1

Online Supporting Information (SI) 1

2 Quantifying High-resolution Spatial Variations and Local Source Impacts of Urban 3

Ultrafine Particle Concentration 4 Provat K. Sahaa, Naomi Zimmermana,d, Carl Malingsa, Aliaksei Hauryliuka, Zhongju Lia, Luke 5 Snellb, R Subramaniana, Eric Lipskya,c, Joshua S. Apteb, Allen L. Robinsona, Albert A. Prestoa* 6

7 aCenter for Atmospheric Particle Studies, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, 8

Pennsylvania 15213, United States 9 bDepartment of Civil, Architectural and Environmental Engineering, University of Texas at Austin, Austin, Texas 10

78712, United States 11 cDepartment of Mechanical Engineering, Penn State Greater Allegheny, McKeesport, Pennsylvania, 15132, United 12

States 13 dnow at: Department of Mechanical Engineering, University of British Columbia, Vancouver, BC V6T 1Z2, Canada 14

15 16 17 18 19 20 21 22 23

24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 *Address correspondence to: Albert A. Presto, 5000 Forbes Avenue, Doherty Hall, Pittsburgh, 46 Pennsylvania 15213, United States. Email: apresto@andrew.cmu.edu, Phone: 412-721-5203 47

2

48

Urban BG

Near local Rd Near local Rd + restaurant

Near arterial Rd

Urban st canyon

Near highway Near industry

49

50 51

Figure S1: Study area map and location of measurements sites. Measurements of particle 52 number and other air pollutant concentrations are collected at these sites during winter of 2017 53 and 2018. 54 55 56

CMU reference site Downtown

Pittsburgh city boundary

N=32 sites

3

Unit Slope Intercept R2

MAGIC #4 0.747 -56 0.99

MAGIC #5 0.705 280 0.996

MAGIC #6 0.56 293 0.997

MAGIC #7 0.54 120 0.96

MAGIC #8 0.51 127 0.85

MAGIC #9 0.56 161 0.996

MAGIC #10 0.62 256 0.995

MAGIC #11 0.45 338 0.872

MAGIC #12 0.58 341 0.99

MAGIC #14 0.57 208 0.99

Figure S2: CPCs inter-comparisons. (a) Example time series of ambient PNC measured by 12 MAGIC CPCs and 2 TSI CPCs collocated at the CMU campus. (b) Scatter plots of MAGIC CPCs versus TSI CPC model 3772 measurements. (c) The chart shows the slope, intercept and R2 values from the linear regression fits of MAGIC CPCs versus TSI CPC model 3772 collocation data (y=ax+b; y = MAGIC CPC measurement, x= TSI CPC 3772 measurement, a= slope, b= intercept).

57 58 59

60 61 62 63 64 65 66 67

68

c)

4

69 Figure S3: Comparison of distributions of PNC measured at an urban street canyon 70 (downtown) site for different sampling lengths (e.g., 6 weeks, 4 weeks, 2 weeks, etc.). (a) Box 71 plots of PNC for different sampling lengths. The box shows data interquartile range, the line 72 inside the box is the median, the length of the whiskers covers 90% of the data, and the star 73 represents the mean. (b) The average diurnal profiles of PNC for different sampling lengths 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91

5

92 Figure S4: Comparison of the distributions of PNC collected at two particular sites during 93 two consecutive winter seasons (i.e., winter 2017 versus winter 2018). (a) Measurements at an 94 urban street canyon (downtown) site, (b) Measurements at an urban background (CMU campus) 95 site. The box shows data interquartile range, the line inside the box is the median, the length of 96 the whiskers covers 90% of the data, and the star represents the mean. 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119

6

120 Figure S5: Monthly variations in the distributions of PNC measured in an urban 121 background site at the CMU campus. The box shows data interquartile range, the line inside 122 the box is the median, the length of the whiskers covers 90% of the data, and the star represents 123 the mean. 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150

7

151

152 153 Figure S6: Box plots of a) PM2.5, b) NO2, and c) CO concentrations measured at 32 sites in 154 Pittsburgh and nearby communities. The box shows data interquartile range, the line inside 155 the box is the median, the length of the whiskers covers 90% of the data. 156 157 158 159 160 161

30

25

20

15

10

5

0

NO

2 (pp

b)

800

600

400

200

0

CO

(ppb

)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Urban BG Near local Rd Nearlocal Rd

+restaurant

Nea

r ind

ustry

Nea

r hig

hway

St c

anyo

n (d

ownt

own)Near

ArterialRd

30

25

20

15

10

5

0

PM2.

5 (µg

m-3

)

PM2.5

NO2

CO

8

162 163 Figure S7: (a) The time series of PNC and wind directions at a near-industrial site (Site# 32; 164 ACHD Lincoln). This site is located about 0.9 miles away from a large metallurgical coke plant. 165 The wind direction data is also shown, a south-west (SW) indicates wind was blowing from the 166 coke plant to our measurement site. (b) Map showing the location of the coke plant and sampling 167 site. 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186

Sampling site

b)

a)

9

187 188 Figure S8: (a) Comparison of PNC measurements from the long-term fixed site CPC and 189 short-term driving FMPS sampling. Stationary measurements are 3-6 weeks of continuous 190 CPC data at a fixed site. Driving measurements are FMPS-measured integrated number 191 concentrations within 50 m buffer area from each stationary site. The driving samples were 192 collected in different parts of a day (e.g., morning, midday, and evening); total number of day 193 visits at each site are indicated in Figure. All measurements shown were collected in winter but 194 are not necessarily concurrent. The box shows data interquartile range, the line inside the box is 195 the median, the length of the whiskers covers 90% of the data, and the star represents the mean. 196 (b) Comparison of FMPS and CPC measured total PNC when they were collocated at the 197 CMU campus. 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218

60 x103

50

40

30

20

10

0

Parti

cle

num

ber c

onc.

(cm

-3)

(3) C

MU

[S]

(3) C

MU

[D]

(14)

Hill

[S]

(14)

Hill

[D]

(19)

Stri

p [S

]

(19)

Stri

p [D

]

(20)

WYE

P [S

]

(20)

WYE

P [D

]

(21)

Law

ren

[S]

(21)

Law

ren

[D]

(27)

Con

ven

[S]

(27)

Con

ven

[D]

local Rdrestaurant N

ear

Arte

rial R

d

St c

anyo

n (d

ownt

own)Near local Rd

+

(5)

(5)

(10)

(4)

(10)

No of dayvisits

NearUrban BG

(5)

Stationary CPC [S]Driving FMPS [D]a) b)

10

219 Table S1: UFP measurement sites 220 221

Site # 1Site type Site name GPS coordinate Notes 1 Urban

Background Fox Chapel 40.511735 -79.869403

2 Beechview 40.412645 -80.024131 3 CMU supersite 40.442067 -79.942534 Ref. site at CMU campus 4

106 Curry Ct. 40.512805 -80.201487 Located 3 km downwind of Pittsburgh international airport

5 4544 Homeridge dr. 40.388364 -79.919219 6 Near local

road Woodstock ave 40.416260 -79.880090

7 Coal Valley Rd 40.318790 -79.901890 8 Dunmoyle Pl 40.447844 -79.927218 9 1202 Goodman st 40.418634 -79.904043 10 Marlborough Rd 40.438890 -79.926660 11 Hatfield St. 40.478412 -79.957016 12 South Dallas 40.450290 -79.909440 13 MtWash library 40.433560 -80.010430 14 Hill district library 40.445500 -79.976530 15 ECCO house 40.463430 -79.916940 16 Near local

road+ restaurant

Ds six pack 40.431920 -79.893930 17 The Cube 40.459805 -79.928604 18 Lehigh Ave 40.457831 -79.925812 19 Strip district 40.452466 -79.984270 20 WYEP (Southside) 40.429880 -79.986670 21 Near arterial

road *ACHD Lawrenceville

40.465680 -79.959480 This is an established EPA monitoring site (AQS# 42-003-0008)

22 Carlow University 40.438360 -79.963560 23 Penn/Homewood 40.449474 -79.901476 24 5th Ave/S Negley 40.450091 -79.929576 25 Comm day School 40.427420 -79.917830 26 Urban street

canyon (downtown Pittsburgh)

Mellon Park 40.441300 -79.997750 27 Convention Center 40.444490 -79.995690 28 First Lutheran

Church 40.441400 -79.995840 Located in a pedestrian alley 10 m away from a restaurant exhaust

29 Near highway

Aspinwall, PA-28 40.494690 -79.907370 10 m upwind of highway PA-28 30

Highway I-376 40.437469 -79.863547

10 m downwind of I-376. This is an established EPA near-road monitoring site (AQS# 42 003 1376)

31 Near industry

ACHD Braddock

40.402410 -79.860860

0.8 mile away from an integrated steel mill, Pittsburgh. This is an established EPA monitoring site (AQS# 42-003-1301)

32

ACHD Lincoln

40.308220 -79.869130

0.9 mile away from large metallurgical coke plant, PA. This is an established EPA monitoring site (AQS# 42-003-7004).

222 1See table S2 for details. *ACHD: Allegheny County Health Department 223 224 225 226 227 228

11

229 Table S2: Site classification 230 231

Site type Urban BG*

Near local road

Near local road + restaurant

Near arterial road

Urban street canyon (downtown)

Near highway

Near large industry

Traffic density (qualitative)

Low Medium Medium High

High Very high Low

1Veh_dens_100m 2(Allegheny County distribution)

<20 (<25 P)

20-80 (25 -75 P)

20-80 (25 -75 P)

80-250 (75 -95 P)

80-250 (75 -95 P)

>250 (> 95P)

<20 (<25P)

Restaurant density (qualitative)

Low Low Medium Low High Low Low

3Rest_count_300m

<5 <5 5-30 <5 >30 <5 <5

No of sites (n) 5 10 5 5 3 2 2 4Site # #1-5 #5-15 #16-20 #21-25 26,27,28 29,30 31,32

232 1Veh_dens_100m: vehicles/day/m within100 meter circular buffer area from monitoring site. 233 2(Allegheny County, PA distribution): different percentile (P) of veh_dens_100m distribution for Allegheny County, PA. E.g., 25th percentile of 234 veh_dens_100m distribution for Allegheny County, PA is 20 vehicles/day/m. 235 3Rest_count_300m: restaurant count within 300 meter circular buffer area from monitoring site. 236 4Site: See Table S1 for details. 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266

12

267 Table S3: Instrumentation, detection limit, data time resolution, and data QA methods for the 268 measurements used in this paper 269 270

Measurement Instrument Raw data time resolution

QA approach

Detection limit/Accuracy

Distributed sites PNC CPC, MAGIC 200P model,

Aerosol Devices Inc. 15 min Colocation calibration against

two TSI CPCs (Fig. S2) 5nm – 2.5 µm; 0.01 to 105 p/cm3

PM2.5 Optical PM2.5: Met-One NPM (forward light scattering laser); PurpleAir PM2.5 monitor (Plantower PMS 5003 laser sensor)

15 min Collocation with BAM and calibration model as a function of temperature and humidity and hygroscopic growth factor

Mean Absolute Error (MAE): below 1 µg/m3 quarterly averages; below 2 µg/m3 for weekly averages

NO2 Electrochemical sensor, Alphasense, ID: NO2-B43F

15 min Colocation against reference monitor and machine learning based calibration model

MAE: 3.4 ppb

CO Electrochemical sensor, Alphasense, ID: CO-B41

15 min Colocation against reference monitor and machine learning based calibration model

MAE: 48 ppb

Reference site PNC TSI CPC model 3772 3 min TSI factory calibration 10 nm – 2.5 µm;

Particle conc accuracy ±10%

Size distribution TSI DMA 3081/TSI CPC 3772

3 min TSI factory calibration 10 nm – 300 nm; Particle conc accuracy ±10%

PM2.5 Beta Attenuation Monitor (BAM), EPA monitoring site, Lawrenceville (AQS#42-003-0008)

1 h US-EPA standard QA protocol

~ 1 µg/m3 for long-term averages

NO2 NO2/NO/NOx Monitor, Model 405 nm, 2B Technologies

1min Calibrated with standard cylinder gas; Airgas

Conc. range: 0-2,000 ppb; Limit of detection (2σ) < 1 ppb

CO Teledyne CO analyzer, Model T300U

1min Calibrated with standard cylinder gas; Airgas

Lower detectable limit < 0.04 ppm

Mobile van Size distribution FMPS, TSI Inc. 1s TSI factory calibration;

Collocated with TSI SMPS 5.6- 560 nm

271 272 273 274 275 276 277 278 279 280 281 282 283 284 285

13

Table S4: Estimation of PNC exposure during traffic rush hours 286 287

Time of day hour Microenvironment Typical outdoor environment

Outdoor PNC Conc. (103 cm-3)

Assumed penetration (P) factor

fraction of daily average exposurec

Midnight- 8 am 8 Indoor/home Urban BG/Near local Rd 7-10 0.4-0.6a 0.18-0.21 8:00-8:30 am 0.5 Outdoor/road/near-road Arterial Rd/Highway 20-35 0.8-1.0b 0.06-0.14 8:30 am- 6:00 pm 9.5 Indoor/work Arterial Rd/downtown 15-20 0.4-0.6 0.45-0.54 6:30-6:30 pm 0.5 Outdoor/road/near-road Arterial Rd/Highway 20-30 0.8-1.0 0.06-0.12 6:30pm- midnight 5.5 Indoor/home Urban BG/Near local Rd 7-10 0.4-0.6 0.12-0.15 Total 24 Rush hour contribution 0.12-0.26

288 aOutdoor/Indoor penetration factor of 0.4-0.6 for overall UFP assumed from Zhu et al. 20051 289 bPenetration factor of 0.8-1.0 assumed for road/roadside outdoor stay during traffic rush hours. 290 cFirst number indicates the lower bound and second number indicates the upper bound of estimates considering different 291 range of outdoor PNC concentrations and assumed penetration factors. 292 293 294 295 References: 296 297 (1) Zhu, Y.; Hinds, W. C.; Krudysz, M.; Kuhn, T.; Froines, J.; Sioutas, C. Penetration of 298

Freeway Ultrafine Particles into Indoor Environments. J. Aerosol Sci. 2005, 36 (3), 303–299 322. 300

301