplingSubset of 3 sites with collocated FRM filters (16.7 LPM)Subset of 3 sites with collocated FRM filters (16.7 LPM)

Tacoma

Sacramento

Birmingham

Atlanta

Gary Bronx

Elizabeth, NJ

11

Joe Tikvart measured > 500 filters CSN Teflon FiltersThis is a non-destructive procedure

12

EC

0

1

2

3

4

5

6

7

8

9

10

ATN, IR

-10 0 10 20 30 40 50 60 70

CSN filters - all 7 sites - 4 are for the full year

NIOSH-type EC vs. Optical Measurements (7 sites)

ATN = 100* ln ( Io / I)(precision = ± 1)

r = 0.89 (n=300)E

C, u

g/m

3 (O

ld C

SN

Met

hod)

2 Wood Smoke sites

(Sacramento and Tacoma)5 Non-

Wood Smoke sites

Optical Measurement (Attenuation units, IR channel)

The OP21 measures at 2 wavelenths (370 and 880nm). I only present results for the infra-red channel.

13

ec_old

0

1

2

3

4

ATN IR

-10 0 10 20 30 40 50 60 70 80 90 100

AIRS Site Code=530530029

ec_old

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

SITE=130890002

Regression Equation:ec_old = -0.140738 + 0.111328*ATN_IR

ec_old

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

SITE=010730023

Regression Equation: ec_old = -0.012905 + 0.122323*ATN_IR

ec_old

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

SITE=060670006

Regression Equation:ec_old = 0.274317 + 0.067532*ATN_IR

ec_old

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

SITE=180890022

Regression Equation: ec_old = 0.203157 + 0.103142*ATN_IR

ec_old

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

SITE=340390004

Regression Equation:ec_old = -0.000646 + 0.107857*ATN_IR

AtlantaBirmingham

Sacramento

Tacoma

Gary

Optical attenuation vs. Optical attenuation vs. ““old ECold EC”” varies by sitevaries by siteWood smoke areas are different!Wood smoke areas are different!

Elizabeth

Magee Transmissometer Study

Similar site-specific relationships in EastStrong correlations mostly > 0.9A few “outliers” to be checked

OPTICAL, ATN

EC

, Old

(ug/

m3)

14

ec_tor_birm

0

1

2

3

4

5

ATN IR

-10 0 10 20 30 40 50 60 70 80 90 100

PLOT ec_tor_birm ec_tor_Sacr ec_tor_Tac ec_tor_Gary ec_tor_Newk ec_tor_Atl ec_tor_Brx

r=0.93 (n=174)

ec_tor = 0.234184 + 0.100823*ATN_IR

EC

, ug/

m3

(IMP

RO

VE

Met

hod)

Optical Measurement, Attenuation units (IR channel)

IMPROVE EC vs. Filter Attenuation (4 CSN sites)

Magee Transmissometer Study

Very similar relationship (±10%)among all 4 sites including Tacoma

Simple Conversion:IMPROVE EC= ~ ATN/10 !!

15

EC

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

AIRS Site Code=010730023

Regression Equation: ec_tor = 0.228174 + 0.099811*ATN_IR

EC

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

AIRS Site Code=180890022

Regression Equation:ec_tor = 0.30591 + 0.093595*ATN_IR

EC

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

AIRS Site Code=360050110

Regression Equation: ec_tor = 0.312954 + 0.093389*ATN_IR

EC

0

1

2

3

4

5

6

7

ATN, IR

-10 0 10 20 30 40 50 60 70 80 90 100

AIRS Site Code=530530029

Regression Equation:ec_tor = 0.158918 + 0.118316*ATN_IR

IMPROVE EC vs. Optical ATN (similar for each site)

Birmingham Gary

Bronx Tacoma

16

birminghamEC

-1

0

1

2

3

4

5

6

7

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

AIRS Site Code=010730023 MSA_NAME=Birmingham,AL

PLOT zero Elemental Carbon

EC

-1

0

1

2

3

4

5

6

7

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

AIRS Site Code=010730023 MSA_NAME=Birmingham,AL

PLOT zero Elemental Carbon atn_EC

Old EC New EC

GA Fire (May 24)

Potential blankTeflon filter?

Birmingham

Magee Transmissometer Study

Red = PM2.5>35 ug/m3

(r=0.91) (r=0.95)

17

EC

-1

0

1

2

3

4

5

6

7

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

AIRS Site Code=130890002 MSA_NAME=Atlanta,GA

PLOT zero Elemental Carbon

EC

-1

0

1

2

3

4

5

6

7

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

AIRS Site Code=130890002 MSA_NAME=Atlanta,GA

PLOT zero Elemental Carbon atn_EC

Atlanta

Potential EC Measurement error?ON Nov 8, Measured EC=6.0 ug/m3CSN mass=17.6, FRM mass=17.1, OC*=10.4

Magee Transmissometer Study

OC* is adjusted for CSN sampling artifact (i.e. blank corrected)

(r=0.82)

EPA NIOSH-type EC

Now adding results Now adding results for the 3 collocated FRM sitesfor the 3 collocated FRM sites

We will normalize CSN and FRM filter We will normalize CSN and FRM filter attenuation for their different sampler attenuation for their different sampler

flow rates and 24flow rates and 24--hr sample volumeshr sample volumes

19

Elem

enta

l Car

bon,

ug/

m3

0

1

2

3

4

5

6

ABS, (A/V*ATN_IR)

-10 0 10 20 30 40 50 60

Elem

enta

l Car

bon,

ug/

m3

0

1

2

3

4

5

ABS, (A/V*ATN_IR)

-10 0 10 20 30 40 50 60

Elem

enta

l Car

bon,

ug/

m3

0

1

2

3

4

5

ATN_IR

-10 0 10 20 30 40 50 60 70 80 90

PLOT gary_c_ec_tor brnx_c_ec_tor tacm_c_ec_tor gary_F_ec_tor brnx_F_ec_tor tacm_F_ec_tor

Elem

enta

l Car

bon,

ug/

m3

0

1

2

3

4

5

6

ATN_IR

-10 0 10 20 30 40 50 60 70 80 90

PLOT gary_c_ec_old brnx_c_ec_old tacm_c_ec_oldgary_F_ec_old brnx_F_ec_old tacm_F_ec_old

Fabs=(A/V) x ATNATN

Fabs=(A/V) x ATNATN

(a) “NIOSH-type” ECvs. ATN

(c) IMPROVE ECvs. ATN

(b) “NIOSH-type” ECvs. Fabs

FRMr = 0.84(n=86)

(d) IMPROVE ECvs. Fabs

FRMr = 0.96(n=98)

CSNr = 0.66(n=148)

CSNr = 0.93(n=174)

(A/V)0.5(FRM)1.2 (CSN)

20

EC, u

g/m

3

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

ABS_IR

-10 0 10 20 30 40 50

PLOT ec_tor2_tacm ec_tor2_brnx ec_tor2_gary

3 FRM sites, IMPROVE EC vs. Optical MeasurementsFilter absorption (Fabs) is much more precise than thermal EC

IMPROVE EC = ~ ATN/20 = ~ Fabs /10

Possible non-linearity, showing potential saturation effects at higher EC concentrations, say > 5 ug/m3

-5.5 -4.5 -3.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 3.5 4.5 5.5 6.5

ATN IR

Fabs = 0.2±1100 field blanks

21

EC= 0.16 + Fabs_IR/11.5

EC_unc= 0.085 *Log(EC) + 0.254 (Mark Green, DRI)

3 FRM sites, IMPROVE EC vs. Fabs_IRWeighted Least Squares Regression

22

OLS Regression: EC= 0+ Fabs_IR/10(SAS “Proc Model” uses MLE)

3 FRM sites, IMPROVE EC vs Fabs_IR

23

Parameter Estimate Std Err t Value Pr > |t|a 0.120029 0.1021 1.18 0.2488b 24.37738 2.3294 10.46 <.0001

R2=0.7793(Gary)

Parameter Estimate Std Err t Value Pr > |t|a 0.032818 0.0672 0.49 0.6285b 20.88619 0.9048 23.08 <.0001

R2 = 0.9417 (Bronx)

Parameter Estimate Std Err t Value Pr > |t|a 0.069648 0.0374 1.86 0.0732b 20.82608 0.7487 27.82 <.0001

R2=0.9651(Tacoma)

Site Specific Relationships

These coeff not normalized for Volume

24

These graphs show that predicted new EC with single equation tracks the IMPROVE measurements among sites

Fabs/10 can help identify potential outliers in measured EC.

Estimated EC using FRM Filter Absorption with single national equation, EC=Fabs/10

EC, u

g/m

3

0

1

2

3

4

5

6

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

Predicted EC using Attenuation of FRM filters and single national equationAIRS Site Code=530530029

EC_ratio = predicted IMPROVE protocol EC / measured EC

EC, u

g/m

3

0

1

2

3

4

5

6

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

Predicted EC using Attenuation of FRM filters and single national equationAIRS Site Code=180890022

EC_ratio = predicted IMPROVE protocol EC / measured EC

EC, u

g/m

3

0

1

2

3

4

5

6

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

Predicted EC using Attenuation of FRM filters and single national equationAIRS Site Code=360050110

EC_ratio = predicted IMPROVE protocol EC / measured EC

Old EC New (IMPROVE)EC

25

Urban Fe & IMPROVE EC CorrelationsGary, r=0.47Bronx, r=0.82Tacoma, r=0.90

ec_old

0

1

2

3

4

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

fe

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

FRM filtersAIRS Site Code=180890022

PLOT ec_old Elemental Carbon TOR ec10s

ec_old

0

1

2

3

4

5

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

fe

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

0.24

0.26

0.28

0.30

0.32

0.34

0.36

0.38

FRM filtersAIRS Site Code=360050110

PLOT ec_old Elemental Carbon TOR ec10s

ec_old

0

1

2

3

4

5

Date

010107 020107 030107 040107 050107 060107 070107 080107 090107 100107 110107 120107 010108

fe

0.00

0.01

0.020.03

0.04

0.05

0.06

0.07

0.08

0.090.10

0.11

0.12

0.13

0.14

0.150.16

0.17

0.18

0.19

0.20

FRM filtersAIRS Site Code=530530029

PLOT ec_old Elemental Carbon TOR ec10s

Gary (r=0.47)

Bronx Tacoma (0.90)(r=0.82)

Old EC Fe

0.9 ug/m3 Fe

0.4 ug/m3 Fe 0.2 ug/m3 Fe

EC IMPROVE

With typical mobile dominated urbanemissions

EC is not the only light absorbing aerosol

Iron can also be important !It may affect measured EC

Gary

Bronx Tacoma

26

PM2.5 Attenuation Coefficients (Teflon Filters)PM2.5 Attenuation Coefficients (Teflon Filters)using Magee using Magee TransmissometerTransmissometer (880nm)(880nm)

1010--12 12 mm22gg--111313--16 16 m2g-1EPA EPA ““NIOSHNIOSH--TypeType”” (TOT)(TOT)

1010--11 11 mm22gg--111010--11 11 m2g-1IMPROVE_A IMPROVE_A (TOR)(TOR)

NonNon--WSWSWood Wood SmokeSmoke

CSN / FRM NetworksCSN / FRM NetworksCarbonCarbon

ProtocolProtocol

These numbers for teflon filters suggest aerosol and carbon-protocol specific differences in PM2.5 absorption

27

Summary and Next StepsSummary and Next Steps

Optical filter absorption is dependent on many factorsOptical filter absorption is dependent on many factorsWavelengthWavelengthThermal Optical ProtocolThermal Optical ProtocolFilter material (Filter material (teflonteflon vsvs quartz) quartz) Fe concentrations Fe concentrations

warrants further study warrants further study

These measurements can help estimate EC without These measurements can help estimate EC without speciation sampling and help QA thermal OCspeciation sampling and help QA thermal OC--ECECEPA is pursuing analysis of archived filters and EPA is pursuing analysis of archived filters and potential incorporation into CSN and FRM networkspotential incorporation into CSN and FRM networks

Can help provide daily EC and higher spatial resolutionCan help provide daily EC and higher spatial resolutionWill be closing coordinating with State/local agenciesWill be closing coordinating with State/local agencies

28

We can’t do this without YOU!

Thanks to the State and local monitoring agencieswho let us examine theirFilters:

••Indiana Department of Indiana Department of Environmental ManagementEnvironmental Management••New York State Department of New York State Department of Environmental ConservationEnvironmental Conservation••Puget Sound Clean Air AgencyPuget Sound Clean Air Agency

It is It is veryvery important for agenciesimportant for agenciesto continue to archive your filtersto continue to archive your filters