Sources of PM2.5 Carbon in the SE U.S.

RPO National Work Group MeetingDecember 3-4, 2002

Outline

• Background

• Simplified Source Matrix Concept

• CMB Analysis of Primary Carbon

• Carbon-14 Analyses

• Emission Factor Approach

Carbon is an Important (often dominant) Component of PM2.5 in

the Southeast

SEARCH Measurements Discrete Particles (24-hour)

– FRM: PM2.5– PCM and Dichot: PM2.5, PM10 and Speciation

Continuous Particles (1-min to 1-hr)– TEOM: mass– R&P 5400: OC/EC– Ammonium/Nitrate– Sulfate– Total Reduced Nitrogen -- Ammonia by Difference (under develop.)

Trace Gases (1-min)– O3, NO, NO2, NOy, HNO3, SO2, CO, CO2

Meteorology (1-min)– WS, WD, T, RH, BP, SR, rainfall

Visibility– Dry Extinction– Adsorption

Rural

Suburban

Urban

6.0

4.77.1

4.4

(3.6)

(5.0)

3.9

(3.9)

FRM Equivalent PM2.5 Organic Matter (ug/m3)10/98-9/01

Note: Project Year Begins 10/1/98, except for PNS, OLF, and OAK (in Parentheses) Begins 10/1/99.

Rural

Suburban

Urban

Best Estimate PM2.5 Composition 10/98-9/01

Details available at atmospheric-research.com

0.0

5.0

10.0

15.0

20.0

25.0

Jan

Feb

Mar

Ap

r

May Ju

n

Jul

Au

g

Sep Oct

No

v

Dec Q 1

Q 2

Q 3

Q 4

An

nu

al

Other

Major Metal Oxides

Organic Matter

Elemental Carbon

Ammonium

Nitrate

Sulfate

Best Estimate Mass

Best Estimate PM2.5 Composition Centreville, AL 2000

Units are µg/m3

0.0

5.0

10.0

15.0

20.0

25.0

Jan

Feb

Mar

Ap

r

May

Jun

Jul

Au

g

Sep Oct

No

v

Dec Q 1

Q 2

Q 3

Q 4

An

nu

al

Other

Major Metal Oxides

Organic Matter

Elemental Carbon

Ammonium

Nitrate

Sulfate

Best Estimate Mass

Best Estimate PM2.5 Composition Oak Grove, MS 2000

Units are µg/m3

Primary

Secondary

Total

Modern

Speciation/CMB*

Speciation, Models,

Difference

C-14

Fossil

Speciation/CMB*

Speciation, Models,

Difference

C-14

Simplified Organic Carbon Source Matrix

*e.g., Zheng et. al., ES&T, 2002; AAAR 2002.

PM2.5 Primary OC Source Identification

Investigators– Dr. Glen Cass– Ms. Mei Zheng

Description– Analyze one-month composite for each season at 8

SEARCH sites for primary organic carbon source tracers (> 100 species)

– Estimate fractional contribution using CMB

Source Contributions to OC in Fine Particles

Source Contributions to OC in Fine Particles

Salient Results of Primary OC Investigations

• Spatial coverage (SE U.S.) and temporal representativeness are limited (seasonal snapshots)– But growing!

• OC is virtually 100% primary in winter (all sites)

• Wood smoke >50% of OC in winter (all sites)– For Oak Grove > 50% winter, spring, fall

• OC is 30-70 primary in summer (all sites)

C-14 Measurement Strategy

• 24-hour or 72-hour quartz filter samples, 3 sites• Analyze OC/EC via TOR (DRI)• Analyze C-14 via accelerator mass spectrometry

(NOSAMS)• Yorkville, GA

– 7/9/01-8/5/01 (n=6)– 12/22/01-1/27/02 (n=13)

• Jefferson St., Atlanta, GA– 7/1/01-8/21/01 (n=12)– 11/13/01-1/19/02 (n=15)

• Oak Grove, MS– 11/19/01-2/26/02 (n=18)

Carbon-14 Measurement Sites

Oak Grove

Centreville

Pensacola

Yorkville

Jefferson St.

N.Birmingham

Gulfport

OLF

rural urban suburban

Carbon-14 and OC Data Atlanta, GA (JST)

0

4

8

12

16

20

7/1/

2001

7/4/

2001

7/7/

2001

7/10

/200

1

7/13

/200

1

7/16

/200

1

7/19

/200

1

7/22

/200

1

7/25

/200

1

7/28

/200

1

7/31

/200

1

8/21

/200

1

11/1

3/20

01

11/1

7/20

01

11/1

8/20

01

12/4

/200

1

12/5

/200

1

12/1

7/20

01

12/2

0/20

01

12/2

6/20

01

12/2

9/20

01

1/4/

2002

1/7/

2002

1/10

/200

2

1/13

/200

2

1/16

/200

2

1/19

/200

2

OC

(u

g/m

3)

0

0.2

0.4

0.6

0.8

1

F M

od

ern

OC F Modern

0.59 0.62

Carbon-14 and OC Data Yorkville, GA

0

2

4

6

8

10

7/9/

2001

7/12

/200

1

7/15

/200

1

7/18

/200

1

7/21

/200

1

8/5/

2001

12/2

2/20

01

12/2

6/20

01

12/2

8/20

01

12/3

1/20

01

1/3/

2002

1/6/

2002

1/9/

2002

1/12

/200

2

1/15

/200

2

1/18

/200

2

1/21

/200

2

1/24

/200

2

1/27

/200

2

OC

(u

g/m

3)

0.5

0.6

0.7

0.8

0.9

1.0

F M

od

ern

OC F Modern

0.83 0.85

Carbon-14 and OC Data Oak Grove, MS

0

2

4

6

8

10

11/1

9/20

01

11/2

2/20

01

11/2

5/20

01

1/15

/200

2

1/18

/200

2

1/21

/200

2

1/24

/200

2

1/27

/200

2

1/30

/200

2

2/2/

2002

2/5/

2002

2/8/

2002

2/11

/200

2

2/14

/200

2

2/17

/200

2

2/20

/200

2

2/23

/200

2

2/26

/200

2

OC

(u

g/m

3)

0.0

0.2

0.4

0.6

0.8

1.0

F M

od

ern

OC F Modern

0.780.96

Biomass Burning Events

Oak Grove, MS Biomass Burning Event

January 31, 2002 FRM Mass – 67.1 ug/m3

24-Hour TEOM Mass – 66.2 ug/m3

Anecdotal Reports of Wood Smoke

Oak Grove Event – 1/31/02

Gas and Particle data together are diagnostic of biomass burning

Continuous Speciation Data Suggest– 8% Black Carbon– <1% SO4– <1 % NO3– 2% NH4– >88% Organic Matter + Water

OC Source MatrixAtlanta, GA – July 2001

* Zheng et al. Source Apportionment of Fine Particles at Atlanta, GA, AAAR 2002

Primary*

Secondary

Total#

Modern

<5

>50

59 +/-5

Fossil

40*

<5

41 +/-5

# from C-14 data

Source Contributions of OC in Fine Particles, 1999

Zheng et al. ES&T, 2002.

100% Primary64% Modern

OC Source MatrixAtlanta, GA – January 2002

* 1999 Data: Zheng et al., ES&T, 2002.

Primary*

Secondary

Total#

Modern

64

<5

61 +/-5

Fossil

36

<5

39 +/-5

# from C-14 data

Source Contributions of OC in Fine Particles, 1999

Zheng et al. ES&T, 2002.

32% Primary16% Modern

OC Source MatrixYorkville, GA – July 2002

* 1999 data: Zheng et al. ES&T, 2002.

Primary*

Secondary

Total#

Modern

16

67

83 +/-6

Fossil

16

<5

17 +/-5

# from C-14 data

Source Contributions of OC in Fine Particles, 1999

Zheng et al. ES&T, 2002.

100% Primary86% Modern

OC Source MatrixOak Grove, MS – February 2002

* 1999 data: Zheng et al. ES&T, 2002.

Primary*

Secondary

Total#

Modern

86

<10

95 +/-5

Fossil

14

<5

5 +/-5

# from C-14 data

Emission Factor Approach

• Use difference between Urban and Biomass signatures to label and quantify source contributions of EC and TC

• Label sources using CO/NOy ratios (2 source assumption)

• Quantify contributions based on EC/CO and TC/CO ratios

Emission Factor (EF) Calculations

[x]event - [x]background

[tracer]event – [tracer]background EF =

where x is component of interest and tracer is CO.

EFs conveniently calculated by linear regression.

Example Total Carbon EF Atlanta SuperSite Experiment

TC = 11.3*CO + 5.3

R2 = 0.77

0

4

8

12

16

0 0.2 0.4 0.6 0.8

CO (ppm)

TC

(u

g/m

3)

Urban and Biomass EFs are Very Different

NOy/CO

(ppb/ppm)

PM2.5/CO

(ug-m-3/ppm)

TC/CO

(ug-m-3/ppm)

Urban 137 38 11

Biomass 23 220 69

Urban/Biomass

Ratio6.0 0.17 0.16

Emission Factor CalculationsCenterville, AL – CY2000

Month BioEC BioTC MobEC MobTC PredEC PredTC ObsEC ObsTC FModern

Jan-00 0.25 1.45 0.03 0.20 0.28 1.65 0.46 2.03 0.88Feb-00 0.86 4.99 0.08 0.53 0.94 5.52 0.85 4.05 0.90Mar-00 0.81 4.70 0.03 0.20 0.84 4.90 0.65 3.06 0.96Apr-00 0.44 2.55 0.03 0.20 0.47 2.75 0.47 1.98 0.93May-00 0.24 1.39 0.05 0.33 0.29 1.73 0.42 2.40 0.80Jun-00 0.23 1.33 0.04 0.27 0.27 1.60 0.53 2.73 0.83Jul-00 0.45 2.61 0.06 0.40 0.51 3.01 0.81 3.90 0.86Aug-00 0.44 2.55 0.03 0.20 0.47 2.75 0.90 3.77 0.93Sep-00 0.79 4.58 0.02 0.13 0.81 4.72 0.61 2.00 0.97Oct-00 1.45 8.41 0.06 0.40 1.51 8.81 1.27 4.39 0.95Nov-00 0.77 4.47 0.04 0.27 0.81 4.73 0.61 2.37 0.94Dec-00 0.82 4.76 0.04 0.27 0.86 5.02 0.63 3.23 0.95

EF calculations suggest EC and TC >80% Biogenic AND yield reasonable F-Modern, BUT tend to overpredict TC, esp. in winter

Summary & Conclusions

• Work in Progress, but results show convergence among techniques• Zheng and Cass primary carbon work shows almost all OC primary

in winter, 30-70 % primary in summer• Also show wood smoke makes up >50% OC in winter (detectable all

seasons)• C-14 data show OC is predominantly modern for all sites and all

seasons.• Urban site exhibits more fossil OC than rural sites.• No evidence of strong summer/winter seasonality in modern/fossil

OC, but an interesting step to >95% modern observed in MS during January 2002 (biomass burning?).

• Emission Factor calculations suggest EC and TC mostly biomass (>80%), but overpredict TC

• Combination of techniques permits semi-quantitative completion of source matrix