Measurement report : The influence of traffic and new particle 1 formation on the size distribution of 1-800 nm particles in Helsinki : 2 a street canyon and an urban background station comparison 3

Most of the anthropogenic air pollution sources are located in urban environments. The contribution of these 14 sources to the population of atmospheric particles in the urban environment is poorly known. In this study, we investigated 15 the aerosol particle number concentrations in a diameter range from 1 to 800 nm at a street canyon site and at a background 16 station within 1 km from each other in Helsinki, Finland. We use these number size distribution data together with 17 complementary trace gas data and develop a method to estimate the relative contributions of traffic and atmospheric new 18 particle formation (NPF) to the concentrations of sub-3 nm particles. During the daytime, the particle concentrations were 19 higher at the street canyon site than at the background station in all analyzed modes: sub-3 nm particles, nucleation mode 20 (3-25 nm), Aitken mode (25-100 nm), and accumulation mode (100-800 nm). The population of sub-3 nm and nucleation 21 mode particles was linked to local sources such as traffic, while the accumulation mode particles were more related to 22 non-local sources. Aitken mode particles were dominated by local sources at the street canyon site while at the background 23 station they were mainly influenced by non-local sources. The results of this study support earlier research showing direct 24 emissions of the sub-3 nm particles from traffic. However, by using our new method, we show that during NPF events, 25 traffic contribution to the total sub-3 nm particle concentration can be small and during daytime (6:00-20:00) in spring it 26 does not dominate the sub-3 nm particle population at either of the researched sites. In the future, the contribution of 27 traffic to particle number concentrations in different urban environments can be estimated with a similar approach, but 28 determining the relationships between the gas and particle concentrations from observations needs to be conducted with 29 longer data sets from different urban environments. 30


30
Aerosol particles are both directly emitted to the atmosphere (primary particles) and formed from gaseous precursors 31 (secondary particles) (Kulmala and Kerminen, 2008). Secondary particles can form by new particle formation (NPF) via article, we present the results of these measurements and compare the particle size distributions and their variation at 78 these two stations. Specifically, we develop and apply a new method to determine the relative contributions of NPF and 79 traffic to the sub-3 nm particle population in different urban environments.     Table 1 indicating the total 96 running time at each station. The detailed working time for each instrument is shown in Table A1. Most of the analysis 97 was conducted separately for workdays and days free of work, i.e. weekends and holidays (1 May 2018 and 10 May 98 2018), which are for simplicity just referred to as 'weekends' in this article. When comparing particle concentrations from 99 two stations, we analyze times when all the instruments measuring particles were performing at each site. This resulted 100 in 120 and 101 hours of measured particle concentration during weekends and workdays, respectively, at the street canyon 101 site. At the background station, it resulted in 217 hours of observed particle concentration during weekends and 398 hours All instruments were corrected for diffusion losses in their inlets except the UCPC measuring at the background station,

173
Lastly, the non-ideal efficiency curve, used for determining the cutoff diameter, makes it possible to sample particles 174 smaller than the cutoff size. When the relative contribution of sub-3 nm particles to the total particle population is high,

175
the uncertainties of cutoff diameter or the shape of the efficiency curve can affect the total concentration measured by

189
Due to the uncertainty of the rate of the reaction between gaseous sulfuric acid and the nitrate ion, the CI-APi-TOF needs  [SA] = C • CR 97 + CR 160 CR 62 + CR 125 + CR 188 (1) where [SA] is SA concentration, C is the calibration coefficient and CRM is a count rate of an ion with a mass M in 196 amu.

197
The SA zero level concentration, determined by measuring filtered air, was subtracted from the measured concentrations.

198
Uncertainties of absolute concentration measured by CI-APi-TOF are in the order of 50%, while the uncertainties of   Table A2. To estimate the influence of traffic and NPF on the sub-3 nm particle population, we analyzed the correlation between  The results of the NPF event classification at the background station for the studied period is shown in examples of an event, non-event, and undefined class are shown in Fig. S1. The overall frequency of NPF event days was 219 12.5%; 21% of weekends and 8% of workdays were classified as events. Due to nucleation mode particles originating 220 from local sources, the majority of days were classified as undefined.

244
Focusing on the smallest particles ( Fig. 3), we observe that at the street canyon the median concentration of sub-3 nm 245 particles is up to 2.4*10 4 cm -3 higher than at the background station (Fig. 3c). The concentration of sub-3 nm particles is 246 higher at the street canyon site regardless the particle loss due to coagulation scavenging being twice as high as at the variation of sub-3 nm particle concentration has a maximum around noon both during weekends and workdays.

255
Nevertheless, a sharp increase of sub-3 nm particles concentration is observed in the morning (6:00) during workdays.

256
Morning raise of sub-3 nm particle concentration at the background station during workdays corresponds to a peak of 278 The diurnal variation of nucleation mode particle concentration is similar to that of sub-3 nm particles at both stations

292
Overall, the influence of traffic on the particle population at the street canyon is clearly visible for sub-3 nm, nucleation 293 mode, and Aitken mode particles, while the accumulation mode is only slightly influenced by traffic. The particle 294 concentrations at the background station are also influenced by traffic, but not as strongly as at the street canyon station.

295
At the background station, the influence of traffic can be observed only for sub-3 nm and nucleation mode particles. These 296 results suggest that sub-3 nm and nucleation mode particle concentrations in the urban environment are mainly influenced 297 by local sources, while the accumulation mode particle concentrations are mostly dominated by transport from non-local 298 sources. Whether the Aitken mode is primarily dominated by local or non-local sources depends on the analyzed location.  canyon station with maximum concentrations of 6.9*10 6 cm -3 and 3.6*10 6 cm -3 during weekends and weekdays, 315 respectively. This difference is likely linked to a bigger fraction of NPF events days during analyzed weekends than 316 workdays (Table 2). Daytime median SA concentrations are slightly higher at the background station than at the street 317 canyon (Fig. 6, Fig. S5), which is probably caused by higher CS at the street canyon site (Fig. 5). In contrast to daytime,  330 09:00-Wednesday 9 May 2018 15:00. The first investigated case is a weekend starting with an NPF event (Fig. 7a,b). The

331
second case contains typical workdays (Fig. 7c,d), which are classified as undefined days in NPF event classification. and background station during these case studies is presented in Fig. S9. During the NPF event (Fig. 8a,b), sub-3 nm 342 particles concentration at the background station is almost a factor of two higher than at the street canyon site. However, 343 nearly simultaneous to the highest peak in sub-3 nm particles and SA concentrations, a peak in particle concentrations 344 across the modes (Fig 7b) as well as in SO2 concentration (Fig. S6) is observed at the background station. This seems not 345 to be a feature of a regional NPF event but could be a plume from e.g. a ship or a coal-fired power plant in Helsinki,

346
which happens to be more efficiently transported to the background station than to the street canyon. This illustrates the 347 interplay of various types of sources on the aerosol concentrations, regional NPF events, local traffic sources, and nearby

362
To investigate in detail the contribution of different sources to the sub-3 nm particles population at both sites, we analyzed 363 correlations of different variables with sub-3 nm particle concentration (Table 3). Sub-3 nm particle concentration at the  suggests that the sub-3 nm particles population at the street canyon site is more influenced by traffic than at the background 372 station. This is discussed more in the next section.

Estimation of NPF and traffic contribution to sub-3 nm particles 381
Our results suggest that the sub-3 nm particle population at the urban background station is mainly influenced by particles dependency between sub-3 nm particles, SA, and NOx concentrations at both sites (Fig. 9, Table S1). We made bivariate 386 fittings to common logarithms of NOx and sub-3 nm particles when the SA concentration was low and reversely we 387 analyzed common logarithms of SA and sub-3 nm particles when the NOx concentration was low. The bins were chosen 388 for fitting so that they were as similar as possible at both stations and contained enough data points. The slopes of the 389 bivariate fit to sub-3 nm particles and SA data for low NOx concentration is close to 1 at both stations ( Fig. 9 a,b). At the 390 same time, the slope of the fit to sub-3 nm particles and NOx data for low SA concentration is considerably smaller at the 391 background station (0.64) than at the street canyon site (1.40) (Fig. 9 c,d). We investigated possible reasons for this 392 difference such as constant background (local source) of sub-3 nm particles at the background station or losses of sub-393 3 nm particles due to CS, or particle growth. Analysis of the correlation between NOx, SA, and total particle concentration 394 (Fig. S11), as well as the correlations between sub-3 nm particles, NOx, SA, and CS ( Fig. S12-S13), implied that neither 395 particle growth out of the sub-3 nm size range nor varying CS can explain the difference in the slopes between stations. concentrations at the background station (Fig. S14). We should have in mind that compared ranges of NOx concentrations 400 are different at each station. Additionally, particle evaporation may affect the comparison.

401
Based on these bivariate fits, we estimated sub-3 nm particle concentration originating from NPF and traffic at the two 402 sites (Table 4). The analysis was done for the time when NOx, SA, and sub-3 nm particle concentrations were measured 403 at each station (Table A1, Fig. S15). The variability of estimated sub-3 nm particle concentration is high, and occasionally 404 estimated concentrations exceed the measured values of sub-3 nm particle while at other times estimated values are clearly 405 lower than the measured values (Fig. 10). However, our estimation captures the temporal variation of the sub-3 nm particle 406 concentrations adequately. We can conclude that during the daytime (6:00-20:00), a similar fraction of sub-3 nm particles 407 originate from traffic (53%) and NPF (47%) at the street canyon site. At the background station, the daytime sub-3 nm

439
In this study, for the first time, the particle size distribution in a diameter range from 1 to 800 nm was analyzed at two

447
We observed a very similar pattern in diurnal variation of SA concentration at both stations. Daytime SA concentrations

448
were slightly higher at the background station, likely due to a lower condensation sink than at the street canyon site.

449
During the nighttime, SA concentration was almost an order of magnitude higher at the street station. High nighttime 450 concentration at the street canyon site is probably caused by two simultaneous processes: direct SA emission from traffic 451 and nighttime SA formation.

452
Additionally, we performed two case studies, in which we analyzed the variation of SA and sub-3 nm particles on a short 453 time scale. Our study supports previous research showing that sub-3 nm particles include direct emissions from traffic.

454
During the NPF event on 5 May 2018, sub-3 nm particle concentrations at both sites were the highest, and traffic 455 contribution to the total sub-3 nm particle concentration at the analyzed stations was small.

456
Furthermore, we analyzed the relation of sub-3 nm particles with trace gases and meteorological variables. We observed 457 that sub-3 nm particles at the background station are mainly related to SO2 and SA, while the sub-3 nm particle population  Calculating the relative contribution of traffic and NPF to sub-3 nm particle population

486
Based on bivariate fittings on the common logarithms of sub-3 nm particles and SA when NOx concentration was low at 487 the street canyon site (Fig. 9a), we determined Eq. A1 estimating the concentration of sub-3 nm particles formed during