Are cities responsible for their air 1 pollution?

13 While the burden caused by air pollution in urban areas is well documented, the origin of this 14 pollution and therefore the responsibility of the urban areas in generating this pollution is still a 15 subject of scientific discussion. Source Apportionment represents a useful technique to quantify 16 the city responsibility but the approaches and applications are not harmonized, therefore not 17 comparable, resulting in confusing and sometimes contradicting interpretations. In this work, we 18 analyze how different source apportionment approaches apply to the urban scale and how their 19 building elements and parameters are defined and set. We discuss in particular the options 20 available in terms of indicator, receptor, source and methodology. We show that different 21 choices for these options lead to very large differences in terms of outcome. In average over the 22 150 EU large cities selected in our study, the choices made for the indicator, the receptor and the 23 source each lead to an average factor 2 difference. We also show that temporal and spatial 24 averaging processes applied to the air quality indicator, especially when diverging source 25 apportionments are aggregated into a single number lead to favor strategies that target 26 background sources while occulting actions that would be efficient at the city center. We stress 27 that methodological choices and assumptions most often lead to a systematic and important 28 underestimation of the city responsibility, with important implications. Indeed, if cities are seen 29 as a minor actor, plans will target in priority the background at the expense of potentially 30 effective local actions.


Introduction
According to a recent estimate (EEA, 2020), about 74 % of the EU-28 urban population are 45 exposed to pollution of fine particulate matter (PM2.5) in concentrations above the WHO Air 46 Quality Guidelines value, this number raises to 99% for ozone (O3) and is about 4% for nitrogen 47 dioxide (NO2). Air pollution is a heavy burden on human health with more than 380,000 48 premature deaths in EU-28 reported in 2017 according to the same EEA estimates. For a wide 49 range of European cities, Khomenko et al. (2021) showed that the health burden due to air 50 pollution varies greatly by city, with annual premature mortality reaching up to 15% for PM2.5 51 and 7% for NO2. The highest mortality burden for PM2.5 occurs in northern Italy, southern 52 Poland and eastern Czech Republic. De Bruyn and de Vries (2020) showed that for all 432 cities 53 in their sample (total population: 130 million inhabitants), the social costs (e.g. hospital 54 admissions, premature mortality) but also due to air pollution exceeded € 166 billion in 2018 for 55 Europe (EU27 plus the UK, Norway and Switzerland). City size was shown to be a key factor 56 contributing to the total social costs: all cities with a population over 1 million features in the 57 Top 25 cities with the highest social costs due to air pollution. 58 59 Given the health and economic burden caused by air pollution in urban areas, it is important to 60 identify the origin of this pollution in order to reduce and control its impact. Identifying the 61 sources of urban pollution and then assigning responsibilities enables a process to implement 62 measures and control air pollution. Assessing the responsibility or share of cities for their 63 pollution has important implications. For being effective, pollution reduction plans must be 64 designed and applied to target the most polluting sectors at the relevant spatial (national, regional 65 and/or local) and with the appropriate temporal scales. In this context, quantifying the share or 66 the city pollutions caused by their own emissions becomes a crucial element to determine 67 whether actions need to be applied locally or at the regional, national country or continental 68 scales. This has important governance consequences for the effective control of air pollution. 69 70 For pollutants like NO2, that mostly originate from traffic sources and have a relatively short 71 lifetime in the atmosphere, there is a general agreement on the fact that cities are the main 72 contributor to this pollutant concentration levels and that acting locally on traffic emissions is the 73 most efficient way of improving NO2 concentration levels in a particular city (Tobias et al., 74 2020). There is available European-wide information such as in Degraeuwe et al. (2019) 75 providing overviews of the potential impact of traffic emission reductions per vehicle type in 76 different European cities. There is also agreement regarding O3 that this secondary pollutant is 77 most effectively reduced by implementing reduction measures at larger spatial scales, involving 78 actions driven at the regional and even continental scales (e.g. Luo et al. 2020). For other 79 pollutants, like PM2.5, complex physical and chemical atmospheric processes with different time 80 scales drive its formation, involving numerous precursors themselves emitted by several sources. 81 The sources of PM2.5 pollution range from local traffic, domestic fuel burning and industrial 82 activities to regional sources such as agriculture in rural areas. Even though the latter emissions 83 do not originate from cities, Thunis et al. (2018) showed that their impact on urban pollution 84 could be important, reaching up to 30% in several European cities. Because of this complexity, 85 there is less consensus regarding the responsibility or share of a city to its pollution when 86 contribution of a given source at a specific receptor for a given indicator (for example the 132 concentration of a given pollutant like PM or NO2). It involves the following steps ( Figure 1): 133 134 (1) defining a relevant indicator, denoted as (I) to characterize air pollution 135 (2) defining the receptor (R) through its spatio-temporal characteristics, i.e. the area ( ̅ ) 136 and time period ( ̅ ) over which the indicator is averaged 137 (3) defining the source (S) through its spatio-temporal characteristics, i.e. the city area 138 (xs) and time period for which the city responsibility is assessed (ts) 139 (4) selecting the source apportionment (SA) methodology to capture the processes that 140 relate the source to the receptor. 141 The first step required to assess the role/responsibility of city emissions with respect to its air 153 pollution, is to define an indicator that identifies the pollution aspect we are interested in. The 154 indicator can be defined in many ways. For example, as the total concentration of a given 155 compound (e.g. PM), or as a specific constituent of that total concentration (e.g. PM2.5 or its 156 primary fraction, PPM), or as a composite based on a mix of different pollutants (e.g. maximum 157 among O3, PM2.5 and NO2 concentrations as in some air quality indexes such as ATMO2003) or 158 as population exposure (i.e. product of population and concentration). 159

Definition of the source (S)
177 The source is defined as the spatio-temporal entity for which we assess the contribution to the 178 indicator. For the purpose of this work, the source is defined as the city, and more precisely as 179 the emissions that originate from a given city. The source emissions (denoted by E) are indeed 180 responsible for the pollution fraction that can be associated to the source/city at the receptor (R). 181 These emissions are characterized by a spatial (xs = extension of the city) and a temporal scale (ts 182 = period of time over which the source activity is assessed). For convenience, we use 183 indifferently the following notations to refer to the source: 184 185 In this work, we analyse in particular the impact of the city extension (xs) on the apportionment 187 outcome. For this purpose, we define cities in two ways: 188 189 (1) as core cities, i.e. the local administrative units, with a population density above 190 1500/km 2 and a population above 50,000, where the majority of the population lives in an 191 urban center and 192 (2) as functional urban areas (OECD, 2012, denoted as "FUA") composed as core cities plus 193 their wider commuting zone, consisting of the surrounding travel-to-work areas where at 194 least 15% of the employed residents work in the city. 195 Details on the FUA and core city areas are available for 150 EU cities in the urban PM2.5 atlas 196 ). Note that other city definitions exist. In the context of the CAMS source 197 allocation analysis, city are defined as an arbitrary number of grid cells in the modelling domain 198 (Pommier et al., 2020). 199 Finally, we define the city background as the sum of all contributions from sources that are not 200 covered by the spatial (xs) and temporal (ts) scales of the city source. 201 202 One main difference between sources and receptors is that for the latter, spatio-temporal 203 characteristics are averaged. Apart from this, temporal and spatial characteristics can also differ 204 in terms of value. For example, the source can be defined as the FUA (xs = FUA) while the 205 receptor is a specific location ( ̅ = ̅ ). Temporally, interest can be on assessing the 206 contribution of the city weekly activity (ts = 1 week) for a given day ( ̅ = ̅ ) at the receptor. In 207 the results presented here, the source and receptor temporal scales are however chosen identical 208 for convenience. 209

210
When the air pollution indicator and the spatio-temporal characteristics of both the receptor and 211 the source have been selected, the next step consists in distinguishing and quantifying the 212 fractions of the indicator related to the city source ( ( )) and to the background ( ( )) at 213 receptor R, respectively. This decomposition is summarized by the following equation: In this case the resulting difference is divided by the reduction percentage to obtain the potential 229 impact ( ( )). A similar approach is used to calculate the background contribution, i.e. by Tagging (TAG): With this approach, species emitted by the city are numerically tagged and 246 followed through the modelled transport, dispersion and chemical transformation processes. 247 When chemical transformations take place, preserved atoms are used as tracers. For example, the 248 nitrogen atom (N) will be used to follow the NO source emissions through its successive 249 transformations into NO2 and HNO3 to reach its final product NO3, that will then be attributed to 250 that source. Example of tagging applications are e.g. Kranenburg

City contribution
Background contribution with the discussion of the results. 296 Based on these sources of information and data, we discuss hereafter the sensitivity of the SA 297 results to the choice of the indicator (Section 3.1), to the choice of the methodology (Section 298 3.2), to the source (Section 3.3) and finally to the receptor (Section 3.4). 299 Quality Directive, AAQD2008) against which health impacts are correlated (WHO2005). In the 308 second case, the indicator is limited to its anthropogenic fraction (PM25 ant), excluding therefore 309 natural contributions (dust, marine salt…). This is motivated by the fact that policies have no 310

Sensitivity to the indicator
impact on this component. According to this indicator, city contributions increase significantly 311 (by about 20% in average) and in some cities where natural dust pollution is important (e.g. in 312 Sicily), the city responsibility shifts from minor to major. If we further restrict the indicator to its  To analyze differences between full and partial impacts, we use a series of EMEP simulations in 369 which we remove totally (PI100) or partly (PI20) the London FUA emissions (source) during an 370 entire year. Figure 3 shows the differences between city contributions obtained with the two PI 371 methods. Differences can be important (up to 25 percentage points for specific days). Although 372 the number of high-difference days is limited (leading to a yearly average difference of few 373 percents), these days might represent high pollution episodes for which assessing the city 374 responsibility is important to act. In general, the higher resolution applied to the temporal and/or 375 spatial averages at the receptor, the largest the differences are among methods. It is also 376 interesting to note that partial potential impacts systematically underestimate full potentials (no 377 negative values). 378 379  SA varies largely from one location to another within Paris. We highlight this with bars that 407 distinguish the city vs. background contributions for locations at different distance from the city 408 centre. We note opposite trends, dominated by the city source (around 60%) at the city center 409 and dominated by the background source towards the periphery (around 80%). While the SA at 410 the city centre is representative of a single cell within the city, this is not the case for SA close to 411 the periphery. This is highlighted by the city rings (below the X-axis) that indicate the area of 412 representativeness of a given SA. When we average spatially an indicator (PM2.5 or population 413 exposure) over a receptor that covers the entire FUA (all 6 rings), these areas of 414 representativeness enter into play. The brown curve indicates the weight (in the spatial average) 415 attached to each city ring, relatively to the city total (i.e. all rings). Weights increase fast when 416 moving towards the periphery because of the larger ring areas. The spatial averaging process 417 leads to over-representing the periphery, which overweight the city center SA by almost a factor 418 40. It is interesting and counter-intuitive to note that with this averaging process, the city 419 responsibility decreases when the city area increases. With population exposure as indicator 420 (weights shown by red curve), the rapid population density decrease balances the ring area 421 increase when moving outward, leading to weights that dominate for middle rings. It is 422 interesting to note, that with average population exposure, the city center weight is yet similar to 423 the weight obtained 28 km away.  Depending on the spatial characteristic of the receptor, some cities will be considered as minor or 436 major actors with respect to their pollution. We discuss this issue further in Section 4.  These lacking or incomplete emission sources will lead to a potential underestimation of the city 486 responsibility as well. 487 488 In the next section, we discuss the consequences of these results on policy, in particular when SA 489 information is used to design air quality plans. 490 4. Implications for air quality strategies 491 Estimating the contribution of a city to its pollution has important consequences in terms of air 492 quality management. Indeed, an important city contribution will be a logic argument to support 493 substantial control measures at the local level to abate pollution. The effectiveness of the control 494 measures then relies on the relevance and accuracy of this city contribution; over-or under-495 estimated city contributions potentially leading to inefficient measures. 496 In previous sections, we have seen that the city contribution largely varies depending on the 497 choices made for the SA setting parameters (definition of the indicator, source, receptor and 498 methodology), hence the challenge to obtain a relevant and accurate estimate to support local 499 action. 500 Given the range of possible SA options and their impact on results, the first recommendation is 501 obviously to report these SA setting choices together with the results to provide policymakers 502 with the full picture and allow them to take informed decisions. This advocates for the use of the 503 proposed nomenclature or a similar one that documents for the choices in the SA approach, 504 providing accountability to the method and enabling correct interpretation of the results. The 505 proposed nomenclature can be understood as a documentation of the SA metadata information. 506 Apart from this point on the importance of documenting SA approach choices, we show below 507 that some of the SA settings are fixed by the purpose of the study. We provide suggestions for 508 the remaining free choices. 509 510 The recommended SA method is potential impacts (PI) 511 512 It is important to recall that not all SA methodologies are equally suited to support air quality 513 planning. As mentioned by several authors ( recommended when non-linear species are involved (which is the case for PM2.5 and PM10 but 516 also for other species like NO2 or O3). It is worth reminding that tagging or incremental 517 approaches are yet erroneously used and believed to be suited for air quality planning purposes 518 (Qiao et  impacts and may be seen as a burden (e.g. lack of additivity, see Appendix), they only reflect the 521 complexity of the real processes that must be accounted for. Although uncertainties associated to 522 the PI approach (e.g. imperfect emission inventory), may lead other SA methods to perform 523 better in some instances because methodological biases compensate uncertainties, this is 524 however coincidental. While uncertainties can be tackled and reduced to improve the approach, 525 this is not the case of methodological biases. These points were extensively discussed in Thunis 526 et al. (2019). 527 528 For the remaining of this section focusing on policy aspects, only potential impact results are 529 discussed. Fixing the methodology however still leaves free options in terms of indicator, 530 receptor and source. This is visualized in Figure 8 that summarizes the variability of the SA 531 results presented in the previous sections (i.e. Figure 2, Figure 4 and Figure 6) for the 150 cities 532 to these possible choices. Differences in terms of city responsibility reach a factor 2 in average 533 for each of these remaining parameters with much larger values for some cities. 534 535 The choice of the indicator is generally motivated by health or environmental considerations. 544 Currently, the WHO guidelines (WHO2005) refer to the total PM2.5 mass as the indicator 545 correlating best with health impacts. These guidelines (or the AAQD limit values) are then the 546 logical and most relevant indicator choice among the options presented in Section 3.1 and shown 547 in Figure 2. As illustrated by Figure 8, evolving knowledge on health-related pollution impacts 548 (i.e. the increased toxicity of some PM2.5 constituents like those related to the traffic and 549 residential activities) might however, drive the choice towards more detailed indicators (e.g. 550 PPM2.5) leading to an increased responsibility for the cities. 551 552 SOURCE: Importance of matching sources with governance levels 553 554 Figure 8 shows that plans limited to city cores would be significantly less efficient than if applied 555 at the FUA scale. In average over all cities, the efficiency decreases by a factor 2 but larger 556 differences occur in many cities. The source does however not represent a free choice in the 557 context of policy practice. Indeed, authorities in charge of AQ plans only have power to act on 558 the area under their responsibility, which sets where measures apply. The same applies for the 559 source temporal characteristic, fixed as the period of time during which measures apply. A good 560 match between the SA settings and the temporal and spatial characteristics of the source is 561 therefore important to provide meaningful support to policy makers. 562 563 RECEPTOR: Drawbacks associated to spatial and temporal averaging processes at the receptor 564 565 As clearly shown in Figure 5, spatial averaging processes lead to a loss of information. In our 566 example, a city average based SA would totally occult the city center SA. It would lead to a 567 strategy that mostly targets the background at the expense of the city center, where the high 568 concentration issues would not be solved. This is well illustrated by Amann et al. (2017) who  569 analyse the responsibility of the city of New Delhi on its air pollution, both at a city center hot-570 spot receptor and in terms of city average population exposure. In the first case, SA suggests 571 acting on local sources while in the second SA suggests acting on regional sources. Spatial 572 averaging drives the balance towards regional actions that will less effective in solving the 573 pollution issue at the city center. The larger the city, the more important this shift will be. As 574 illustrated by Figure 8, there is more than a factor 2 between city-averaged and hot spot 575 indicators. Similar considerations apply to temporal averages. Figure 7 clearly shows that yearly 576 average values hide the potential for effective local actions during wintertime and even more on 577 specific days. 578 579 Averaging implies merging, into one single number, locations and time instants that are 580 characterized by different and sometimes opposite SA. This may lead to strategies that will not 581 be efficient everywhere all the time. Whenever the final objective is to reduce a temporally 582 or/and spatially averaged indicator (e.g. average population exposure), strategies would gain in 583 efficiency with the following process: (1) perform SA and hierarchize the raw (not averaged) SA 584 results into homogeneous spatio-temporal clusters; (2) design strategies on the basis of these 585 clusters; (3) assess the strategy efficiency against the averaged indicator. The key is here to 586 design strategies on raw or clustered results rather than on averaged ones, to prevent information 587 loss. 588 589 Note that designing a unique strategy based on multiple SA results (point 2 above) does not 590 necessarily complicate the analysis, as these different SA will likely suggest action on different 591 sectors of activity that can be combined at the final strategy. 592 593

594
Although air quality has improved in Europe over the last decades, in great part thanks to 595 effective measures and consistent EU-wide legislation, pollution hot spots yet remain in many 596 European cities. The extent by which city emissions are causing these elevated urban pollution 597 levels is however still a subject of scientific discussion. Source apportionment represents a useful 598 technique to quantify the city responsibility but the approaches and applications are however not 599 harmonized, therefore not comparable, resulting in confusing and sometimes contradicting 600 interpretations. 601 602 In this work, we analyzed how different SA approaches apply to the urban scale and how their 603 building elements and parameters are defined and set. We identified the possible settings 604 associated to four key steps in SA: indicator, receptor, source and methodology. We showed that 605 different choices for these settings lead to very large differences in terms of results. In average 606 over the 150 European large cities selected as example, the choices made for the indicator, the 607 receptor, and the source each lead to an average factor 2 difference in terms of city 608 responsibility. These various options and the large differences that result, highlight the difficulty 609 of comparing results from different studies and stress the need to document the SA approach 610 with its related metadatathat documents the choices made for the key four steps. 611 612 This work advocates for the use of a harmonized nomenclature to support the comparability of 613 SA approaches. We propose the use of indexes and subindexes attached to the 4 key steps in any 614 SA approach in a harmonized way to uniquely document the approach and enable correct 615 interpretation of the results. We believe that the adoption of this nomenclature will provide 616 clarity to the scientific discussion on different results and enable the correct interpretation of the 617 results for policy applications. Even though this is applied to the specific case of PM2.5, the 618 concepts presented here can easily be generalized to other pollutants. 619 620 In the context of supporting urban air quality plans, the SA configuration and most setting 621 parameters are driven by the purpose of the AQ plan itself and by its associated constraints. 622 While environmental and/or health related considerations guide the choice of the indicator, the 623 spatio-temporal characteristics of the source are strongly correlated to governance aspects. In 624 other words, the source characteristics should reflect the governance levels to facilitate 625 interpretation. Finally, the recommended SA method should be based on "potential impacts", to 626 prevent misleading interpretations in terms of expected AQ plan outcome. 627 628 At the receptor level, temporal and spatial averaging processes lead to a loss of information, 629 especially when diverging SA results are aggregated into a single number. Averaging process, in 630 particular spatial, often lead to favor strategies that target background sources while neglecting 631 actions that would be efficient at the city center. In our 150 cities example, the impact of spatial 632 averaging leads to an average factor 2 difference in terms of city responsibility. Not only results 633 differ from one city to the other, and from one location to another in a given city, they also differ 634 through time. To cope with this variability, we recommend using non-averaged SA results for the 635 design of AQ strategies. Once clustered in homogeneous spatio-temporal classes, these can serve 636 to understand where and when actions are most efficient. When implemented, the efficiency of 637 abatement measures can then be assessed via spatially and temporally averaged indicator (e.g. 638 city average population exposure). 639 640 The responsibility of a city to its pollution is obviously city dependent. But even for a given city, 641 SA studies using different approaches and parameter settings will deliver very different 642 outcomes. It is important to note that a departure from the methodological recommendations 643 listed above, additional uncertainties and assumptions will most often lead to a systematic and 644 important underestimation of the city responsibility. We showed that in average over 150 645 European cities, departures in terms of source, receptor, and indicator may lead for each to a 646 factor 2 underestimation. This comes with important implications: if cities are seen as a minor 647 actor, plans will target in priority the background at the expense of potentially effective local 648 actions. 649 650 Future work will consist in comparing spatially/temporally averaged SA results with SA results 651 that are clustered in homogeneous spatio-temporal classes and assess the implications in terms of 652 AQ strategy. 653 654 In our example, the NO3 fraction attributed to the city depends on the ratio of the available NOx 934 precursor  Table A1: Formulations for the potential impacts, increments and tagging approach for the example presented in Figure  approach. This is due to the non-fulfilment of one of its underlying assumptions, i.e. the 957 lack of spatial homogeneity of the background which affects differently the rural and city 958 locations (indicated by "cc" and "bg" in Figure , respectively). 959 960