Trends in Source Impacts at Long- Running PM2.5 CSN...
39
Trends in Source Impacts at Long- Running PM 2.5 CSN Monitoring Sites in the Pacific Northwest and Intermountain West Robert Kotchenruther NW-AIRQUEST Meeting June 11-13, 2019
Transcript of Trends in Source Impacts at Long- Running PM2.5 CSN...
Trends in Source Impacts at Long-Running PM2.5 CSN Monitoring
Sites in the Pacific Northwest and Intermountain West
Robert KotchenrutherNW-AIRQUEST Meeting
June 11-13, 2019
Motivation:
➢ Some PM2.5 monitors have a long record of chemically speciated data.
➢ These data can be used for source apportionment modeling (e.g., PMF)
➢ We can use source apportionment results to look at changes in source impacts over time.
➢ Some sources have a long history of being targeted for emissions reductions (e.g., residential wood combustion, sulfur in fossil fuels)
➢ ??? How successful have we been?
➢ ??? Are there any unexpected results?
This Approach:
Data:1) Review currently active sites that collect chemically speciated PM2.5 data.2) Select sites that are up to date in data submissions to EPA’s AQS system (data
through summer 2018)3) Select sites that have a data record going back to when EPA switched carbon
measurement methods to match the IMPROVE methodology (2007-2009).
Modeling:1) Run sites’ data in PMF and get source apportionment results (each site run
independently)2) Analyze year to year changes in resulting sources.
Note:• For this presentation I focus on year to year results for winter months Nov –
Feb (but all data was used in modeling)• A ‘winter year’ is Nov – Feb. So, for example, ‘2015 winter’ is Nov-Dec 2015
and Jan-Feb 2016
How does the PMF source apportionment model work?
The model looks for systematic patterns in the day-to-day chemical variations and quantifies a smaller set of ‘factors’ that can explain the overall data variability.
ChemicallySpeciated
PM2.5
Data(> ~100
samples)
PMFModel
PM2.5 Factor/Pattern 1
PM2.5 Factor/Pattern 2
PM2.5 Factor/Pattern 3
PM2.5 Factor/Pattern 4
Pattern Interpretation
soil dust
secondary SO4
secondary NO3
wood combustion
PMF Model Output
7 Sites Analyzed
What were the common types of chemical patterns / factors / sources found from PMF analysis?
Factor / PatternBoise,
IDButte,
MTPortland,
ORBountiful,
UTSeattle, WA (Beason Hill)
Tacoma, WA (South L St.)
Yakima, WA
Fresh Wood Smoke ✓ ✓ ✓ ✓ ✓ ✓ ✓
Aged Wood Smoke ✓ ✓ ✓ ✓ ✓ ✓ ✓
Soil / Dust ✓ ✓ ✓ ✓ ✓ ✓
Gas Engines ✓ ✓ ✓ ✓ ✓ ✓
Ammonium Sulfate ✓ ✓ ✓ ✓ ✓
Ammonium Nitrate ✓ ✓ ✓ ✓
Diesel Engines ✓ ✓ ✓ ✓
Sulfate Dominant ✓ ✓ ✓
Sea Salt ✓ ✓ ✓
Residual Oil Combustion ✓ ✓
Nitrate Dominant ✓ ✓
Industrial (Sulfate & Metals) ✓
Mixed ✓ ✓
Format of results analysis - Example of Yakima WAChanges in winter (Nov-Feb) PM2.5* from 2007-2017.*only using PM2.5 data coincident with CSN monitoring
Yakima Winter PM2.5 Yakima Winter PM2.5, top 25%
2007-2009 2015-2017
Mean (ug/m3)
Stdev (ug/m3)
Median (ug/m3)
Mean (ug/m3)
Stdev (ug/m3)
Median (ug/m3)
WMW# p-value ( significant?)
% Change in Mean
Yakima Winter PM2.5 16.0 9.8 14.5 12.6 8.1 10.9 0.050 -21.3
Yakima Winter PM2.5 (top 25%) 28.6 7.6 25.9 22.0 8.5 18.8 0.003 -23.0
Comparing 2007-2009 with 2015-2017
2007-2009 2015-2017 2007-2009 2015-2017
0
50
PM
2.5
(u
g/m
3)
0
50
PM
2.5
(u
g/m
3)
#WMW = Wilcoxon-Mann-Whitney significance test (non-parametric)
Changes in winter (Nov – Feb) PM2.5
2007-2009 to 2015 - 2017
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
ha
ng
e in
Me
an
PM
2.5
200
7-2
009
to
20
15
-201
7
-100
-80
-60
-40
-20
0
20 % Change in Mean
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
% C
ha
ng
e in
Me
an
PM
2.5
[to
p 2
5%
of ye
arly d
ata
]
200
7-2
009
to
20
15
-201
7Change in 3-year average winter PM2.5* [2007-2009 to 2015-2017]*only using PM2.5 data coincident with CSN monitoring
8.9 12.3 18.6 10.5 6.4 13.8 16.0 Beginning 2007-2009 average winter PM2.5 mass (ug/m3)
Change in 3-year average winter PM2.5* [2007-2009 to 2015-2017] for top 25% of yearly data.*only using PM2.5 data coincident with CSN monitoring
17.5 25.0 30.1 19.8 12.0 28.6 28.6Beginning 2007-2009 average winter PM2.5 mass (ug/m3) for top 25% of yearly data.
All Winter CSN Data
Top 25% Winter CSN Data
Results for Winter Wood Smoke PM2.5
(2007-2009 to 2015 – 2017)
AL
CA
FE SI
TI
BR
CL
NA
+M
G
CR
CU
PB
ZN
MN V NI
NH
4+
NO
3-
SO
4--
NS
S KO
MC
1(1
.4)
OM
C2(1
.4)
OM
C3(1
.4)
OM
C4(1
.4)
OM
P(1
.4)
EC
1
EC
2E
C3
1E-4
0.001
0.01
0.1
1 Fresh Wood Smoke Factor (12 site average)
Fra
ction
al C
on
trib
utio
n
AL
CA
FE SI
TI
BR
CL
NA
+M
G
CR
CU
PB
ZN
MN V NI
NH
4+
NO
3-
SO
4--
NS
S KO
MC
1(1
.4)
OM
C2(1
.4)
OM
C3(1
.4)
OM
C4(1
.4)
OM
P(1
.4)
EC
1
EC
2E
C3
1E-4
0.001
0.01
0.1
1 Aged Wood Smoke and Secondary Organic Carbon
Factor (11 site average)
Fra
ction
al C
on
trib
utio
n
How was Wood Smoke Identified in the Chemical Data?(Figures from R.A. Kotchenruther / Atmospheric Environment 142 (2016) 210-219)
Fresh wood smoke• OC and EC dominate• OC components shifted to lower boiling
fractions • K contribution• Cl contribution
Aged wood smoke• OC and EC dominate• OC components shifted to higher boiling
fractions (more oxidative processing)• Higher OC/EC ratio then fresh smoke• K contribution• No Cl contribution
In winter, fresh + aged = RWC In summer, fresh + aged = wild, Rx, Ag fire & SOA
Boxplot examples, winter wood smoke PM2.5 (fresh + aged)
Yakima, WA
Bountiful, UTTacoma, WA
0 0
0
15
7.5
20
PM
2.5
(u
g/m
3)
PM
2.5
(u
g/m
3)
PM
2.5
(u
g/m
3)
Change in 3-year average winter PM2.5 from wood smoke (fresh + aged)[2007-2009 to 2015-2017]
Percent contribution of beginning 2007-2009 winter PM2.5 mass
Change in 3-year average winter PM2.5 from wood smoke (fresh + aged)[2007-2009 to 2015-2017] for top 25% of yearly data
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
ha
ng
e in
Me
an
PM
2.5
200
7-2
009
to
20
15
-201
7
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
% Change in Mean
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
% Change in Mean
(Stat. Sig. at 95% conf.)
-100
-80
-60
-40
-20
0
20
% Change in Mean
% C
ha
ng
e in
Me
an
PM
2.5
[to
p 2
5%
of ye
arly d
ata
]
200
7-2
009
to
20
15
-201
7
46 20 68 42 25 57 50
39 15 82 49 25 68 52
Percent contribution of beginning 2007-2009 winter PM2.5 mass (top 25% of yearly data)
All Winter Data
Top 25% Winter Data
Results for Winter Ammonium Nitrate PM2.5
(2007-2009 to 2015 – 2017)
How was Ammonium Nitrate Identified in the Chemical Data?(Figure from R.A. Kotchenruther / Atmospheric Environment 142 (2016) 210-219)
AL
CA
FE SI
TI
BR
CL
NA
+M
G
CR
CU
PB
ZN
MN V NI
NH
4+
NO
3-
SO
4--
NS
S KO
MC
1(1
.4)
OM
C2(1
.4)
OM
C3(1
.4)
OM
C4(1
.4)
OM
P(1
.4)
EC
1
EC
2E
C3
1E-4
0.001
0.01
0.1
1 Ammonium Nitrate Factor
(10 site average)
Fra
ction
al C
on
trib
utio
n
Boxplot examples, winter ammonium nitrate PM2.5
Yakima, WA
Boise, ID Bountiful, UT
0
PM
2.5
(u
g/m
3)
10
0
0
15
10
PM
2.5
(u
g/m
3)
PM
2.5
(u
g/m
3)
Change in 3-year average winter PM2.5 from ammonium nitrate[2007-2009 to 2015-2017]
Percent contribution of beginning 2007-2009 winter PM2.5 mass
Change in 3-year average winter PM2.5 from ammonium nitrate[2007-2009 to 2015-2017] for top 25% of yearly data
Percent contribution of beginning 2007-2009 winter PM2.5 mass (top 25% of yearly data)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
ha
ng
e in
Me
an
PM
2.5
200
7-2
009
to
20
15
-201
7
-100
-80
-60
-40
-20
0
20
% Change in Mean
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
% Change in Mean
% C
ha
ng
e in
Me
an
PM
2.5
[to
p 2
5%
of ye
arly d
ata
]
200
7-2
009
to
20
15
-201
7
37 51 12 34
41 60 12 37
All Winter Data
Top 25% Winter Data
Results for Winter Ammonium Sulfate PM2.5
(2007-2009 to 2015 – 2017)
How was Ammonium Sulfate Identified in the Chemical Data?(Figure from R.A. Kotchenruther / Atmospheric Environment 142 (2016) 210-219)
AL
CA
FE SI
TI
BR
CL
NA
+M
G
CR
CU
PB
ZN
MN V NI
NH
4+
NO
3-
SO
4--
NS
S KO
MC
1(1
.4)
OM
C2(1
.4)
OM
C3(1
.4)
OM
C4(1
.4)
OM
P(1
.4)
EC
1
EC
2E
C3
1E-4
0.001
0.01
0.1
1 Ammonium Sulfate Factor
(6 site average)
Fra
ctional C
ontr
ibutio
n
Boxplot examples, winter ammonium sulfate PM2.5
Yakima, WA
Boise, ID Portland, OR
0
PM
2.5
(u
g/m
3)
43
1
PM
2.5
(u
g/m
3)
PM
2.5
(u
g/m
3)
0
0
Change in 3-year average winter PM2.5 from ammonium sulfate[2007-2009 to 2015-2017]
Percent contribution of beginning 2007-2009 winter PM2.5 mass
Change in 3-year average winter PM2.5 from ammonium sulfate[2007-2009 to 2015-2017] for top 25% of yearly data
Percent contribution of beginning 2007-2009 winter PM2.5 mass (top 25% of yearly data)
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
ha
ng
e in
Me
an
PM
2.5
200
7-2
009
to
20
15
-201
7
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
% Change in Mean
(Stat. Sig. at 95% conf.)
-100
-80
-60
-40
-20
0
20
% Change in Mean
% C
ha
ng
e in
Me
an
PM
2.5
[to
p 2
5%
of ye
arly d
ata
]
200
7-2
009
to
20
15
-201
7
12 7 4 6 3
13 8 4 6 2
All Winter Data
Top 25% Winter Data
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
hange in M
ean P
M2
.5
2007-2
009 to 2
015-2
017
-100
-80
-60
-40
-20
0
20 % Change in Mean
Change in 3-year average winter PM2.5 [2007-2009 to 2015-2017]
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
hange in M
ean P
M2
.5
2007-2
009 to 2
015-2
017
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
% Change in Mean
46 20 68 42 25 57 50
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
hange in M
ean P
M2
.5
2007-2
009 to 2
015-2
017
-100
-80
-60
-40
-20
0
20
% Change in Mean
37 51 12 34
-100
-80
-60
-40
-20
0
20 % Change in Mean
(Stat. Sig. at 95% conf.)
% C
hange in M
ean P
M2
.5
2007-2
009 to 2
015-2
017
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20 % Change in Mean
12 7 4 6 3
Wood Smoke (fresh + aged)
Ammonium Nitrate
Ammonium Sulfate
Many important contributors to PM2.5 going down, why not total PM2.5 so much?
One chemical pattern in the PM2.5 data has been increasing. What is it?
AL
CA
FE SI
TI
BR
CL
NA
+M
G
CR
CU
PB
ZN
MN V NI
NH
4+
NO
3-
SO
4--
NS
S KO
MC
1(1
.4)
OM
C2(1
.4)
OM
C3(1
.4)
OM
C4(1
.4)
OM
P(1
.4)
EC
1
EC
2E
C3
1E-5
1E-4
0.001
0.01
0.1
1 CSN Network Factor (11 site average)
Fra
ction
al C
ontr
ibutio
n
AL
CA
FE SI
TI
NA
CL-
AS
BR
CR
CU
PB
MG
MN NI V P
SE
SR
ZN
Am
NO
3A
mS
O4 K
OM
C1(1
.8)
OM
C2(1
.8)
OM
C3(1
.8)
OM
C4(1
.8)
OM
P(1
.8)
EC
1E
C2
EC
3
1E-5
1E-4
0.001
0.01
0.1
1 IMPROVE Network Factor (6 site average)
Fra
ction
al C
ontr
ibutio
n
(Figure from R.A. Kotchenruther / Atmospheric Environment 142 (2016) 210-219) [19 sites analyzed]
(Figure from R.A. Kotchenruther / Atmospheric Environment 151 (2017) 52-61) [22 sites analyzed]
What do we know?• Ubiquitous, appears in PMF analysis at every
CSN and IMPROVE site (either on own or mixed with other patterns/sources)
• OC & EC dominate, -> some kind of fuel combustion.
• Not associated with wood smoke factors.• Trace metals, Fe, Ti, Cu, Zn.
• IMPROVE factor, OC shifted to higher thermal evolution components (OC4 & OP) -> Likely more atmospheric oxidative processing (more aged from source)
• IMPROVE factor, fewer sites where factor clearly resolved -> further form source
➢ My take on what source fits this pattern …
Gas Vehicles
Boxplot examples, winter gas vehicles PM2.5
Yakima, WA
Tacoma, WA Bountiful, ID
0
PM
2.5
(u
g/m
3)
9
0
PM
2.5
(u
g/m
3)
10
0
PM
2.5
(u
g/m
3)
5
Change in 3-year average winter PM2.5 from gas vehicles[2007-2009 to 2015-2017]
Percent contribution of beginning 2007-2009 winter PM2.5 mass
Change in 3-year average winter PM2.5 from gas vehicles[2007-2009 to 2015-2017] for top 25% of yearly data
Percent contribution of beginning 2007-2009 winter PM2.5 mass (top 25% of yearly data)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
0
100
250260270280290300
% Change in Mean
(Stat. Sig. at 95% conf.)
% C
hange in M
ean P
M2
.5
2007-2
009 to 2
015-2
017
0
10
20
30
40
50
60
70
80
90
100
250260270280290300
% Change in Mean
0
100
200
300600
700
800
% Change in Mean
(Stat. Sig. at 95% conf.)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
0
100
200
300600
700
800 % Change in Mean
% C
hange in M
ean P
M2
.5 [to
p 2
5%
of yearly d
ata
]
2007-2
009 to 2
015-2
017
15 12 29 36 19 10
11 6 25 31 9 3
What have other source apportionment efforts found?
Multi-site PMF analysis of New York State sites from 2005 – 2016
Squizzato et al., A long-term source apportionment of PM2.5 in New York State during 2005–2016 Atmospheric Environment 192 (2018) 35–47Masiol et al., Long-term trends (2005–2016) of source apportioned PM2.5 across New York State Atmospheric Environment 201 (2019) 110–120
Similar story
Masiol et al., 2019 [from abstract]“Spark-ignition vehicles were the only source type experiencing upward annual trends at all urban sites with slopes ranging from 0.02 μg/m3/y to ~0.2 μg/m3/y …”
Same factor / pattern as identified in this work
Masiol et al., 2019 [from supplemental]
Squizzato et al., 2018 [from supplemental]
PMF chemical pattern / factor for ‘Spark–ignition’
Why is this chemical pattern / source increasing?
Masiol et al., 2019 attribute increases in this factor to:• a similar patter in increases in registered vehicles • gas vehicle emissions are a significant source of secondary organic
aerosol precursors
Some questions about the gas vehicle pattern …• Why such a large increase in Yakima? • Can it really all be attributed to increasing vehicle population?• Why is ammonium nitrate going down in Yakima (52% reduction
in mean from 2007-2009 to 2015-2017), if this pattern is going up?(i.e., if most of NOx is from vehicles)(note: 2007-2009 to 2015-2017 mean monitored NO3 down only 17%, mean monitored NH4 down 58%, mean SO4 down 47%)
• Maybe there is a shift in winter nitrate chemistry, away from ammonium nitrate production, and towards organic nitrate production?
Your thoughts?
Thank You!
Questions?
Supplementary Slides
Yakima winter bulk NO3 Yakima winter bulk NH4
Yakima winter bulk SO4
2007-2009 mean = 3.1 ug/m3 2015-2017 mean = 2.5 ug/m3% change = -17%wmw p-value = 0.07
2007-2009 mean = 0.68 ug/m3 2015-2017 mean = 0.36 ug/m3% change = -47%wmw p-value = <0.01
2007-2009 mean = 1.13 ug/m3 2015-2017 mean = 0.47 ug/m3% change = -58%wmw p-value = <0.01
Change in Yakima winter monitored NO3, NH4, and SO4
Results for Winter Soil Dust PM2.5
(2007-2009 to 2015 – 2017)
How was Soil Dust Identified in the Chemical Data?(Figure from R.A. Kotchenruther / Atmospheric Environment 142 (2016) 210-219)
AL
CA
FE SI
TI
BR
CL
NA
+M
G
CR
CU
PB
ZN
MN V NI
NH
4+
NO
3-
SO
4--
NS
S KO
MC
1(1
.4)
OM
C2
(1.4
)O
MC
3(1
.4)
OM
C4
(1.4
)
OM
P(1
.4)
EC
1
EC
2E
C3
1E-4
0.001
0.01
0.1
1 Soil Dust Factor (10 site average)
Fra
ctional C
ontr
ibution
Boxplot examples, winter soil dust PM2.5
Yakima, WA
Seattle, WA Boise, ID
0
PM
2.5
(u
g/m
3)
1
0
PM
2.5
(u
g/m
3) 1
0
PM
2.5
(u
g/m
3)
1
Change in 3-year average winter PM2.5 from soil dust[2007-2009 to 2015-2017]
Percent contribution of beginning 2007-2009 winter PM2.5 mass
Change in 3-year average winter PM2.5 from soil dust[2007-2009 to 2015-2017] for top 25% of yearly data
Percent contribution of beginning 2007-2009 winter PM2.5 mass (top 25% of yearly data)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
40
60
% Change in Mean
(Stat. Sig. at 95% conf.)
% C
ha
ng
e in
Me
an
PM
2.5
200
7-2
009
to
20
15
-201
7
-100
-80
-60
-40
-20
0
20
40
60 % Change in Mean
-100
-80
-60
-40
-20
0
20
40
60
% Change in Mean
(Stat. Sig. at 95% conf.)
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
40
60 % Change in Mean
% C
hange in M
ean P
M2
.5 [to
p 2
5%
of yearly d
ata
]
2007-2
009 to 2
015-2
017
3 4 4 1 4 3
2 3 2 1 4 2
All Winter Data
Top 25% Winter Data
Results for Winter Diesel Engine PM2.5
(2007-2009 to 2015 – 2017)
How were Diesel Engines Identified in the Chemical Data?
SO4 + EC2
Al
Ca
Fe Si
Ti
Na+ Cl
Br
Cr
Cu
Zn
NH
4
NO
3
NS
S
SO
4 K
OC
1
OC
2
OC
3
OC
4
OP
EC
1
EC
2
EC
3
0
20
40
60
80
100
Al
Ca
Fe Si
Ti
Na+ Cl
Br
Cr
Cu
Zn
NH
4
NO
3
NS
S
SO
4 K
OC
1
OC
2
OC
3
OC
4
OP
EC
1
EC
2
EC
3
1E-5
1E-4
0.001
0.01
0.1
1
Fra
ctional C
ontr
ibution
Boise, Diesel
Specie
s P
erc
ent (%
)
Al
Ca
Fe Si
Na+ Cl
Br
Cr
Cu
Mn
Zn
NH
4
NO
3
NS
S
SO
4 K
OC
1
OC
2
OC
3
OC
4
OP
EC
1
EC
2
EC
3
0
20
40
60
80
100
Al
Ca
Fe Si
Na+ Cl
Br
Cr
Cu
Mn
Zn
NH
4
NO
3
NS
S
SO
4 K
OC
1
OC
2
OC
3
OC
4
OP
EC
1
EC
2
EC
3
1E-5
1E-4
0.001
0.01
0.1
1
Fra
ctional C
ontr
ibution
Butte, Diesel
Specie
s P
erc
ent (%
)
Al
Ca
Fe Si
Na+ Cl
Br
Cu
Mg
Zn
NH
4
NO
3
SO
4 K
OC
1
OC
2
OC
3
OC
4
OP
EC
1
EC
2
0
20
40
60
80
100
Al
Ca
Fe Si
Na+ Cl
Br
Cu
Mg
Zn
NH
4
NO
3
SO
4 K
OC
1
OC
2
OC
3
OC
4
OP
EC
1
EC
2
1E-5
1E-4
0.001
0.01
0.1
1
Fra
ctional C
ontr
ibution
Bountiful, Diesel
Specie
s P
erc
ent (%
)
SO4EC2
SO4
SO4
EC2
EC2
Boxplot examples, winter diesel engine PM2.5
Seattle, WA
Boise, ID Bountiful, UT
0
PM
2.5
(u
g/m
3)
1
1
2
PM
2.5
(u
g/m
3)
PM
2.5
(u
g/m
3)
0
0
Change in 3-year average winter PM2.5 from diesel engines[2007-2009 to 2015-2017]
Percent contribution of beginning 2007-2009 winter PM2.5 mass
Change in 3-year average winter PM2.5 from diesel engines[2007-2009 to 2015-2017] for top 25% of yearly data
Percent contribution of beginning 2007-2009 winter PM2.5 mass (top 25% of yearly data)
All Winter Data
Top 25% Winter Data
-100
-80
-60
-40
-20
0
20
40
% Change in Mean
(Stat. Sig. at 95% conf.)
% C
hange in M
ean P
M2
.5
2007-2
009 to 2
015-2
017
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
40
% Change in Mean
Bo
ise
Bo
un
tifu
l
Bu
tte
Po
rtla
nd
Sea
ttle
Ta
co
ma
Yak
ima
-100
-80
-60
-40
-20
0
20
40
% Change in Mean
(Stat. Sig. at 95% conf.)
-100
-80
-60
-40
-20
0
20
40
% Change in Mean
% C
hange in M
ean P
M2
.5 [to
p 2
5%
of yearly d
ata
]
2007-2
009 to 2
015-2
017
4 2 1 8
2 0.3 0.4 7
