the impact of traffic on air quality in ireland: insights ...€¦ · 15 5department of civil,...
TRANSCRIPT
1
Supplementary for
The impact of traffic on air quality in Ireland: insights from the
simultaneous sampling of submicron aerosol at the kerbside and
residential site
5
Chunshui Lin1,2,3, Darius Ceburnis1, Wei Xu1,2, Eimear Heffernan4, Stig Hellebust4, John Gallagher5, Ru-
Jin Huang1,2,3*, Colin O’Dowd1*, and Jurgita Ovadnevaite1
1School of Physics, Ryan Institute’s Centre for Climate and Air Pollution Studies, National University of Ireland Galway.
University Road, Galway. H91 CF50, Ireland 2State Key Laboratory of Loess and Quaternary Geology and Key Laboratory of Aerosol Chemistry and Physics, Chinese 10
Academy of Sciences, 710061, Xi’an, China 3Center for Excellence in Quaternary Science and Global Change, Institute of Earth Environment, Chinese Academy of
Sciences, Xi’an 710061, China 4School of Chemistry and Environmental Research Institute, University College Cork, Cork, Ireland 5Department of Civil, Structural & Environmental Engineering, Trinity College Dublin, the University of Dublin, Ireland. 15
Correspondence to: Ru-Jin Huang ([email protected]) and Colin O’Dowd ([email protected])
2
Supplementary section of the chemical composition of PM1, meteorological parameters, and gases (NOx and CO)
20
Figure S1. Time series of submicron organic aerosol (OA), sulfate (SO4), nitrate (NO3), ammonium (NH4), chloride (Chl), and black
carbon (BC) at the kerbside (top panel), residential site (mid panel), and EPA Rathmines station from 4 September to 9 November
2018. P1, from 10 September to 17 September, was selected to represent non-heating periods while P2, from 27 October to 4
November 2018, was selected to represent the heating period. Note that the ACSM and BC data were hourly averaged while the time
resolution of PM2.5 measurement was 1 h. The gaps at the measurement was due to ACSM malfunction at TCD. 25
160
140
120
100
80
60
40
20
0
Mas
s co
nc.
(µ
g m
-3)
4/9 9/9 14/9 19/9 24/9 29/9 4/10 9/10 14/10 19/10 24/10 29/10 3/11 8/11
2018
160
140
120
100
80
60
40
20
0
Mas
s co
nc.
(µ
g m
-3)
160
140
120
100
80
60
40
20
0
Mas
s co
nc.
(µ
g m
-3)
PM2.5
(c) Rathmines
Org SO4
NO3
NH4
Chl BC
(a) Kerbside
(b) Residential
P1P2
3
Figure S2. Time series of wind direction (wd), wind speed (ws), relative humidity (RH), and temperature (T) during P1 (left panel)
and P2 (right panel). 30
Figure S3. Time series of the mixing ratio of NOx (ppbv) during P1 and P2 at the kerbside (a, c) and residential site (b, d). Also
shown are the day of the week, including Monday (Mon), Tuesday (Tue), Wednesday (Wed), Thursday (Thu), Friday (Fri), Saturday 35 (Sat), and Sunday (Sun).
360
270
180
90
0
wd
(o)
10/9 11/9 12/9 13/9 14/9 15/9 16/9 17/9
2018
15
10
5
0
ws (m
/s)
100
90
80
70
60
50
RH
(%
)
25
20
15
10
5
0
T (
oC)
Mon Tue Wed Thu Fri Sat Sun360
270
180
90
0
wd
(o)
27/10 28/10 29/10 30/10 31/10 1/11 2/11 3/11
2018
15
10
5
0
ws (m
/s)
100
90
80
70
60
50
RH
(%
)
15
10
5
0
-5
T (
oC)
Sat Sun Mon Tue Wed Thu Fri
P1: P2:
300
200
100
0
NO
x (
ppb)
10/9 11/9 12/9 13/9 14/9 15/9 16/9 17/9
2018
30
20
10
0
NO
x (
ppb)
Mon Tue Wed Thu Fri Sat Sun
(a) Kerbside
(b) Residential
500
400
300
200
100
0
NO
x (
ppb)
27/10 28/10 29/10 30/10 31/10 1/11 2/11 3/11
2018
200
150
100
50
0
NO
x (
ppb)
Sat Sun Mon Tue Wed Thu Fri
(c) Kerbside
(d) Residential
P1: P2:
4
Figure S4. Time series of the mixing ratio of CO (ppmv) at the kerbside during P1. Note that the CO measurement was not available
at the residential site. Also shown are the day of the week, including Monday (Mon), Tuesday (Tue), Wednesday (Wed), Thursday 40 (Thu), Friday (Fri), Saturday (Sat), and Sunday (Sun).
1.0
0.8
0.6
0.4
0.2
0.0
CO
(p
pm
)
10/9 11/9 12/9 13/9 14/9 15/9 16/9 17/9
2018
Mon Tue Wed Thu Fri Sat Sun
5
Supplementary section of OA source apportionment
OA factors at the kerbside during P1. 45
To resolve the OA sources, unconstrained PMF (or free PMF) was firstly conducted on the OA matrix with a range of solutions
from three to five factors. The solution that best represented the data was the three-factor solution because the solutions with
more factors provided no meaningful but splitting from the already resolved factors (Fig. S5). The three-factor solution
contained a hydrocarbon-like OA (HOA) factor from traffic, cooking-like OA (COA) factor from cooking sources, and an
oxygenated OA (OOA) factor corresponding to the secondary processes. 50
Although the three-factor solution from free PMF could reasonably represent the data, there was still a certain extent of
mixing between the factors because the cooking-like OA (COA) factor contained no signal at m/z 44 and a very low signal at
43 (Fig. S8) which was not physically meaningful (Mohr et al., 2012). To reduce the mixing between factors, the a value
approach within ME-2 (Canonaco et al., 2013) was utilized by constraining the reference profile of HOA and COA (Crippa et
al., 2013). A tight constraint or a small a value (i.e., 0.1) was applied for HOA because HOA was not expected to vary too 55
much as indicated by the small variation of HOA profiles obtained from different sites across the world (Ng et al., 2011a). In
contrast, COA could vary to some extent due to different styles of cooking and different types of food cooked (He et al., 2010;
Mohr et al., 2012). To explore the solution space, an a value of 0-0.5 with a step of 0.1 was used to constrain the reference
COA profile while keeping the a value of 0.1 constant for the HOA reference profile. The ME-2 results showed that the relative
contributions of HOA (26% - 29%), COA (35% - 39%), and OOA (36% - 37%) to the total OA varied only by a few percents 60
within the considered a values (Fig. S9), indicating the magnitudes (i.e., the a value of 0-0.5) of constraints on COA had a
relatively small effects on the apportionment results. However, larger a values correspond to higher freedom of the solution
space which led to potential mixing as the extreme case of a value of 1 which is equivalent to free PMF while extremely tight
constraint (e.g., a value of 0) resulted in large residues (Lin et al., 2017). Therefore, the solution with an a value of 0.1 was
chosen as the most appropriate one. 65
Figure S6 shows the mass spectra of HOA, COA, and OOA, as well as their diurnal patterns and relative contribution at a
value of 0.1. The HOA profile was dominated by peaks at m/z of 27, 29, 41, 43, 55, and 57, characteristic of the hydrocarbon
ion series of [CnH2n+1]+ and [CnH2n-1]+. The COA profile is characterized with a f55 to f57 ratio of 2.6 (where f55 and f57 are
the fraction of m/z 55 and 57 to the total organic mass, respectively) which was higher than that for HOA (0.9) but was in the
range of 2.2-2.8 typical found for COA (Mohr et al., 2012). The f55 vs. f57 plot (Fig. S9) provided further evidence that 70
downtown Dublin was strongly affected by cooking activities as many OA points lied on the left-hand side of the triangular
space defined by Mohr et al. (2012), consistent with the fact that there were many restaurants around the sampling site in
downtown Dublin. The OOA profile featured a high f44 value of 0.22 and did not contain significant contributions from marker
peaks of other organic aerosol classes such as at m/z 55 or 60. The OOA profiles resemble more to the less volatile OOA (LV-
OOA) which is often taken to represent well-aged SOA (Canonaco et al., 2013; Crippa et al., 2014). However, the diurnal 75
6
pattern of OOA in Dublin showed a clear pattern that was strongly influenced by local sources and was most likely from fresh
SOA instead of well-aged SOA.
OA factors at the kerbside during P2
The diurnal pattern of OA at the kerbside showed the evening OA peak was dominant during P2, suggesting large contributions
from residential heating as discussed in Sect. 3.2. Consistently, free PMF found a fourth factor associated with biomass burning 80
(i.e., BBOA) during P2, in addition to HOA, COA, and OOA (Fig. S10). The presence of BBOA factor was confirmed by the
elevated f60 (fraction of m/z 60 to total OA) of ~0.009 during the night (Fig. S11), as f60 is usually taken as a marker for
biomass burning especially for periods when f60 is above background level of ~0.003 (Alfarra et al., 2007; Cubison et al.,
2011).
Wood and peat are two types of biomass that were consumed by <5% of the households as the heating sources in Dublin 85
(CSO, 2016), with the rest ~95% households using electricity and natural gas which, however, were not expected to emit POA.
Moreover, the mass spectra of the emissions from wood and peat burning both contained f60 as found in our previous
combustion experiment (Lin et al., 2017). Free PMF failed to separate the peat factor from wood because they were co-emitted
from residential heating, demonstrating temporal covariation which was hard for free PMF to resolve (Canonaco et al., 2013;
Lin et al., 2017). To evaluate their contribution, ME-2 was utilized to constrain both peat and wood with a tight constraint (a 90
value of 0.1). Note that the evening emissions were local and relatively fresh, and thus their mass spectral profiles were not
expected to vary too much from the reference profiles retrieved from the combustion experiments (Lin et al., 2017). As for
coal burning, no mass spectral marker was available but the reference coal profile demonstrated higher fraction at high m/z’s
compared to HOA, peat, and wood (Lin et al., 2017). The five-factor solution from free PMF contained a factor profile with
higher distributions at high m/z’s and its diurnal showed peaks in the evening, suggesting a coal factor. Similarly, an a value 95
of 0.1 was applied for the coal factor.
The peat profile featured peaks at m/z of 27, 29, 41, 43, 55, and 57 (Fig. S12), which was similar to HOA. However, the
differences in f60 between the peat factor and HOA suggested different sources. f60 in the peat profile was 0.016 which was
higher than that for HOA (<0.003), confirming its biomass nature (Cubison et al., 2011). Note that peat is an accumulation of
partially decayed plant material which is harvested as an important source of fuel in Ireland (Tuohy et al., 2009). The 100
incomplete decay of vegetation resulted in an increase f60 when burned. However, f60 in peat profile was lower than wood
(0.072) as wood is relatively fresh vegetation, containing a higher fraction of m/z 60-related material e.g., levoglucosan (Alfarra
et al., 2007; Lin et al., 2017).
7
105
Figure S5. Mass spectral profiles (left panels) and diurnal (mid panels), as well as the relative contribution of the free PMF three-
(top) and four- (bottom) factor solutions at the kerbside during P1. The four-factor solution shows a HOA splitting factor because
its diurnal pattern was similar to HOA. For the diurnal cycles, the colored lines represent the median per hour of the day, and the
shaded area the range between the 1st and the 3rd quartile of the data.
110
Figure S6. Mass spectra (a) and diurnal pattern (b) of HOA, COA, and OOA at the kerbside during P1. An a value of 0.1 was used
to constrain the reference profiles of HOA and COA. The shaded area in (a) spans the area between lower and higher constrains
with the center of the area representing the reference profile. For the diurnal cycles (b), the colored lines represent the median per
hour of the day, and the shaded area the range between the 1st and the 3rd quartile of the data.
115
2.0
1.5
1.0
0.5
0.0
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
3.0
2.0
1.0
0.0
µg
m-3
2.0
1.5
1.0
0.5
0.0
0.10
0.08
0.06
0.04
0.02
0.00
10080604020
m/z
0.08
0.06
0.04
0.02
0.00
Fra
ctio
n
0.20
0.15
0.10
0.05
0.00
a-value: 0.1
a-value: 0.1
HOA
COA
OOA
(a) (b)
8
Figure S7. The relative contribution of HOA, COA, and OOA as a function of a value.
120
Figure S8. Triangular space of f55 and f57 color coded by hour of the day. The dash line was defined by Mohr et al. (2011).
125
1.0
0.8
0.6
0.4
0.2
0.0
Fra
ctio
n
0.1, 0,
0.1, 0.1,
0.1, 0.2,
0.1, 0.3,
0.1, 0.4,
0.1, 0.5,
a-value
HOA COA OOA
0.12
0.10
0.08
0.06
0.04
0.02
0.00
f55
0.0800.0600.0400.0200.000
f57
20
15
10
5
0
hrs
9
Figure S9. Mass spectral profiles (left panels) and diurnal (mid panels), as well as the relative contribution of the free PMF four-
(top) and five- (bottom) factor solutions at the kerbside during P2. The four-factor solution shows a HOA splitting factor because
its diurnal pattern was similar to HOA. For the diurnal cycles, the colored lines represent the median per hour of the day, and the
shaded area the range between the 1st and the 3rd quartile of the data. 130
Figure S10. Diurnal pattern of f60 at the kerbside during P2. The colored lines represent the median per hour of the day, and the
shaded area the range between the 1st and the 3rd quartile of the data. The dash line represents the background of f60 as found by
Cubison et al. (2011). 135
0.015
0.012
0.009
0.006
0.003
0.000
f60
0 6 12 18 24Hours
0.15
0.10
0.05
0.00
10080604020
m/z
0.12
0.08
0.04
0.00
Fra
ctio
n
0.12
0.08
0.04
0.00
0.080.060.040.020.00
factor_pr_1 factor_pr_2
factor_pr_3 factor_pr_4
OOA
HOA
COA
BBOA
6
4
2
0
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
6
4
2
0
µg
m-3 4
3210
86420
3.02.01.00.0
86420
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
8
6
4
2
0
µg
m-3
4
3
2
1
0
86420
fact 133.5%2.24
fact 224%1.6
fact 321.4%1.43
fact 421.2%1.41
fact 130.4%2.03
fact 218.1%1.21
fact 320.4%1.37
fact 418.3%1.23 fact 5
12.8%0.85
0.150.100.050.00
10080604020 m/z
0.250.200.150.100.050.00
Fra
ctio
n 0.12
0.080.04
0.00
0.080.060.040.020.00
0.120.080.040.00
factor_pr_1 factor_pr_2 factor_pr_3 factor_pr_4 factor_pr_5
OOA
HOA
COA
BBOA
BBOA/HOA splittingand/or coal factor
10
Figure S11. Mass spectra (a) and diurnal pattern (b) of HOA, COA, peat, coal, wood, and OOA at the kerbside during P2. An a
value of 0.1 was used to constrain the reference profiles of HOA, COA, peat, coal, and wood. The shaded area in (a) spans the area
between lower and higher constrains with the center of the area representing the reference profile. For the diurnal cycles (b), the 140 colored lines represent the median per hour of the day, and the shaded area the range between the 1st and the 3rd quartile of the
data.
145
0.10
0.00
10080604020
m/z
0.08
0.04
0.00
Fra
ctio
n
0.10
0.00
0.10
0.00
0.10
0.00
0.20
0.10
0.00
a-value: 0.1
a-value: 0.1
a-value: 0.1
a-value: 0.1
a-value: 0.1
HOA
COA
Peat
Wood
Coal
OOA
(a) (b) 4
2
0
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
5
0
µg
m-3
8
4
0
2.0
1.0
0.0
1.0
0.05
0
HOA
COA
Peat
Wood
Coal
OOA
0.20
0.15
0.10
0.05
0.00
10080604020
m/z
0.08
0.06
0.04
0.02
0.00
Fra
ctio
n
factor_pr_1 factor_pr_2
OOA
Peat
0.8
0.6
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
0.6
0.4
0.2
0.0
µg
m-3
fact 163.9%0.39
fact 236.1%0.22
0.25
0.20
0.15
0.10
0.05
0.00
10080604020
m/z
0.08
0.06
0.04
0.02
0.00
Fra
ctio
n
0.5
0.4
0.3
0.2
0.1
0.0
factor_pr_1 factor_pr_2 factor_pr_3
OOA
Peat
m/z 29 splitting
0.6
0.4
0.2
0.0
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
0.6
0.4
0.2
0.0
µg
m-3
0.4
0.3
0.2
0.1
0.0
fact 144.7%0.29
fact 232.6%0.21
fact 322.8%0.15
11
Figure 12. Mass spectral profiles (left panels) and diurnal (mid panels), as well as the relative contribution of the free PMF two- (top)
and three- (bottom) factor solutions at the residential site during P1. The two-factor solution shows a m/z 29 splitting from other
factors. For the diurnal cycles, the colored lines represent the median per hour of the day, and the shaded area the range between
the 1st and the 3rd quartile of the data.
150
Figure S13. Diurnal pattern of f60 at the residential site during P1. The colored lines represent the median per hour of the day, and
the shaded area the range between the 1st and the 3rd quartile of the data. The dash line represents the background of f60 as found
by Cubison et al. (2011).
155
Figure S14. Mass spectra (left panel) and diurnal pattern (right panel) of HOA, peat, and OOA at the residential site during P1. An
a value of 0.1 was used to constrain the reference profiles of HOA and peat. The shaded area in (a) spans the area between lower
and higher constrains with the center of the area representing the reference profile. For the diurnal cycles (b), the colored lines
represent the median per hour of the day, and the shaded area the range between the 1st and the 3rd quartile of the data. 160
0.012
0.009
0.006
0.003
0.000
f60
0 6 12 18 24
Hours
0.10
0.08
0.06
0.04
0.02
0.00
10080604020
m/z
0.100.080.060.040.020.00
Fra
ctio
n
0.20
0.15
0.10
0.05
0.00
a-value: 0.1
a-value: 0.1
0.4
0.3
0.2
0.1
0.0
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
0.4
0.3
0.2
0.1
0.0
µg
m-3
0.8
0.6
0.4
0.2
0.0
HOA
Peat
OOA
12
Figure S15. Mass spectra (left panel) and diurnal pattern (right panel) of HOA, peat, coal, wood, and OOA at the residential site
during P2. An a value of 0.1 was used to constrain the reference profiles of HOA and peat. The shaded area in (a) spans the area
between lower and higher constrains with the center of the area representing the reference profile. For the diurnal cycles (b), the 165 colored lines represent the median per hour of the day, and the shaded area the range between the 1st and the 3rd quartile of the
data.
0.100.080.060.040.020.00
10080604020
m/z
0.100.080.060.040.020.00
Fra
ctio
n 0.100.080.060.040.020.00
0.120.100.080.060.040.020.00
0.12
0.08
0.04
0.00
a-value: 0.1
a-value: 0.1
a-value: 0.1
a-value: 0.1
HOA
Peat
Wood
Coal
OOA
86420
0 1 2 3 4 5 6 7 8 91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
hours
1612
8
4
0
µg
m-3
43210
2.52.01.51.00.50.0
6
4
2
0
HOA
Peat
Wood
Coal
OOA
13
Supplementary section of comparison between the kerbside and residential site
170
Figure S16. Diurnal cycles of the urban increment ratio for OA, BC, and NOx. The colored lines represent the median per hour of
the day, and the shaded area the range between the 1st and the 3rd quartile of the data
Figure S17. Linear regression for organic aerosol (OA), sulfate (SO4), nitrate (NO3), ammonium (NH4), and chloride (Chl) between
the kerbside (y-axis) and residential site (x-axis) over the entire campaign. All were hourly averaged and were in μg m-3. 175
30
20
10
0
BC
16
12
8
4
0
OA
0 4 8 12 16 20 24Hours
Ker
b i
ncr
emen
t (R
atio
of
ker
b :
res
iden
tial
)
14
Figure S18. Linear regression between HOA and BC with a slope of 0.16 and R2 of 0.6.
References:
Cubison, M. J., Ortega, A. M., Hayes, P. L., Farmer, D. K., Day, D., Lechner, M. J., Brune, W. H., Apel, E., Diskin, G. S., 180
Fisher, J. A., Fuelberg, H. E., Hecobian, A., Knapp, D. J., Mikoviny, T., Riemer, D., Sachse, G. W., Sessions, W., Weber, R. J.,
Weinheimer, A. J., Wisthaler, A., and Jimenez, J. L.: Effects of aging on organic aerosol from open biomass burning smoke in
aircraft and laboratory studies, Atmospheric Chemistry and Physics, 11, 12049-12064, 10.5194/acp-11-12049-2011, 2011.
Mohr, C., Richter, R., DeCarlo, P. F., Prévôt, A. S. H., and Baltensperger, U.: Spatial variation of chemical composition and
sources of submicron aerosol in Zurich during wintertime using mobile aerosol mass spectrometer data, Atmospheric 185
Chemistry and Physics, 11, 7465-7482, 10.5194/acp-11-7465-2011, 2011.
20
15
10
5
0 H
OA
252015105
BC
2018/9/11
2018/9/13
2018/9/15y=0.16x - 0.21, R
2=0.6