Source Contribution to PMSource Contribution to PM2.52.5 and and

Visibility Impairment in Two Class I Visibility Impairment in Two Class I Areas Using Positive Matrix Areas Using Positive Matrix

FactorizationFactorization

Keith RoseKeith Rose

EPA, Region 10EPA, Region 10

June 22, 2005June 22, 2005

Introduction Introduction

PMF was used to analyze IMPROVE data from PMF was used to analyze IMPROVE data from 1991-95 and 2000-03 time periods for Mt. Rainier 1991-95 and 2000-03 time periods for Mt. Rainier and Yosemite National Park Class I Areasand Yosemite National Park Class I Areas

PMF identified six source profiles and the time-PMF identified six source profiles and the time-dependent contributions of these sources to PMdependent contributions of these sources to PM2.52.5 concentrations in these Class I Areasconcentrations in these Class I Areas

PMF source concentrations were converted to PMF source concentrations were converted to light extinction to determine the visibility light extinction to determine the visibility impairment due to each sourceimpairment due to each source

The PMF ModelThe PMF Model

PMF is a form of factor analysis with non-negative PMF is a form of factor analysis with non-negative factor elementsfactor elements

PMF looks for co-variation of species over time to PMF looks for co-variation of species over time to identify source profiles and source time-dependent identify source profiles and source time-dependent contributions which best fit the datacontributions which best fit the data

PMF uses data uncertainty to “weight” the data PMF uses data uncertainty to “weight” the data (i.e. greater uncertainty = less weight)(i.e. greater uncertainty = less weight)

The PMF Model The PMF Model (continued)(continued)

PMF works best with a large (>300) number of PMF works best with a large (>300) number of samples and several metal tracer speciessamples and several metal tracer species

The appropriate number of sources can be The appropriate number of sources can be determined by analyzing the data first with another determined by analyzing the data first with another model (UNMIX), or comparing source profiles with model (UNMIX), or comparing source profiles with previously identified profilespreviously identified profiles

The model was run in the “robust” mode in the The model was run in the “robust” mode in the “heuristically-computed” error mode“heuristically-computed” error mode

Data SelectionData Selection

IMPROVE data from 1991-95 and 2000-03IMPROVE data from 1991-95 and 2000-03 Species with a substantial number (>50%) of Species with a substantial number (>50%) of

values below the MDLs were eliminatedvalues below the MDLs were eliminated Data reported as “0” were replaced by ½ MDLData reported as “0” were replaced by ½ MDL Species used in this analysis included: Ca, Cu, Species used in this analysis included: Ca, Cu,

EC1-2, Fe, K, H, Na, Pb, OC2-4, NO3, S, SO4, Si EC1-2, Fe, K, H, Na, Pb, OC2-4, NO3, S, SO4, Si and Zand Z

Number of samples used: Mt. Rainier – 476/435 Number of samples used: Mt. Rainier – 476/435 and Yosemite – 473/421and Yosemite – 473/421

Determining the Number of SourcesDetermining the Number of Sources

Used the UNMIX model to identify the number of Used the UNMIX model to identify the number of sources and source profiles for Mt. Rainiersources and source profiles for Mt. Rainier

Compared PMF source profiles with identified Compared PMF source profiles with identified source profiles from Seattle and Columbia Gorgesource profiles from Seattle and Columbia Gorge

Examined “goodness of fit” of linear regression Examined “goodness of fit” of linear regression between measured and calculated source between measured and calculated source concentrationsconcentrations

Sources and Average Percent Contribution to PMSources and Average Percent Contribution to PM2.52.5

Concentrations at Mt. RainierConcentrations at Mt. Rainier (1998-2001 data)(1998-2001 data)

SourceSource PMFPMF UNMIXUNMIX

BiomassBiomass 38.938.9 43.143.1

SulfateSulfate 31.631.6 31.131.1

NitrateNitrate 8.68.6 11.511.5

MobileMobile 4.84.8 4.74.7

SoilSoil 9.49.4 5.75.7

MarineMarine 6.76.7 4.04.0

Definition of Sources Definition of Sources

Biomass = wood stoves/fireplaces, agricultural Biomass = wood stoves/fireplaces, agricultural burning, wildfires, prescribed burning burning, wildfires, prescribed burning

Mobile = on-road and off-road gasoline and diesel Mobile = on-road and off-road gasoline and diesel powered mobile vehiclespowered mobile vehicles

Secondary sulfate = diesel fuel, home heating fuel, Secondary sulfate = diesel fuel, home heating fuel, pulp mills, oil refineriespulp mills, oil refineries

Secondary nitrate = oil refineries, commercial Secondary nitrate = oil refineries, commercial boilers, power plants, on and off-road mobile boilers, power plants, on and off-road mobile sources (50-66%?) sources (50-66%?)

ResultsResults

Mt. Rainier and Yosemite source profilesMt. Rainier and Yosemite source profiles Time-dependent concentrations for each Time-dependent concentrations for each

sourcesource Average source light extinction (Bext)Average source light extinction (Bext) Light extinction by source for 1991-95 and Light extinction by source for 1991-95 and

2000-03 time periods2000-03 time periods 20% worst vs. 20% best days for 200220% worst vs. 20% best days for 2002

Biomass Burning Profile - Mt. Rainier 1991-95

0

5

10

15

20

25

30

35

40

Species

Per

cen

t C

on

cen

trat

ion

Biomass Burning Profile - Yosemite 1991-95

0

5

10

15

20

25

30

35

Species

Per

cen

t C

on

cen

trat

ion

Secondary Sulfate Profile - Mt. Rainier 1991-95

0

10

20

30

40

50

60

70

Species

Per

cen

t C

on

cen

trat

ion

Secondary Sulfate Profile - Yosemite 1991-95

0

10

20

30

40

50

60

70

Species

Per

cen

t C

on

cen

trat

ion

Secondary Nitrate Profile - Mt. Rainier 1991-95

0

5

10

15

20

25

30

35

40

45

50

Species

Per

cen

t C

on

cen

trat

ion

Secondary Nitrate Profile - Yosemite 1991-95

0

10

20

30

40

50

60

70

Species

Per

cen

t C

on

cen

trat

ion

Mobile Source Profile - Mt. Rainier 1991-95

0

5

10

15

20

25

30

35

Species

Per

cen

t C

on

cen

trat

ion

Mobile Source Profile - Yosemite 1991-95

0

5

10

15

20

25

30

35

Species

Per

cen

t C

on

cen

trat

ion

Soil Profile - Yosemite 1991-95

0

5

10

15

20

25

30

35

Species

Per

cen

t C

on

cen

trat

ion

Biomass Source Trend - Mt. Rainier 1991-95

0

2

4

6

8

10

12

1/5/91

4/5/91

7/4/91

10/2/91

12/31/91

3/30/92

6/28/92

9/26/92

12/25/92

3/25/93

6/23/93

9/21/93

12/20/93

3/20/94

6/18/94

9/16/94

12/15/94

3/15/95

6/13/95

9/11/95

12/10/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Secondary Sulfate Trend - Mt. Rainier 1991-95

0

1

2

3

4

5

6

7

8

9

10

1/5/91

4/5/91

7/4/91

10/2/91

12/31/91

3/30/92

6/28/92

9/26/92

12/25/92

3/25/93

6/23/93

9/21/93

12/20/93

3/20/94

6/18/94

9/16/94

12/15/94

3/15/95

6/13/95

9/11/95

12/10/95

Dates

Co

nce

ntr

atio

ns

(ug

/m3)

Secondary Nitrate Trend - Mt. Rainier 1991-95

0

0.5

1

1.5

2

2.5

3

3.5

1/5/91

4/5/91

7/4/91

10/2/91

12/31/91

3/30/92

6/28/92

9/26/92

12/25/92

3/25/93

6/23/93

9/21/93

12/20/93

3/20/94

6/18/94

9/16/94

12/15/94

3/15/95

6/13/95

9/11/95

12/10/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Mobile Source Trend - Mt. Rainier 1991-95

0

0.5

1

1.5

2

2.5

3

3.5

1/5/91

4/5/91

7/4/91

10/2/91

12/31/91

3/30/92

6/28/92

9/26/92

12/25/92

3/25/93

6/23/93

9/21/93

12/20/93

3/20/94

6/18/94

9/16/94

12/15/94

3/15/95

6/13/95

9/11/95

12/10/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Soil Trend - Mt. Rainier 1991-95

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1/5/91

4/5/91

7/4/91

10/2/91

12/31/91

3/30/92

6/28/92

9/26/92

12/25/92

3/25/93

6/23/93

9/21/93

12/20/93

3/20/94

6/18/94

9/16/94

12/15/94

3/15/95

6/13/95

9/11/95

12/10/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Biomass Source Trend - Yosemite 1991-95

0

2

4

6

8

10

12

14

1/2/91

4/2/91

7/1/91

9/29/91

12/28/91

3/27/92

6/25/92

9/23/92

12/22/92

3/22/93

6/20/93

9/18/93

12/17/93

3/17/94

6/15/94

9/13/94

12/12/94

3/12/95

6/10/95

9/8/95

12/7/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Secondary Sulfate Trend - Yosemite 1991-95

0

1

2

3

4

5

6

1/2/91

4/2/91

7/1/91

9/29/91

12/28/91

3/27/92

6/25/92

9/23/92

12/22/92

3/22/93

6/20/93

9/18/93

12/17/93

3/17/94

6/15/94

9/13/94

12/12/94

3/12/95

6/10/95

9/8/95

12/7/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Secondary Nitrate Trend - Yosemite 1991-95

0

1

2

3

4

5

6

1/2/91

4/2/91

7/1/91

9/29/91

12/28/91

3/27/92

6/25/92

9/23/92

12/22/92

3/22/93

6/20/93

9/18/93

12/17/93

3/17/94

6/15/94

9/13/94

12/12/94

3/12/95

6/10/95

9/8/95

12/7/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Mobile Source Trend - Yosemite 1991-95

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1/2/91

4/2/91

7/1/91

9/29/91

12/28/91

3/27/92

6/25/92

9/23/92

12/22/92

3/22/93

6/20/93

9/18/93

12/17/93

3/17/94

6/15/94

9/13/94

12/12/94

3/12/95

6/10/95

9/8/95

12/7/95

Dates

Co

nce

ntr

atio

n (

ug

/m3)

Soil Trend - Yosemite 1991-95

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1/2/91

4/2/91

7/1/91

9/29/91

12/28/91

3/27/92

6/25/92

9/23/92

12/22/92

3/22/93

6/20/93

9/18/93

12/17/93

3/17/94

6/15/94

9/13/94

12/12/94

3/12/95

6/10/95

9/8/95

12/7/95

Dates

Co

nce

ntr

atio

ns

(ug

/m3)

Light Extinction CalculationLight Extinction Calculation

Bext = (3mBext = (3m22/g)Ft(RH)[sulfate] + /g)Ft(RH)[sulfate] + (3m(3m22/g)Ft(RH)[nitrate] + (4m/g)Ft(RH)[nitrate] + (4m22/g)[OC] + /g)[OC] + (10m(10m22/g)[EC] + (1m/g)[EC] + (1m22/g)[soil]/g)[soil]

Ft(RH) = annual average relative humidity factorFt(RH) = annual average relative humidity factor

Average Light Extinction by Source at Yosemite

0

2

4

6

8

10

12

Biomass Sulfate Nitrate Mobile Soil

Sources

Bex

t (1

/Mm

)

1991-95 2000-03

Average Light Extinction by Source at Mt. Rainier

0

2

4

6

8

10

12

14

16

18

Biomass Sulfate Nitrate Mobile Soil

Sources

Bex

t (1

/Mm

)

1991-95 2000-03

Light Extinction of 20% Worst and 20% Best Days - Mt. Rainier 2002

0

5

10

15

20

25

Biomass Sulfate Nitrate Mobile Soil

Sources

Bex

t (1

/Mm

)

20% Worst Days 20% Best Days

Light Extinction of 20% Worst and 20% Best Days -Yosemite 2002

0

5

10

15

20

25

30

35

40

45

Biomass Sulfate Nitrate Mobile Soil

Sources

Bex

t (1

/Mm

)

20% Worst Days 20% Best Days

ConclusionsConclusions

At Mt. Rainier, the highest Bext is due to secondary sulfate At Mt. Rainier, the highest Bext is due to secondary sulfate and the second highest is due to biomass burning and the second highest is due to biomass burning

At Yosemite, the highest Bext is due to biomass burning At Yosemite, the highest Bext is due to biomass burning and the second highest is to secondary sulfate and the second highest is to secondary sulfate

At Mt. Rainier, Bext from biomass, sulfate, nitrate and At Mt. Rainier, Bext from biomass, sulfate, nitrate and mobile sources decreased between 1991-95 and 2000-03mobile sources decreased between 1991-95 and 2000-03

At Yosemite, average Bext from biomass and mobile At Yosemite, average Bext from biomass and mobile sources increased, while secondary sulfate and nitrate sources increased, while secondary sulfate and nitrate sources decreased between 1991-95 and 2000-03sources decreased between 1991-95 and 2000-03

Conclusions (continued)Conclusions (continued)

Bext of the 20% worst days at Mt. Rainer in 2002 Bext of the 20% worst days at Mt. Rainer in 2002 was due to a combination of secondary sulfate was due to a combination of secondary sulfate (44%) and biomass burning (27%)(44%) and biomass burning (27%)

Bext of the 20% worst days at Yosemite in 2002 Bext of the 20% worst days at Yosemite in 2002 was dominated by biomass burning (66%)was dominated by biomass burning (66%)

PMF is a valuable tool in determining source PMF is a valuable tool in determining source contribution to visibility impairment, and changes contribution to visibility impairment, and changes in visibility impairment over time in Class I Areasin visibility impairment over time in Class I Areas