On the long term impact of emissions from central European cities on regional air-quality

Introduction Conclusions References

tions was carried out for the 2001-2010 period either with all urban emissions included (base case) or without considering urban emissions.Further, the sensitivity of ozone production to urban emissions was examined by performing reduction experiments with −20 % emission perturbation of NO x and/or non-methane volatile organic compounds (NMVOC).
The modeling system's air-quality related outputs were evaluated using AirBase and EMEP surface measurements showed reasonable reproduction of the monthly variation for ozone (O 3 ), but the annual cycle of nitrogen dioxide (NO 2 ) and sulfur dioxide (SO 2 ) is more biased.In terms of hourly correlations, values achieved for ozone and NO 2 are 0.5-0.8 and 0.4-0.6,but SO 2 is poorly or not correlated at all with measurements (r around 0.2-0.5).The modeled fine particulates (PM 2.5 ) are usually underestimated, especially in winter, mainly due to underestimation of nitrates and carbonaceous aerosols.
European air-quality measures were chosen as metrics describing the cities emission impact on regional air pollution.Due to urban emissions, significant ozone titration occurs over cities while over rural areas remote from cities, ozone production is modeled, mainly in terms of number of exceedances and accumulated exceedances over the threshold of 40 ppbv.Urban NO x , SO 2 and PM 2.5 emissions also significantly contribute to concentrations in the cities themselves (up to 50-70 % for NO x and SO 2 , and up to 60 % for PM 2.5 ), but the contribution is large over rural areas as well (10-20 %).Although air pollution over cities is largely determined by the local urban emissions, considerable (often a few tens of %) fraction of the concentration is attributable to other sources from rural areas and minor cities.For the case of Prague (Czech Republic capital), it is further shown that the inter-urban interference between large cities does not play an important role which means that the impact on a chosen city of emissions from all other large cities is very small.tion in road transportation and energy production.Carbon monoxide (CO) is a product of incomplete combustion and is dominantly emitted in African and Asian cities reflecting the older-than-average technologies used (Streets and Waldhoff, 2000).Non-methane volatile organic compounds (NMVOCs) are products of road transport and solvents use in North American and European cities; however in Africa and Asia, they originate mainly from domestic combustion (Denier van der Gon et al., 2010).SO 2 emissions are released mainly due to energy production and industry and they are relatively low in European cities.
Emissions of NO x and VOC are predominantly affecting photochemistry and depending on their ratio, the photochemical regime in and around cities is either NO x -controlled or VOC-controlled (Xue et al., 2014).When the concentrations of NO x are much higher than of VOCs (NO x -saturated case), the ozone (O 3 ) formation is controlled by the changes of VOCs: ozone increases with increasing VOCs while if 85 NO x increases, ozone decreases by titration.This regime is called VOC-controlled.On the other hand when VOCs/NO x ratio is high, ozone production depends on the change of nitrogen oxides: with increasing NO x concentration ozone increases as well and a NO x -controlled regime occurs (Sillman, 1999).The ratio NO x /VOC is usually high in North-American agglomerations, many eastern Asian cities and in European agglomerations like Athens, Paris, Milan or Berlin as well and ozone is usually titrated over these cities (Beekmann and Vautard, 2010).However, according to actual meteorological conditions, pollution from cities can be transported over large distances where the aged plume from the city mixes with additional VOC sources and can become NO x sensitive leading to ozone production (Beekmann and Derognat, 2003).The overall effect of city emissions on 100 ozone production/destruction can further depend on model's resolution.Thunis et al. (2007), analyzing Berlin, Milan, Paris and Prague found that while models with large spatial step usually predict ozone production due to emissions from cities, high resolution modeling studies attribute VOC-105 controlled regime to cities that leads to ozone destruction....part of the introduction deleted here!!!... Emissions of gaseous pollutants from cities can further perturb the aerosol burden.Sulfur dioxide, nitrogen (di)oxide and ammonia emissions lead, in presence of water vapor, 110 to formation of secondary inorganic aerosols: ammoniumsulfate-nitrate particles (Martin et al., 2004).The primary precursor for sulfate aerosol (PSO 4 ) formation is sulfur dioxide.Barth and Church (1999) investigated the sulfate formation due to SO 2 originating from Mexico City and cities from southeastern China, still the largest SO 2 emitter regions nowadays.They found significant perturbation of the global sulfate aerosol burden due to these two regions and cities located therein.NO x emissions do not affect only photochemistry (and the consequent ozone formation/destruction) but 120 also the formation of nitrate aerosol (PNO 3 ).If the meteorological conditions are favorable, nitrate oxide emissions from cities can enhance background nitrate aerosol levels significantly (Lin et al., 2010).Emissions of ammonia (NH 3 ) from cities are an efficient contributor to formation of sul-125 fate and nitrate aerosol (by forming ammonium-sulfates and ammonium-nitrates) and its importance in connection with cities emissions are studied recently by many (Behera and Sharma, 2010, and references therein).Generally, the thermodynamic system of ammonium-sulfate-nitrate-water solu-130 tion is rather complicated and its equilibrium state is highly dependent on the initial ratio of SO 2 -NO x -NH 3 given by their emissions, and the governing meteorological conditions (Martin et al., 2004), thus the contribution of different cities to these particles can be very variable.

135
Finally, organic gaseous material (volatile, intermediateand semi-volatile VOC) released from cities can contribute to formation of secondary organic aerosols and significantly enhance the total aerosol burden in urban, as well as the downwind environment, as showed by Paredes-Miranda et al.
Numerous studies were dealing with the impact of emissions from cities on air-quality over local, regional and even global scale.Many of them were based on measurements within and outside of the urban plumes from particu-145 lar cities (Freney et al., 2014;Lin et al., 1996;Gaffney et al., 1999;Molina et al., 2010;Kuhn et al., 2010;Wang et al., 2006).There has been also model based effort to estimate the cities fingerprint on the atmospheric chemistry across multiple scales: on global scale, Lawrence et al. (2007), But-150 ler and Lawrence (2009), Folberth et al. (2010) and Stock et al. (2013) gave estimates on the city emissions impact on the surrounding environment.On regional scales, many studies focused on European urban centers, especially those in the Mediterranean region (e.g.Im et al., 2011a, b;Es-155 cudero et al., 2014;Finardi et al., 2014), but also covering London and the Ruhr area (Hodneborg et al., 2011), or Paris (Skyllakou et al., 2014;Markakis et al., 2015).The importance of multi-model modelling approach for investigating the megacities impact on air-quality and climate was ana-160 lyzed in detail by Baklanov et al. (2010) in the framework of the European FP7 project MEGAPOLI.Within another European project, FP7 project CITYZEN, Im and Kanakidou (2012) investigated the impact of emissions from eastern Mediterranean megacities, Athens and Istanbul.

165
Here we present a study that is inspired by a wider effort to describe quantitatively the urban/climate/air-quality interactions over the target area of central Europe.Previously, Huszar et al. (2014) presented the impact of urban landsurface forcing on climate.Here, we link to this study and look 170 at a further aspect of the urban impact on environment: we aim to provide a chemistry transport model based estimate of the long term impact of emissions from cities in central Europe on the regional air-quality.The study brings four novelties: (1) the above listed studies over Europe focused either on the region of Mediterranean, which encounters dry warm climate, and/or on large megacities only (London, Paris, Is-tanbul, Athens).In contrary, our target region is central Europe with different climate (temperate maritime to continental) and without any megacity.(2) Previously, model based estimates of urban emission impact over Europe considered relatively short time periods (1-2 months) often separately for winter and summer seasons (e.g.Im et al., 2011a;Im and Kanakidou, 2012;Finardi et al., 2014).These periods are however short to eliminate the potential influence of specific 185 meteorological conditions during those time periods.Therefore, we have proposed to conduct continuous, 10 yr long simulations which decreases the uncertainty originating in the driving meteorological conditions.This choice was preferred also by Katragkou et al. (2010) and Zanis et al. (2011) or Markakis et al. (2015) (3) Most of the above listed regional studies focused on only one or two megacities (and their impact).Here we consider all large city within the region in focus.This is an important step, as the combined impact of emissions from all cities may, due to chemical nonlinearities, significantly differ from the cumulative impact evaluated separately for each city.(4) Our study evaluates the impact on policy relevant metrics that include also exceedances above a threshold, instead of evaluating simply seasonal averages that often lack information on extreme pollution.

200
The study has two main goals: (1) to evaluate the presentday contribution of city emissions to the regional air pollution over central Europe.(2) To calculate the potential impact of mitigation strategies by testing the regional fingerprint of urban emissions reductions.The possible climate impact of 205 the presented urban-induced chemical perturbation of the atmosphere will be addressed as well in future paper.Within the first goal, the study tries to answer two questions: (a) what is the contribution of urban emissions to the air-quality over rural areas further from cities, (b) to what extent is the urban air-quality influenced by non-urban emissions.Regarding the second goal, the question asked is which urban emission reductions are the most effective in controlling regional scale ozone pollution.
The impact will be evaluated in terms of surface concentrations and exceedances of key gaseous pollutants (O 3 , NO 2 , SO 2 ) and fine aerosol (size < 2.5 µm, PM 2.5 ).

Emissions
Emissions used in the study are the TNO emissions prepared for year 2005 in the framework of the FP 7 MEGAPOLI project (Kuenen et al., 2010).This high resolution (1/8

225
For the purpose of calculating the impact of urban emissions, emission mask had to be built for selected cities.These were built according to the administrative borders of the particular city in combination with the subgrid urban land-surface data used in Huszar et al. (2014), originally extracted 230 from the Corine2006 database (EEA, 2012).The selection of certain cities, in general, comprises cities considered to be large within the particular region.As such we chose the threshold of 500 000 inhabitants representing a "large" city.This threshold was reduced to 200 000 inhabitants over se-235 lected regions (Czech Republic, Slovakia, Hungary, Romania, partly Poland, Austria, Italy).Figure 1 presents the distribution of the annual emissions over selected cities for the main pollutants: CO, NMVOC, NO x , NH 3 , SO 2 and PM 2.5 .It clearly reveals the emission density differences between 240 the urban centers and suburban areas and that the emissions are mostly comprised of CO, NO x and NMVOC, which can reach 500, 100 and 100 Mg km −2 yr −1 , respectively, especially in urban centers.
Figure 2 plots the absolute annual emissions for the whole 245 domain, for all the cities and for six selected cities (namely, Vienna, Budapest, Berlin, Prague, Munich, Warsaw).The plot shows that in most of the sectors, urban emissions form roughly 10 % of all emissions, while they cover slightly more than 3.5 % of the area of the focused region.The sector to 250 which they contribute less is the agriculture where they emit less than 0.5 % of all emissions.
In general, road transportation is the sector contributing most to urban emissions, followed by non-industrial combustion in Central Europe.However, large differences are iden-255 tified between cities.While emissions from sector SNAP 8 that include ship and airport traffic are generally small, in selected cities with major international airports or intense vessel traffic (on rivers), these can be of comparable magnitude with the road transportation (e.g.Munich), or even exceed 260 road traffic (Vienna).
The most contributing substance to city emissions is carbon monoxide with an approximately 56 % contribution (in mass units) in average, followed by NO x and NMVOC both with around 14 %.SO 2 makes 12 % of all the city emissions 265 in average, being somewhat higher in eastern European urban centers (almost 20 % in Budapest, and 18 % in Warsaw).
3 Models and experimental design 3.1 The regional climate model RegCM4.2270 As a meteorological driver, we used the regional climate model RegCM version 4.2 (hereafter referred to as RegCM4.2) developed by The International Centre for Theoretical Physics.Although the up-to-date version of RegCM is 4.5 (June 2015), the development of the modeling tools 275 for this study started earlier when the newest version was 4.2.RegCM4.2 and its evolution from RegCM3 is fully described by Giorgi et al. (2012).Its dynamical core is based on the hydrostatic version of the NCAR-PSU Mesoscale Model version 5 (MM5) (Grell et al., 1994).The radiation is solved  within the Community Climate Model version 3 (CCM3) (Kiehl et al., 1996).The large-scale precipitation and cloud processes are calculated following Pal et al. (2000) and for convection parameterization we use the Grell scheme (Grell, 1993) using the Fritch and Chappell (1980) closure assumption in this study.RegCM4.2includes two land-surface models: Biosphere-Atmosphere Transfer Scheme (BATS) originally developed by Dickinson et al. (1993) and the CLM3.5 model (Oleson et al., 2008).In this study, the BATS scheme is activated.The single layer urban canopy model coupled to RegCM4.2 introduced by Huszar et al. (2014) was not applied assuming that the urban-meteorological influence on the emissions impact will be minor and further, to meet the computational demand of long climate simulations.

295
The chemistry simulations were carried out with the chemistry transport model CAMx (version 5.4).CAMx is an Eulerian photochemical dispersion model developed by ENVI-RON Int.Corp. (http://www.camx.com).CAMx includes the options of two-way grid nesting, multiple gas phase chem-300 istry mechanism options (CB-IV, CBV, CBVI, SAPRC99), evolving multi-sectional or static two mode particle size treatments, wet deposition of gases and particles, plumein-grid (PiG) module for sub-grid treatment of selected point sources, Ozone and Particulate Source Apportionment Technology, mass conservative and consistent transport numerics, parallel processing.The ISORROPIA thermodynamic equilibrium model (Nenes and Pandis, 1998) is implemented in CAMx to calculate the composition and phase state of an ammonia-sulfate-nitrate-chloride-sodiumwater inorganic aerosol system in equilibrium with gas phase precursors.A detailed description of the model (the version used here) can be found at http://www.camx.com/files/camxusersguide_v5-40.pdf.

The coupled model RegCMCAMx4
To achieve the goals of the study, a coupled system was designed consisting of RegCM4.2 and CAMx (denoted RegCMCAMx4) following the technique of online access coupling defined by Baklanov (2010).It represents an interactive two-way coupled modeling framework where chemistry is driven by the climate model and the calculated concentrations of the radiatively active gases and aerosols are fed back to the climate model's radiation code.
RegCMCAMx4 is more advanced version of the original RegCMCAMx couple described by Huszar et al. (2012).The update interval for the meteorology from RegCM remained 1 h which is sufficient (Grell and Baklanov, 2011).However, the original update interval for the species in the radiation code of 6 h was too coarse for describing the diurnal species evolution, therefore it has been reduced to 1 h as well.
The original RegCMCAMx considered only the direct effect of sulfates and primary organic and black carbon.RegCM-CAMx4 introduces the indirect effect of secondary inorganic aerosols (both sulfates and nitrates).For sulfates, it follows the work of Giorgi and Qian (2003) where the cloud droplet 335 concentration and effective droplet radius is modified according to the aerosol concentration.For nitrates both direct and indirect radiative effects are computed with the same method as for sulfates but with slightly modified optical properties following the works of McMeeking et al. (2005) and Wang RegCMCAMx4 further replaces the O'Brien (1970) method for calculating the coefficients of vertical turbulent diffusion (which is required by CAMx) with the newer Byun (1999) scheme (as used in CMAQ model), which provides 345 better agreement of model results with measurements, as shown by Eben et al. (2005) in a CAMx application over the same region and at similar horizontal resolution like in this study.
The added value of using an online coupled climate-350 chemistry modelling system is the possibility to calculate radiative feedbacks and impacts on temperature (and climate in general).We also assumed, based on previous validation studies involving RegCM and CAMx that the capability of these models reproducing the state of the atmosphere (both 355 meteorology and chemistry) will not change significantly if coupling them online with respect to the case when they are coupled offline (e.g.Huszar et al., 2012).

Experimental set-up
The period of 2001-2010 was chosen to analyze the present 360 day impact of urban emissions on the air-quality over central Europe.Calculation with RegCMCAMx4 were carried out on 10 km × 10 km horizontal resolution domain centered over Prague, Czech republic of 160 × 120 × 24 (in x, y, and z direction) gridboxes for the climate model up to 50 hPa, 365 while the chemistry model was integrated only on the lowermost 16 levels (approximately up to 300 hPa or 9000 m).
The integration time step for the climate model was 30 s and 10 min for the chemistry model.The ERA Interim reanalysis (Simmons et al., 2010) was 370 chosen as driving meteorological conditions while for the chemical model, chemical boundary conditions (CBC) were taken from a similar 10 year run performed by RegCM-CAMx4 over a larger, 30 km × 30 km domain covering whole Europe.

375
As already mentioned, the TNO 2005 emissions were chosen to cover the studied period.Over the focused region, their resolution is about 7 km × 7 km, so sufficient for a 10 km × 10 km computational grid.TNO are sector based annual emission data were first regridded into the model grid.

380
Than for each sector, specific temporal disaggregation factors and NMVOC speciation profiles were used to decompose the annual sums into hourly emissions following the inventory (Winiwarter and Zueger, 1996).The temporal profiles they human activity profiles regarding transport, combustion, production etc. Biogenic emission of isoprene and monoterpenes were calculated following Guenther et al. (1993).
A number of experiments was carried out to examine the effect of city emissions on the regional air quality.These are summarized in Table 1.The total impact of all city emissions is evaluated as the difference between experiments 05BASE and 05ZERO (the "05" means that the 2005 emission were used).We were also interested in the impact the emissions from all other cities have on a selected city.To achieve this goal, we performed a run where all city emissions are removed expect those from Prague.Apart from the total impact, it is also of interest to see how the individual species emitted contribute to the overall impact.We therefore evaluate also the partial impact of major gaseous pollutants, namely NO x , NMVOC.As the interest of policy makers is to estimate the consequences of possible emission reduction in present day cities, we propose to evaluate this partial impact in a framework of a sensitivity test where the emissions of the above mentioned pollutants will be reduced by 20 %.
For the sensitivity runs, no radiative feedbacks were calculated and the same meteorological conditions were thus used as a driver for these simulations.The assumption made here was that the main driver for the air-quality changes are emissions, as the meteorological impact of the online coupled ozone and aerosols are expected to be small, having small feedback back to the chemistry.

Model validation
In order to justify the model's applicability for the presented goal, a detailed quantitative validation is provided for surface concentrations of O 3 , NO 2 , SO 2 and PM 2.5 .We also included a minimal validation of the meteorological results: the near surface temperature is compared with the European EOBS climate dataset (Haylock et al., 2008) that is currently extended until year 2014.A detailed validation for the meteorological output is planned in a follow-up study, which intends to present the meteorological feedbacks of the presented chemical perturbation induced by urban emissions.Fig. 3 presents the comparison of model seasonal near sur-425 face temperatures averaged over the 2001-2010 period with the E-OBS gridded dataset.It indicates a negative model bias, which is highest in spring (around -1 to -2 K over large areas) and lowest in summer (0 to -1.5 K).This is attributable to overestimated cloudiness in RegCM, as already concluded by Huszar et al. (2014), who used the same model and set-up.Over mountainous areas like Alpes, southeastern Carpathians, biases reach values from -5 to 5 K, however, this is most probably related to coarse model representation of specific terrain features (valleys, ridges etc.) that highly determine the surface temperatures.A striking feature is the overestimation of temperature on eastern edge of the domain.As the boundary-close cells of the domain are strongly influenced with the boundary conditions, this may indicate that the reanalysis used (ERA Interim) is somewhat warmer than the 440 gridded observational data.
For the chemical validation, the AirBase version 8 data (http://www.eea.europa.eu/data-and-maps/data/airbase-the-european-air-quality-database-8) provided by the European Environmental Agency, is used.We se-445 lected only rural background stations which are more consistent with the model provided value that represents a 10 km × 10 km average.Further the stations ale filtered to exclude high elevation stations (above 2000 m).In the end, 328 stations for O 3 , 280 for NO 2 , 200 for SO 2 and 53 for 450 PM 2.5 were selected for comparison with model results.For the gaseous pollutants, hourly, daily and monthly averages are considered while for aerosols only daily and monthly data.The validation is done separately for winter (DJF), summer (JJA) months and for the whole year.The statistical 455 measures evaluated were the correlation coefficient (r), root mean square error (RMSE), normalized mean square error (NMSE), the ratio of standard deviations (σ r , calculated as σ observation divided by σ model ) and fractional bias (FB), as defined by Borrego et al. (2008) and adopted by 460 Juda-Rezler et al. (2012).They identified these metrics as the most important in assessing air-quality model accuracy.Choosing these metrics further ease the comparison of the RegCMCAMx4 model performance with its former version presented in Huszar et al. (2012) who applied the same 465 metrics.The experiment 05BASE gave the base for the validation.
The above mentioned statistical measures are collected in Table 2 for O 3 , NO 2 , SO 2 and PM 2.5 .3-3-3 columns are dedicated for hourly, daily and monthly data averaged over 470 the whole year, DJF and JJA, respectively.
The average monthly and hourly cycles for DJF and JJA were selected, which provide measure of the model's ability to capture the basic chemical climatology of key-species concentrations.Figure 4 plots the average monthly variation 475 (left column) of the gaseous species O 3 , NO 2 and SO 2 .The middle and right column provide the average diurnal cycle of these species for DJF and JJA, respectively.
Further, the monthly mean values of PM 2.5 and its major components were compared to observations (Fig. 5) dis-480 tinguishing between DJF and JJA.For PSO 4 and PNO 3 , measurements from the already mentioned AirBase database were used.For carbonaceous aerosol, the monthly data from the BC/OC measurement campaign data described by Yttri et al. (2007) covering July 2002-June 2003 were used 485 with the assumption that the basic climatology of these data is similar to the 2001-2010 average.These measurements considered BC/OC from PM 10 aerosol (size < 10 µm), our model (CAMx) that uses a two bin approach (fine and coarse particles), calculates them as fine particles (size < 2.5 µm).We applied the factor of 0.8 to the measured values to estimate the PM 2.5 fraction.This value is compiled from Chen et al. (1997), Offenberg and Baker (2000) and Samara et al. (2014) as an average of different seasons and character of the measurement site (cold vs. warm season and urban vs. rural).
We assessed the model bias further for urban stations as well, although it is well accepted that these stations are not suitable for standard chemistry transport model evaluation (at such resolution as in this study).We selected 10 urban background stations for Berlin, Budapest, Frankfurt, Katowice, Ljubljana, Milan, Münich, Prague, Vienna and Warsaw.The same statistical metrics were chosen as above but only for hourly (daily) averages for gases (particle matter).The results are collected in Table 3.

Ozone
The correlations of modeled ozone data with measurements are highest for the monthly means reaching 0.77 when considering the whole period.It is generally lower in DJF than during JJA and decreases for shorter averaging periods.For the hourly means, it is about 0.57 for the whole period, and 510 about 0.41 and 0.53 for DJF and JJA.Relatively high RMSE and NMSE values are modeled for the hourly values and get smaller for longer averaging period (around 17 µg m −3 for the monthly means).The ratios of standard deviation are slightly higher than 1 indicating that the measured ozone val-515 ues have higher variability than the modeled ones.In terms of fractional bias, the model underestimates ozone for both DJF and JJA (FB being around −16 and −2 %) giving an overall underestimation of FB = −4.3% for the whole period.
The negative ozone bias is clearly seen in terms of  The correlations for individual cities are lower in general but the RMSE are of similar value.FBs indicate a slight overestimation in JJA, in contrary with the rural station values.
The DJF negative bias is stronger for urban stations than for their rural counterparts.

Nitrogen dioxide
The modeled NO 2 values are less correlated with the measured ones than in case of ozone.Again, they are highest for the monthly means (0.68, 0.59 and 0.62 for the whole period, for DJF and JJA months, respectively).The RMSE values are lower than for O 3 , being highest for DJF and for the hourly means (around 19 µg m −3 ).The observation-model agreement is in general best for JJA.However, during JJA, the model exhibits much larger standard deviation than the measured values, while during DJF, the ratio of standard deviations is close to 1.In general, the model tends to underestimate both DJF and JJA NO 2 values with FB values around −14.8, −23.7 and −10.5 % for all the months, for DJF and JJA, respectively.This is also well described by the average monthly and diurnal variation plots (Fig. 4, middle row): the DJF NO 2 values are underestimated by up to 5 µg m −3 but 550 a fair agreement is modeled during late spring to early autumn months with only a slight underestimation around 1-2 µg m −3 .The diurnal course for DJF is captured in the model as well, but it peaks around 18:00 UTC compared to 555 20:00 UTC in observations.The diurnal amplitude is of comparable magnitude due to both maximum and minimum values lower in model than in observations.Especially the nighttime NO 2 values are underestimated during DJF (by more than 5 µg m −3 ).In JJA, the observations reveal two peaks, 560 in the morning and early evening hours, but the model reproduces (although poorly) only the evening peak with an overestimation around 2 µg m −3 .
Similarly to ozone, the observation-model correlation for individual cities is much smaller, or there is no correlation 565 at all.The model remains negatively biased but it is highest for JJA.The variability in urban stations is much underestimated, in opposite to the rural stations above.
Table 3.Comparison of model data with measurements: city based evaluation of the correlation coefficient (r), root mean square error (RMSE; in µg m −3 ), normalized mean square error (NMSE), the ratio of standard deviations (σr) and fractional bias (FB; in %) for hourly (for gases) and daily (for PM2.5) averages for winter (DJF) and summer (JJA) months and for the whole year ("annual") for pollutants O3, NO2, SO2 and PM2.5.The overestimation of SO 2 in DJF is apparent from the average monthly and diurnal cycle in DJF (Fig. 4, bottom left).The model predicts much larger concentrations than the observed ones especially in December, and, in terms of the hourly variation, during mid-day.During JJA, SO 2 is underestimated by 2-3 µg m −3 , especially during noon hours.
From an urban station perspective, the model is very poorly correlated with measurements (correlations not exceeding 0.4) and the values are strongly overestimated (often by more than 100 % in terms of FB) in both JJA and DJF, in contrary to model performance evaluated for rural stations.

PM 2.5 and components
The modeled PM 2.5 values are low correlated with measurements, higher for monthly values and for DJF (up to 595 r = 0.45).In terms of RMSE and NMSE, the model performs better during JJA, especially for the monthly means.In DJF, the standard deviation of the observed values is largely underestimated by the model, while in JJA, only half of the observed variability is predicted.PM 2.5 is largely under-600 estimated in DJF (with around −90 % fractional bias) and slightly overestimated in JJA (FB = 17 %).This is further well seen on the monthly scatter plot in Fig. 5 (upper row, left).The DJF values (orange) do not exceed 20 µg m −3 in DJF in the model, while they often reach 40 µg m −3 in ob-605 servations.In JJA, the range of values is similar, but the correlations are low, as already seen in Table 2.
The monthly scatter plots of individual components of the PM 2.5 aerosol are plotted in Fig. 5 as well.A relatively good agreement is achieved for PSO 4 during DJF with values 610 ranging in both measurements and model up to 8 µg m −3 .However, sulfates are often over-predicted by the model in JJA.A different situation occurs for PNO 3 , where the model exhibits a negative bias, especially for DJF, when values often above 6 µg m −3 are measured, in contrary with the mod-615 eled monthly values.The modeled BC and OC fractions of PM 2.5 are usually underestimated with a few exceptions.In case of OC, a slightly better agreement is achieved for DJF.
In JJA, however, the model is unable to reproduce values over 5 µg m −3 , often seen in measured data.

620
Over urban station, the model is very low correlated with measurements and tends to underestimate the fine particulate matter concentrations in JJA, in contrary to the rural stations.In terms of other metrics, the model performs similarly than over rural stations.

Impact of city emissions on air-quality
The impact of urban emissions from large cities on the regional air quality is evaluated in terms of selected air quality measures.For quantifying the exposure of the ecosystems, particularly crops, to elevated ozone levels, a widely 630 used measure, the accumulated exposure over the threshold (AOT), introduced by Fuhrer et al. (1997), can be used.In this study, we evaluated the present AOT for the threshold of 40 ppbv for crops and forests (AOT40crop/forest) where the integration is done from May to July and from April to September, respectively.Further, the number of exceedances above a certain threshold is evaluated for daily maximum 8 h running ozone mean, the hourly NO 2 and SO 2 and the daily SO 2 values.Finally, the mean JJA O 3 , mean DJF and annual SO 2 and the mean annual PM 2.5 surface concentrations are considered.These measures are established in the EC Directive on ambient air quality and cleaner air for Europe (2008/50/EC) and are implemented also in the Czech legislation.These are summarized in Table 4.
In further, we will present the spatial distributions of the

645
(1) absolute change of the chosen metrics by the introduction of city emissions (calculated as experiment 05BASE minus 05ZERO) and the (2) relative change which is calculated differently for ozone and for other pollutants.In case of ozone, which (as we will see in further) both increases and decreases, the change is shown relative to the no-urban emission case (05ZERO).For all other species, the change is shown relative to the all-emission case (05BASE), i.e. we are interested in the relative contribution.We also calculated the all-city-average of the maximum impact which coincide with the location of the cities themselves, summarized in Table 5.

Ozone
The impact of urban emissions on the average JJA surface ozone concentration (Fig. 6) is characterized by a clear reduction peaking over city centers from −4 ppbv over smaller 660 cities up to −12 ppbv over western Germany urban agglomerations (e.g.Rhur area).This corresponds to more than 30 % ozone decrease.Further inland or over the southern part of the domain, the influence of city emissions is smaller in relative sense with change around −20 % while the city influ-665 ence peaks around −4 to −6 ppbv.Over rural regions, JJA ozone tends to decrease slightly for the western part of the domain.However, over southern and eastern part of the domain, the mean JJA ozone concentrations increase due to city emissions by up to 0.5 ppbv, representing an 1 % in-670 crease.The average decrease over cities is −5.1(±3.3)ppbv, or −34.1(±18.3)%.
The impact on AOT40 values are, similarly to the JJA average ozone, characterized by a significant decrease over and around cities up to −4000 ppbv h (for both impact on 675 crops and forests) or −40 to −60 % in relative sence.An opposite impact is modeled over areas neighboring or even further from cities. City emissions increase AOT40 values up to 600 ppbv h over many regions, meaning an 5-10 %.In the vicinity of many cities (Milan, Zagreb, War-680 saw), the both AOT40s increase by up to 1000 ppbv h while the above mentioned decrease occurs just a few 10 kms towards the city center.The averaged decrease over cities is −1800(±1300) ppbv h or −29.1(±18.3)% for crops and −2460(±1800) or −30.7(±18) for forests, respectively.

685
A similar picture to previous ones is obtained when evaluating the number of days with maximum 8 h ozone greater than 120 µg m −3 .City emissions clearly decrease this number over and near cities (by up to −15 days yr −1 ), but further from them, the increase of extreme ozone days is evi-690 dent (up to 6 days yr −1 ).This corresponds to from 10 % increase over many parts of western Europe up to more than 40 % enhancement in central Europe with selected regions encountering even higher, up to 100 % increase.The all-cityaverage decrease of the number of exceedances was calcu-695 lated to −2.9(±4.0)or −28.7(±39.8)%.

Nitrogen oxides
Due to systematic negative bias the model was unable to predict exceedances over 200 µg m −3 , therefore only the impact of city emissions on the annual NO 2 concentration is 700 shown (Fig. 7).The annual mean change can be as high as 30 µg m −3 over the cities themselves, making around 50 % contribution to the absolute values over the western part of the domain, while over Central European cities (e.g.Berlin, Warsaw, Vienna, Budapest) it can reach 70 %.Over areas 705 further from cities, the contribution quickly decreases making less than 10 % of the absolute NO 2 values but remaining above 5 % over much of the domain.Comparing the absolute  Table 5.The averaged maximal impact of urban emission on air-quality (in terms of the quantities from Tab. 4) over cities.The 2nd and the 5th column stands for the absolute impact, the 3rd and the 6th column for the relative impact (for ozone related quantities) and contribution (for other species).The standard deviation of the all-city-average is included as well.NO 2 change over cities and the JJA ozone decrease in previous figure it is clear that larger urban NO 2 perturbation the more pronounced is the ozone suppression.Indeed, linear fit between these two quantities (the plot not shown here) has a coefficient of determination R 2 = 0.86 indicating a strong link.The averaged urban induced NO 2 increase over cities is 12.8(±6.8)µg m −3 , corresponding to 42.7(±18.3)% contri-715 bution to the total value.

Sulfur dioxide
Figure 8 shows the impact on SO 2 .The annual mean increase due to city emissions reaches 50 µg m −3 over Eastern European cities and can be as high as 12 µg m −3 over Western Europe (e.g. the Ruhr area).In relative manner, the contribution peaks at 80 % and is above 70 % over many cities all over the domain.However, it can stay higher even further from the cities: over large parts of northern Germany, the contribution to the annual SO 2 values is between 10 and 20 %.A similar picture is revealed when looking at the DJF SO 2 impact with up to 20 µg m −3 increase over the cities themselves giving relative contribution of similar magnitude as in case of the annual means (up to 80 % in cities).The all-city-average increases of annual and winter values are 5.5(±6.3)and 8.2 (±9.8) µg m −3 , or, 41.4(±24.4)and 38.6(±23.8)% as contributions, respectively.The urban emissions contribution to the daily SO 2 exceedances over the 125 µg m −3 threshold is again highest over cities making more than 20 days yr −1 contribution.Over Eastern Europe, larger regions are affected with city emissions increasing the number of exceedances by 1-2 days.In relative sense, larger areas around cities are affected by higher daily SO 2 values often reaching 90-100 % meaning that vast majority of the high SO 2 occurrences are due to emissions from cities.Even further from cities, especially over Eastern Europe, up to 10 % of all the elevated daily SO 2 values are due to city emissions.The impact on hourly SO 2 exceedances is again highest over cities (so in line with the emissions) with up to 100 h yr −1 contribution to 745 the absolute number of exceedances and, in general, Central and Eastern Europe is affected the most.Increases in hourly exceedances due to urban emissions are modeled at even larger distances from cities (similarly to daily exceedances) up to 1-2 h yr −1 . in relative numbers, the contribution is around 80-90 % over cities, but quickly reduces below 20 % further from them.The averaged increases in exceedances over cities are very variable: 8.6(±18.9)and 15.1(±43.0)for the daily and hourly averaging period, giving 40.1(±44.8)% and 23.8(±40.0)% relative contribution.

PM 2.5
According to Fig. 9, urban emissions increase annual PM 2.5 levels by 4-8 µg m −3 over cities with the highest impact over the Ruhr area and Warsaw (up to 15 µg m −3 ).These cor-respond to about 20-60 % contribution to the total PM 2.5 760 levels.Above rural areas further from cities, the impact goes rapidly below 1 µg m −3 .However, in a relative manner, it remains around 5-10 %, e.g. over Northern Germany or Central Europe.The averaged urban induced PM 2.5 increase over cities is 4.2(±3.7)µg m −3 , corresponding to 765 24.3(±12.8)% contribution to the total value.

Impact on a particular city
It was seen in Figs.6-9 that urban emissions impact airquality mainly over the cities themselves and the influence on rural air is much smaller.The question is how the emis-770 sions from other cities contribute to the impact over a particular city, or in other words, what fraction of the total impact (due to all cities) is attributable to the impact of the local emissions.To examine this, we selected the city of Prague lying in the center of the domain and representing a middle 775 sized city from the domain.Figure 10 presents the total impact (i.e. from all urban emissions; left column) and the impact of emissions from the rest of the cities not considering the emissions from Prague (right column).We evaluated this in terms of the average quantities from Table 4 (annual, DJF 780 and JJA means), showing only the relevant part of the domain.The total impact (right column) actually corresponds to a detail of the impact presented in Figs.6-9.and gives −7 ppbv, 15, 6, 8, 3 µg m −3 as the maximum change over Prague due to all urban emissions for JJA O 3 , annual NO 2 , 785 annual and DJF SO 2 and annual PM 2.5 .The same quantities from the right column give, for Prague, approximately 0.2 ppbv, 0.5, 0.4, 0.6, 0.5 µg m −3 .This represents about 3, 3, 7, 7, 16 % of the impact due to all urban emissions.

Sensitivity experiments 790
The response to possible urban emission reductions of selected ozone related measures presented in Table 4 is evaluated here.The results are presented in Fig. 11.Reducing city NO x emissions by 20 %, due to limited reaction with NO, JJA ozone concentrations are enhanced by around 1.5-795 2 ppbv over city centres but O 3 increases over larger areas as well, although by much smaller magnitude (0.1-0.2 ppbv increase over Western Germany).The AOT40s responded in similar manner: due to reduced ozone titration, elevated AOT40s are modeled over and around cities (by up to 500-800 1500 ppbv h).However, over Central and Southern Europe, AOT40s tend to slightly decrease (by up to −200 ppbv h) with decreasing urban NO x emissions.The reduced titration is evident on the change in the number of ozone exceedances which increases over cities often by more than 2-805 3 days yr −1 .Further from urban centers, however, less NO x emitted tend to decrease ozone exceedances (especially over Central Europe, by up to −1.5 days yr −1 in average).
While NO x reduction increased ozone over cities, and caused small decreases elsewhere, especially in terms of  The simultaneous reduction of both NO x and NMVOC leads to similar changes in JJA ozone means values: only the peak changes over cities -caused by decreased titration, are smaller, up to 1.5 ppbv.In terms of AOT40crop/forest, 820 reduction of urban NO x + NMVOC emissions leads to increases over cities, but again by a smaller magnitude than in case of purely NO x reduction.On the other hand, over rural areas, AOT40s decreased more than due to NO x reduction alone.The increases of 8 h ozone exceedances due to 825 decreased NO x + NMVOC go up to 2-3 days yr −1 , which is again less than for NO x reduction.On the other hand, again, the decrease in the number of exceedances over rural areas is slightly larger in case of simultaneous NO x + NMVOC reduction than due to NO x emission reduction only.

5 Discussion
The validation of the modeling system The validation showed that the modeling system captures the observed annual cycle of ozone with negative bias encountered in each months except late summer and autumn.

835
The chemical boundary conditions used by our model were taken from a 10 yr simulation from a larger domain, which however was forced with time invariant, spatially constant boundary conditions (40 ppbv) and this artificial constraint could propagate to the inner domain.Katragkou et al. (2010) 840 showed that the final ozone levels greatly depend on the imposed boundary conditions.This constant constrain is also evident in the underestimation of the standard deviation for each averaging period, especially for DJF.The diurnal cycle underlines the monthly model bias, giving lower hourly values in DJF and better agreement in JJA, but with an underestimation of the JJA daily maximum values.The lower afternoon values of ozone in JJA can be attributed also by higher afternoon and evening NO 2 values, as a result of the NO+O 3 reaction.A very similar result is provided by Huszar et al. (2012) both in terms of monthly and hourly variation.
Recently, Akritidis et al. (2013) were investigating the impact of CBC on the simulated ozone concentrations using the same two models as in our study (an offline couple of models RegCM and CAMx).They found a clear improvement in the correlation coefficient when using global chemistry model (ECHAM5/MOZART) based CBC.Their correlation of monthly ozone values using time/space invariant CBCs is 0.74 that compares very well to our value (0.77).Introducing the MOZART based CBCs, the correlations increased by often more than 0.1.
Compared to Huszar et al. (2012), our modeling system performs better in terms of correlation, with r = 0.6 and r = 0.67 for hourly and daily values against 0.51 and 0.53 in the later study who used an earlier version of the RegCM-CAMx coupled system.This improvement probably lies in the duration of the data of comparison: in a 10 year timeframe, the main drivers of the variability are the diurnal and monthly variations which contribute to overall correlation in a significant way (Hogrefe et al., 2001).Zanis et al. (2011), who used the same models in offline couple for also a 10 yr experiment over Europe, achieved similar values of correlations.In general, our model performs for ozone better during JJA, when photochemistry is more intensive.This is true also for RMSE, NMSE and FB.Katragkou et al. (2010) and Zanis et al. (2011) came to the same conclusion.
Very low correlations are achieved in case of NO 2 , especially for the hourly values.In general, it is difficult to achieve higher degree of agreement in case of precursor species for at least two reasons.The driving meteorology, 880 which greatly influences the hour-to-hour evolution of the NO 2 concentrations, is a result of a 10 yr climate model run.The climate model does not need to accurately reproduce the hour-to-hour, day-to-day weather pattern; however it has to reconstruct the climate close to reality in terms of aver-885 aged quantities and capability to capture extremes (Halenka et al., 2006).Further, the emission decomposition into hourly values is based on numerous assumptions about the typical temporal evolution of a certain activity sector and the actual emissions may differ for a particular hour.At last, ozone pre-890 cursor species are modeled always with higher degree of uncertainty with great differences between models and set-ups (including chemistry mechanism) while they give very similar results in terms of final ozone concentrations (Kuhn et al., 1998).

895
Another striking feature is the underestimation of the modeled NO 2 values.Huszar et al. (2012), who encountered a similar negative model bias, concluded that one reason lies in the overall underestimation of emissions and in the suppressed NO to NO 2 conversion due to volatile organic 900 compounds in CB-IV mechanism.Our configuration invoked the CB-V chemistry mechanism which was to remove this erroneous feature in the earlier version of the CB mechanism (Sarwar et al., 2008).However, the negative bias persists in our simulations, which in consequence could mean 905 that the emissions are probably underestimated for the region modeled.Further, seen in the hourly plots, the underestimation mainly occurs during night-time similarly as in P. Huszar et al.: Urban emissions impact on air-quality Huszar et al. (2012).Many other studies argued that chemistry in air-quality models performs less biased during daylight (Zanis et al., 2011).At last, biomass burning emission were not accounted for in our simulations while it is an important contributor to NO 2 burdens, especially for southern stations (Baldasano et al., 2011).
The strong DJF overestimation of the SO 2 levels in Huszar et al. (2012) Huszar et al. (2012).Schaap et al. (2004) obtained comparable values as well and at-tributed this negative model bias to deficiencies in describing coating processes which are burdened by large uncertainties 965 and directly determine the BC lifetime (BC has to become hydrophilic to get washed out by the wet deposition).Even more striking underestimation occurs for the organic carbon, although this bias is reduced compared to Huszar et al. (2012).This is probably due to different emission data used 970 here for primary OC.However, the largest source for low modeled OC values probably lies in: (1) in modeling the gasto-particle partitioning that is affected with uncertainty with large number of tunable parameters (Simpson et al., 2007), (2) disregarding biomass burning aerosol that occurred in 975 2003 in eastern Europe affecting the measurements of Yttri et al. (2007).
As expected, for selected urban stations, our modeling system is, in general, less accurate, especially in terms of correlations.Urban station are often influenced by local or 980 nearby emissions sources, far below the models spatial resolution.The relatively coarse input emissions data cannot resolve the variability of these sources leading to much worse observation-model agreement compared to rural background stations.In case of ozone, the model over urban stations 985 is positively biased in summer, which can be explained by the instant dilution of concentrated urban emission into the 10 km model grid which tends to ozone production overestimation (Hodneborg et al., 2011).The opposite holds for the modeled NO 2 concentrations which, due to instant dilution, 990 are negatively biased in the model within urban environment.The model performance for SO 2 and PM 2.5 is worse as well over urban stations compared to rural ones, probably for the same reasons as for ozone and NO 2 .
The urban emissions impact on air-quality 995 Generally, the impact of city emissions on ozone is characterized by two main features: over cities, all examined metrics decreased.Enhancements are encountered further from urban centers and are of lower magnitude than the decreases.Im et al. (2011a, b) performed regional chemistry simula-1000 tions over Istanbul and Athens and arrived to similar results: decrease of O 3 over urban areas due to reaction with NO and, as a consequence of NMVOC transport, a smaller production of O 3 at downwind areas due to increasing NMVOC/NO x ratio.Previously, Poupkou et al. (2008) showed that regional 1005 transport plays and important role in carrying urban pollution to larger distances leading to O 3 formation downwind from cities. Im and Kanakidou (2012) focused on both Istanbul and Athens and found up to 27 and 5 ppbv decreases of ozone due to urban emissions from these cities.Although they are not covered by our domain and are characterized by warmer climate, the changes are consistent with our JJA mean changes, especially in case of Athens, which is affected by higher background pollution (Kanakidou et al., 2011) as typical for cities over our domain as well.
Our results further suggest, that while the enhancement of average ozone further from cities (as a result of downwind transport) is relatively small (up to 0.5 ppbv), however, much larger increase is detected when considering metrics describing accumulated and extreme ozone values.In conclusion, the importance of cities' impact on ozone levels lies in higher potential for extreme ozone pollution over downwind areas during favorable meteorological conditions rather than in increase of average levels.This is seen especially in changes in the number of exceedances, but the AOT40 levels show often large enhancement around cities as well.
The impact on NO 2 levels is important only over cities themselves indicating that in urban plume further from urban areas, the NO x ages to HNO 3 decreasing the contribution to the total NO 2 levels.The contribution in cities goes up to 50 to 70 % indicating that large part (i.e.30-50 %) of the urban NO 2 pollution is of non-urban origin.This supports region or country wide emission control strategies as their emissions undergo regional transport.Im and Kanakidou (2012) found for Athens and Istanbul even larger contribution around 95-96 % over these cities.Earlier, Guttikunda et al. (2005) calculated the eastern Asia megacities contribution to NO z levels and found values around 10-30 % over cities, however NO z contains species produced during plume aging further from city centers causing this lower contribution.
In terms of average quantities (annual and DJF mean), the urban sulfur dioxide contribution is similar to NO 2 contribution.Urban SO 2 emissions are responsible for up to 70-80 % of pollution in over urban areas themselves.In other words, 20-30 % of urban SO 2 pollution comes from other other areas (rural and minor cities, villages) giving importance on the regional emission transport.Guttikunda et al. (2003) found over 50 to 75 % contribution near megacities of eastern Asia and 10-30 % over large areas in eastern China far from megacities.Similar values are obtained in our simulations (around 10-20 %) over large parts of the domain.They argue that large contribution within the cities indicate that the industry is concentrated within urban areas.This is often the case for eastern European cities where we obtained the highest urban SO 2 contributions.The urban contribution to SO 2 daily and (mainly) hourly exceedances is slightly higher than the contribution to average values and much more resembles the emission pattern indicating the enhanced importance of local urban emission in high air pollution episodes (when these exceedances occur) compared to the inter-urban pollution transport.
The annual average PM 2.5 impact calculated in our simulations (up to 10-15 µg m −3 ) are in line with values obtained for Istanbul and Athens from Im and Kanakidou (2012): 18 and 12 µg m −3 , respectively.However, their relative contribution is higher due to probably lower background pollution (around 62 and 55 %, respectively) compared to ours (30-60 % contribution over cities).
The emission reduction sensitivity test showed that in cities, the chemical regime is NO x -saturated meaning it is dominated by the reaction of NO with O 3 .This causes that NO x -oriented emission reduction actually worsen the ozone levels above cities and its close environment.This holds for the AOT40s and exceedance change as well, where decreases are modeled only far from cities, where the urban 1075 plumes undergoes photochemical aging.The response to 20 % NMVOC emission reduction is less dominant and in terms of all metrics it leads to reduction of ozone.Im et al. (2011b) tested the ozone response to 30 % reduced NO x or NMVOC emissions and found the ozone concentrations 1080 more sensitive to NO x emissions than to VOC emission.They also showed that above and around cities, reduced (increased) NO x emissions led to enhanced (suppressed) ozone by up to 8-10 % (in both direction), which is similar to the percentage change extracted from our simulations giving 1085 about 5-8 % ozone increase due to 20 % NO x emission reduction.The smaller numbers can be partly due to the lower emission reduction scenario.In case of NMVOC emission reduction, our numbers (up to −2 %) are only half of the relative ozone reduction achieved in their study (up to −4 %).

1090
The small ozone decreases due to NO x reduction in terms of AOT40s and exceedances further from cities in our simulations are probably caused by less NO x available in urban plumes when it mixes with biogenic NMVOC emissions over downwind areas (Im et al., 2011a, b;Finardi et al., 2014).

1095
Interestingly, the simultaneous NO x + NMVOC reduction scenario led to very similar ozone response than the NO x reduction alone.This can be explained by the dominating effect of ozone titration due to NO x emissions.Indeed, the NMVOC reduction caused only minor ozone changes.Con-1100 sequently, the effective ozone reduction strategy depend on the targeted area.Urban emission reduction of NO x and NMVOC improves the air-pollution over rural areas, however over the cities themselves, this usually leads to worsening of the ozone levels.According to our simulations, the 1105 only effective emission reduction strategy to decrease ozone levels is to reduce NMVOC emissions significantly while changing the NO x emissions only slightly or not at all.
Compared to the impact of all (100 %) emissions, the 20 % NO x + VOC emission change approximately equals to 1110 the 1/5th of the total impact (mainly for JJA ozone and AOT40s).This justifies the 100 % emission perturbation approach instead of a smaller perturbation introduced to maintain linearity.A similar approach was used recently for aviation emission impact (Huszar et al., 2013).Conclusions separatelly ... Answering the questions raised in the introduction, the air quality over cities is largely determined by the urban emissions but considerable (often a few tens of %) fraction of the 1120 surface concentration is attributable to other sources from rural areas, minor cities (which we did not consider here) or transported from distant areas (via the boundary conditions).On the other hand, the contribution of urban emission to surface pollution over rural areas is in general lower, around 5eling framework on the same domain as here, show that the urban impact on temperature is localized, meaning that there is only a minor influence of neighboring cities on a certain city (Prague, in their case).Here, for the impact of emissions on chemistry, it is shown for the case of Prague that the impact from "all-except-Prague" urban emissions on Prague is rather small (a few %, to over 10 % in case of PM 2.5 ) meaning that the air pollution over Prague is determined mainly by local urban sources rather than urban emissions from other cities.The inter-urban influence is largest for fine aerosol which has, in general, longer lifetime giving more importance to long range transport.
In summary, we showed that air-pollution over urban areas is a combination of the local urban emissions and those from rural areas without large cities with this later having often 1145 more than 50 % contribution.This implies that to meet the air-quality standards over cities, emission from the surrounding rural areas and long range transport have to be considered as well.Further it is shown, that the inter urban air-pollution is minor meaning that emissions from large cities do not in-1150 fluence each other in a significant way, at least as a long-term average.
It has to be emphasized that the long-term impact was evaluated here.Depending on the meteorological conditions (wind direction, temperature, boundary layer state), the impact of urban emissions represented by the urban-plume can be much larger than showed here, corresponding to the averaged impact.
Finally, numerous caveats of the modeling system were identified that has to be taken into account in future develop-1160 ments.The most important ones are: inclusion of time/space variant chemical boundary conditions that highly affected ozone performance, improvement in the representation of the annual cycle of emissions, considering biomass burning emissions, inclusion of dust emissions and their radiative ef-1165 fects as well as those of secondary organic aerosol.

Figure 2 .
Figure 2. First two columns: annual emissions (2005) per sector for the entire domain in Tg yr −1 , for all the cities in Tg yr −1 and for six selected cities from the domain in Gg yr −1 .Right two columns: the same as the first two one, but for the relative contribution of individual pollutants in %.

Figure 4 .
Figure 4. Left column: comparison of the 2001-2010 mean monthly variation of O3, NO2 and SO2 averaged over all stations with vertical error bars indicating the standard deviation of the average.Middle column: comparison of the 2001-2010 DJF mean diurnal variation for the same species with error bars indicating the standard deviation of the average.Right column: same as middle column but for summer months (JJA).

Figure 5 .
Figure 5.Comparison of monthly values of PM2.5 and its major components (sulfate-, nitrate aerosol, black and organic carbon) for DJF (orange) and JJA (dark blue) months.For carbonaceous aerosol, square stands for BC and triangle for OC.Linear trend lines are also shown.
520 the monthly means and is highest during early spring (−20 µg m −3 ) and reaching almost zero during August, September and October.On an hourly basis, the model is always negatively biased in DJF (by 10-15 µg m −3 ) showing minimum diurnal variations (in accordance with the mea-525 surements).During JJA, the model reasonably captures the timing of the ozone daily maximum values but underestimates the daily amplitude by giving smaller daytime peak values by almost 20 µg m −3 . 625

Figure 6 .
Figure 6.Impact of city emissions on ozone related air-quality measures listed in Table 4. Upper row presents the absolute change averaged over 2001-2010.The lower row corresponds to the change relative to the zero urban emission case (05ZERO).

Figure 7 .
Figure 7. Impact of city emissions on the annual NO2 concentration.Left figure presents the absolute change averaged over 2001-2010, while the relative contribution due to city emissions is shown on the right.

Figure 8 .
Figure 8. Impact of city emissions on SO2 related air-quality measures listed in Table 4: upper row shows the absolute change, the lower row the relative contribution from the urban emissions.

14P.
Figure 9. Impact of city emissions on the annual PM2.5 concentration.Left figure presents the absolute change averaged over 2001-2010, while the relative contribution due to city emissions is shown on the right.

Figure 11 .
Figure 11.Impact of 20 % reduction of NOx (upper row), 20 % reduction of NMVOC (middle row) and 20 % reduction of both NOx and NMVOC (bottom row) on ozone related air-quality measures: average JJA ozone, AOT40 for crops and forest, and, the average number of days with maximum 8 h ozone running mean greater than 120 µg m −3 .

Table 1 .
Summary of the conducted experiments including the experiment name, the time period, the emissions considered and whether radiative feedbacks on meteorology are considered.

Table 2 .
Comparison of model data with measurements: evaluation of the correlation coefficient (r), root mean square error (RMSE; in µg m −3 ), normalized mean square error (NMSE), the ratio of standard deviations (σr) and fractional bias (FB; in %) for hourly, daily and monthly averages for winter (DJF) and summer (JJA) months and for the whole year ("annual") for pollutants O3, NO2, SO2 and PM2.5.For PM2.5, no hourly data were available.

Table 4 .
The EC air-quality standards and AOT40 (in µg m −3 and ppbv h, respectively) for different averaging interval.(+): the average concentration is evaluated instead of the number of the exceedances.(-): no threshold value defined.
Huszar et al. (2012)007)uate treatment of emissions in considering them as only area sources.An important improvement in the RegCMCAMx4 model against its earlier version was the treatment of part of SO 2 emissions as elevated source which better compiles with the reality.How-920 ever, our results suggest minor improvement in DJF and the positive bias, although smaller, remained.On the other hand, summer encounters a clear negative bias.This could indicate that both the incorrect monthly disaggregation of the annual SO 2 emissions and the overestimated conversion to sul-925 fate aerosol (in JJA) play role here as well.Reduced deposition can contribute to SO 2 overestimation(Baker and Scheff, 2007)as well, as concluded byHuszar et al. (2012)who applied the same deposition scheme as in this study.To understand the model performance concerning the 930 PM 2.5 levels, we have to look at the comparison of the main fine particle matter components.In DJF, PM 2.5 is largely underestimated: the main contributors to this bias is the underestimation of nitrate aerosol and both black and organic carbon.The DJF sulfate aerosol is in acceptable agreement with the observations, which, given that SO 2 is overestimated, means that the SO 2 to SO 4 conversion is underestimated, leading to fair observation-model agreement.During JJA on the other hand, probably too strong SO 2 to SO 4 tran-