Climate forced air-quality modeling at urban scale: sensitivity to model resolution, emissions and meteorology

Introduction Conclusions References

processing method (6.5%) and the horizontal resolution of the air quality model (~5%) while 23 annual PM2.5 levels are particularly sensitive to changes in their primary emissions (~32%) and 24 the resolution of the emission inventory (~24%). The air quality model horizontal and vertical 25 resolution have little effect on model predictions for the specific study domain. In the case of 26 modelled ozone concentrations, the implementation of refined input data results in a consistent 27 decrease (from 2.5% up to 8.3%), mainly due to inhibition of the titration rate by nitrogen 28 oxides. Such consistency is not observed for PM2.5. In contrast this consistency is not observed 29 for PM2.5. In addition we use the results of these sensitivities to explain and quantify the 30 discrepancy between a coarse (~50km) and a fine (4km) resolution simulation over the urban 31

Description of the sensitivity simulations 28
Through a number of test cases we study the ability of the model to predict present-time decadal 29 air-quality with respect to emission and meteorological input as well as the CTM's horizontal 30 and vertical resolution. For that purpose we conduct five sets of 10yr long simulations (1996-31 2005) over a 4km resolution grid covering the IdF region (see Table 1). In all our comparisons 1 we use as a measure of sensitivity of modeled ozone and PM2.5 the absolute difference between 2 the mean of daily averaged concentrations (|∆c|) as well as the absolute change in the skill score 3 S. For ozone we also compare the MNB, MNGE and for PM2.5 the MFB and MFE. All scores are 4 calculated to represent an average of all urban, suburban or rural stations. For PM2.5 for which 5 only observations from urban stations are available we represent the results for summer, winter 6 and in annual basis of urban stations. 7 The first sensitivity case focuses on the climate bias due to the meteorological forcing. It is well 8 established that ozone and certain particulate matter species are sensitive to temperature changes 9 afternoon peak to numerous model processes and inputs for a typical summer episode in Paris 12 and found that temperature and wind speed were the most influential parameters to the observed 13 changes. For our test we utilize meteorological input that stems from a WRF run employing 14 ERA40 reanalysis data over a 0.44º resolution regional scale grid (ERA05) and compare with the 15 REF simulation utilizing climate model meteorology. Both configurations share identical 16 emission inventories (AIRPARIF) and vertical resolution (8 σ-p layers). Modeled meteorological 17 fields are further interpolated over the 4km-resolution IdF grid for the air-quality simulation. We 18 note here, that interpolating the 0.44 o resolution meteorology over the 4km resolution CHIMERE 19 grid adds a source of uncertainty in modeled pollutant concentrations, but due to the flat 20 topography of the area and as shown in previous research studies in the same region, increasing 21 the resolution of the meteorological input does not improve model performance (Menut et al., 22 2005;Valari and Menut, 2008). To study the impact of the resolution of the input meteorology 23 here, we conduct a second sensitivity run where meteorological input stems from a WRF 24 simulation using ERA40 reanalysis data over a finer resolution mesh with grid spacing of 0.11º 25 (ERA01) and compare with the ERA05 run. 26 The third sensitivity test addresses the issue of the CTM's vertical resolution (VERT). A 27 previous sensitivity analysis conducted with the same air-quality model showed only small 28 changes in modeled ozone and PM10 concentrations over the IdF region due to increase in the 29 CTM's vertical resolution (Menut et al., 2013b). On the other hand Menut et al. (2003) showed 30 that vertical diffusivity was one of the most influential parameters to the observed daily peak 31 concentrations of ozone for a typical summertime episode in IdF. Here, we undertake a similar 1 analysis but in a climate modeling framework, where enhanced meteorological bias is expected. 2 VERT implements a 12 vertical σ-p layers instead of 8. The major difference between the two 3 configurations (REF vs. VERT) is not the number of layers but the depth of the first model layer, 4 which is reduced from 20 to 8 m in VERT. We note that because the WRF meteorology 5 (resolved in 31 layers) is interpolated to the CTM's vertical grid, technically, increasing the 6 number of vertical layers in CHIMERE from 8 to 12 will result in a refinement of the 7 meteorological input used for the chemical simulations as well. 8 The fourth sensitivity case estimates the discrepancy in modeled ozone and PM2.5 concentrations 9 between two runs where emission totals stem from different inventories, namely the local 10 AIRPARIF inventory and the ECLIPSE regional-scale dataset. In Menut et al. (2003) it was 11 shown that the sensitivity of ozone concentrations in the afternoon peak hour due to surface 12 emissions was the second largest after the sensitivity associated with meteorology. In Markakis 13 et al. (2014) we compared the two approaches as for their ability to correctly represent ozone 14 photo-chemical production under typical anticyclonic summer conditions and also found 15 important differences. In the present work we push the analysis a step further and quantify model 16 response to the emission input over longer timescales. For this purpose we compile a new 4km 17 resolution emission dataset over the IdF domain (ANN) in which annual emission fluxes match 18 the ECLIPSE emissions (0.5 o resolution) but are downscaled spatially and temporally to obtain 19 4km-resolution and hourly emissions based on the local scale information implemented in the 20 bottom-up approach of the AIRPARIF emission inventory. The same approach is applied on the 21 chemical speciation of the inventory's pollutants to obtain emissions for all the species required 22 by the CTM's chemical mechanism. Therefore the only difference amongst the two runs stem 23 from the use of different annual quantified emission fluxes for the region (Table 1). To give a 24 sense of the discrepancies between the two inventories over IdF we compare the annual domain-25 wide fluxes of NOx, NMVOCs and PM2.5 (Fig. 2). NMVOCs emissions are considerably higher 26 in the ECLIPSE inventory while NOx emissions are lower than AIRPARIF. In terms of 27 photochemical ozone production, this makes ECLIPSE more favourable of NOx-limited 28 conditions than the bottom-up AIRPARIF inventory, which is consistent with the findings of 29 Markakis et al. (2014). Fine particle emissions are 2.4 times more in ECLIPSE, which probably 30 stems from the use of a population proxy to spatially allocate wintertime emissions from wood-31 burning. We note here, that the interest of comparing the two emission inventories is strictly to 1 quantify the added value of implementing local scale information in city-scale climate studies 2 and not by any means to compare qualitatively the two datasets. It should be made clear that 3 ECLIPSE dataset is not meant to accurately represent emissions at such fine scales. 4 In the fifth sensitivity case we study the impact of the post-processing methodology e.g., the 5 process followed in order to split the annual emission totals into hourly emission fluxes for all 6 the species and vertical layers required by the air-quality model. modeled concentrations to the refinement of the spatial and temporal allocation of input 12 emissions and found that the model was as sensitive to these improvements as to the vertical 13 mixing parameterization. Also they conclude that the temporal distribution of emissions in 14 particular, could be very important in stable urban atmospheres and that this sensitivity is 15 reduced with increased mixing conditions. For our test emission totals must match between the 16 two emission datasets. We compile a new emission dataset (POST) where the ECLIPSE annual 17 totals are spatially (both horizontally and vertically) and temporally downscaled on the 4km-18 resolution IdF grid. This procedure is based on coefficients extracted from the ECLIPSE post-19 processed inventory which in turn derive from the EMEP model. Comparing between the POST 20 and ANN runs (Table 1) we can model the impact on pollutant concentrations of integrating a 21 bottom-up approach in regional emission modeling. 22 Finally the impact of model horizontal resolution is a crucial issue for air-quality modeling. As show improvement but their results include the effect of increasing the resolution of the 1 meteorological input as well. Valari and Menut. (2008) showed that the impact of the resolution 2 of emissions on modeled concentrations of ozone may be higher than the model resolution itself. 3 This question has not yet been raised in the framework of climate driven atmospheric 4 composition modeling at the local scale. In our study we disentangle the impact of the resolution 5 of the emission dataset from the effect of model resolution itself by conducting two more tests. 6 In the first test we employ the 0.5º resolution simulation (REG hereafter) from which all 7 aforementioned simulations take their boundary conditions. We also compile the AVER database 8 which uses as a starting point the modeled concentrations at 4km resolution from the POST run 9 spatially averaged over the 0.5 o grid-cells of the REG resolution mesh. REG vs. AVER (see 10   Table 1) can provide information on the influence of model resolution while comparing AVER 11 against POST provides the sensitivity to the resolution of the emission inventory. where the bias is reduced by a factor of more than 2 compared to MET_CLIM. Summertime 4 temperature is adequately represented in the climate dataset whereas a wintertime weak cold bias 5 (-0.3 o C) is observed. A strong hot bias during the winter is found for the reanalysis meteorology. 6 A warmer climate can increase ozone formation through thermal decomposition of PAN 7 releasing NOx (Sillman and Samson, 1995). RH is generally well represented in both cases. 8 Finally we notice that the finer resolution reanalysis dataset (MET_ERA01) is not able to reduce 9 the observed domain-wide biases of the coarse meteorological run with the exception of specific 10 locations such as the Montsouris station in Paris where the bias in wintertime precipitation and 11 wind speed bias is reduced by 22% and 40% respectively. 12 13

Evaluation of the reference simulation (REF) 14
Mean modeled daily surface ozone and the daily maximum of 8-hour running means (MD8hr) 15 are compared against surface measurements in urban, suburban and rural stations (Fig. 3a). The 16 results presented are averaged over the ozone period (April-August). We also use odd oxygen MNGE are +20.6 and 38.9 respectively. 30 We observe an overestimation of mean daytime suburban ozone (+5ppb). The small bias in Ox 1 (+0.6ppb) suggests that the problem stems from the representation of local titration and more 2 specifically daytime titration; the daily average ozone bias drops to +3.9ppb while Ox is 3 accurately represented in this case (-0.2ppb). Suburban stations present the lowest skill score 4 (0.63) compared to urban and rural. Model performance over rural stations is adequate, with an 5 overestimation in mean daily ozone of 8.2% (bias=+2.8ppb) and a good skill score (0.73). The 6 two major downwind locations in the IdF domain which present the lowest biases (less than 7 0.1ppb and 1.1ppb for the south-west and north-east directions respectively). The bias of the 8 daytime average reaches +2.1ppb. 9 Ozone daily maxima in the urban and rural stations are underestimated by 10% (-4.2ppb) and 7% 10 (-3.2ppb) respectively but we consider the magnitude of the underestimation small given the 11 climate framework of the simulation. Daily average ozone is better represented than daily 12 maxima, highlighting model sensitivity to accumulated errors (Valari and Menut, 2008). 13 Modeled peak concentrations are particularly sensitive to temperature compared to the daily 14 averages as shown in Menut at al. (2003). This could also be due to the fact that 4km is still an 15 insufficient model resolution. 16 The evaluation of PM2.5 (Fig. 3b)

Sensitivity to climate model driven meteorology (REF vs. ERA05) 30
This case study estimates the discrepancy between an air-quality model run where regional 1 meteorology is downscaled with WRF from reanalysis data (ERA05) and a simulation where 2 meteorology is downscaled from a global scale climate model (REF). The wet bias in 3 MET_CLIM meteorology is significantly reduced with meteorology from reanalysis data (Sect. 4 3.1). This is expected to have a significant role in the modeled PM concentrations. Another 5 influential factor is the colder bias found in summertime temperature in the MET_ERA05 6 dataset. This could lead to decreased reaction rates, less biogenic emissions and consequently to 7 less ozone. The lower bias in 10m wind speed under MET_ERA05 is bound to increase surface 8 concentrations through reduced dispersion. We also compare the average modeled boundary 9 layer height (PBL) for the summer and winter periods between the two datasets: PBL is reduced 10 by 5% and 12% in summer and winter respectively (not shown) when reanalysis data are used 11 instead of climate model output. This may result in less dilution of emissions and therefore 12 higher surface concentrations for primary emitted species, such as PM and NOx. 13 Comparing the results of the two air-quality model runs for ozone ( Fig. 4a and Table 3) we find 14 only a small sensitivity to using meteorology from a climate model or reanalysis data over all 15 three types of monitor sites (|∆c|~1ppb or 3.4%). The small improvement of model performance 16 with the reanalysis dataset (ozone decreases through higher NOx emissions following the PBL 17 scheme described above) is due to the fact that titration is more realistically represented in 18 ERA05 (the difference is Ox between the two runs is negligible). The response of urban daily 19 maximum values to the meteorological dataset is also negligible (|∆c|=0.1ppb or 0.3%). We conclude that using climate model driven meteorology has a small impact on modeled ozone 3 whereas larger sensitivity is observed for wintertime PM2.5 levels due to the accuracy of modeled 4 precipitation. 5 6 4.2 Sensitivity to the resolution of the meteorological input (ERA01 vs. ERA05) 7 Here we model the sensitivity of modeled ozone and PM2.5 concentrations to the resolution of the 8 meteorological input (Fig. 5 and Table 3). Daily average ozone shows a very weak response over 9 urban and rural sites (|∆c|<0.4ppb or <0.8%) and daily urban maxima improve slightly with the 10 ERA01 run (|∆c|=0.4ppb or 1%). At the suburban area the impact, though small (|∆c|=1.4ppb or 11 4.3%), is definitely higher than over urban or rural sites. Ox change at the suburban area (not 12 shown) is much weaker compared to ozone (|∆c|~0.5ppb or 1.2%) showing that the increase in 13 the resolution of meteorology has an impact on the representation of ozone titration leading to 14 improved model performance. The skill score over suburban sites increases by 9% while NMB 15 improves by 22% from 26.1 in ERA05 to 20.3 in ERA01. Interestingly, the response of suburban 16 ozone to the resolution of the meteorological input is the strongest modeled sensitivity for this 17 variable amongst all studied cases. 18 Weak sensitivities are modeled for PM2.5 (Table 3)  This study addresses the impact of the resolution of the CTM's vertical mesh and more 8 specifically of the thickness of the first CTM layer, on modeled ozone and PM2.5 concentrations 9 (Fig. 6). Mean daily ozone is practically insensitive to the refinement of the vertical mesh at the 10 urban, suburban and rural areas (Table 3). Similarly, maximum ozone at the urban area changes 11 by only 0.5ppb (1.4%) with increased bias in the VERT run. Changes in summertime and annual 12 modeled PM2.5 concentrations are also small, while the wintertime daily average shows some 13 weak sensitivity (|∆c|=0.5μg/m 3 or 2.2%). Scores are hardly affected. 14 Interestingly, the impact of the refinement of the vertical grid on daily averaged Ox is much 15 stronger that on ozone: Ox, changes by 0.9ppb in the urban and suburban areas. The change in Ox 16 is reasonable since in VERT, NOx emissions are released within a surface layer thinner by 60% 17 compared to REF (from 20m to 8m) leading to higher NOx concentrations. That should normally 18 affect titration which is the driver of urban ozone concentrations. The fact that ozone remains 19 insensitive to the change in NOx concentrations suggests that some other modeled processes 20 counteracts titration. To further investigate this issue we study the change in dynamical processes 21 such as vertical mixing and dry deposition. We extract the vertical diffusion coefficient Kz (m 2 /s) 22 and dry deposition rates (g/m 3 ) for ozone, NO2 and PM2.5 for all grid cells that include an urban 23 monitor site and look how modeled sensitivities change as a function of these parameters (Fig.7). 24 NO2 concentrations increase with the refinement of the first vertical layer of the CTM for all 25 vertical mixing conditions (Fig. 7a). However it is only under low vertical mixing (1< Kz<5 26 m 2 /s) that ozone sensitivity becomes positive (Fig. 7b). Under stronger turbulence (Kz > 5 m 2 /s), 27 the 12-layer setup leads to higher first-layer NO2 concentrations (stronger titration) leading to 28 negative values for ozone sensitivity (such conditions account for the 93% of the simulated 29 period). On the other hand the refinement of the vertical mesh primarily affects NO2 deposition 30 rates which accelerate by 14.3% but leaving ozone deposition rates unaffected. We may assume 31 that under low mixing conditions, the increased deposition rate of NO2 slows down the increase 1 in NO2 concentration due to the emission effect and dynamical processes become more 2 influential than titration. As a result the surface layer is enriched in ozone by getting mixed with 3 air from higher atmospheric layers (Menut et al., 2013b). 4 For almost the entire Kz range, PM2.5 concentrations increase with VERT (Fig. 7c). This is due to 5 the fact that emissions are released in smaller volumes as discussed above. On the other hand, 6 here too, the refinement of the vertical resolution of the CTM, enhances deposition rate. These 7 two conflicting effects explain the small impact of the CTM's vertical resolution on PM2.5 8 concentrations. 9 We conclude that both ozone and PM2.5 sensitivities to the refinement of the vertical mesh are 10 small. Our analysis suggests that in both cases this is the result of two competing processes, 11 either titration against vertical mixing (ozone) or emission versus deposition (PM2.5). Although 12 in the Ile-de-France area (low topography) the overall effect is insignificant, it may not be the 13 case in other regions with more complex topography. 14 15

Sensitivity to the annual emission totals (REF vs. ANN) 16
This case study compares modeled concentrations between two runs where annual emission 17 totals stem from either the AIRPARIF inventory (REF) or the ECPLISE dataset (ANN). Changes 18 in modeled urban daily average ozone concentrations are small (|∆c|=0.8ppb or 2.5%) with the 19 regional inventory (ECLIPSE) to tend to increase the bias of the REF run (Fig.8a and Table 3). 20 This is due to the fact that when passing from the AIRPARIF to the ECLIPSE inventory (see 21 also in suburban Ox (|∆c|=0.1ppb or 0.3%) suggests that this area benefits more than the urban area 28 from the improvement in the titration process. The skill score associated to the REF run is also 29 higher by 8% (Fig. 8a). Changes in daytime averages at both urban and suburban areas are 30 similar to those in the daily averages suggesting that modeled sensitivity stems mainly from 31 daytime titration. Rural ozone is practically unaffected (|∆c|= 0.3ppb or 1%). It is noteworthy 1 that the absolute change in modeled ozone concentrations is in the order of 1ppb or less despite 2 the large differences in ozone precursors' emissions between the local and the regional 3 inventory. 4 Changes in the daily average fine particle concentrations in summertime, wintertime and in the 5 annual basis are much stronger than ozone (|∆c|=4.1μg/m 3 or 33%, 6.6μg/m 3 or 33.8% and 6 5.5μg/m 3 or 31.9% respectively). PM2.5 concentrations modeled with the ANN run are 7 significantly higher than those modeled with the REF run (Fig. 8b). Wintertime bias in ANN 8 reaches +5.8μg/m 3 showing that fine particle emissions from the ECLIPSE inventory are 9 overestimated (see also Fig. 2). The main source of primary wintertime PM2.5 emissions over the 10 IdF region as well as in Paris in the ANN run is wood burning (see discussion in Sect. 2.4), 11 which is unrealistic for a city like Paris and stems directly from the use of the population proxy 12 to spatially allocate national totals over the finer scale. This is consistent to the fact that the 13 summertime bias in the ANN run is much lower (+1.4μg/m 3 ). In fact, in this case the ANN bias 14 is even smaller than the REF bias (-2.8μg/m 3 ) enhancing our hypothesis that summertime fine 15 particle emissions in the AIRPARIF inventory are underestimated (see also Sect. 2.1). The skill 16 score in REF is higher than in ANN in wintertime and lower in summertime. 17 We conclude that ozone sensitivity to the annual emission totals is low but strong for fine 18 particles. 19 that don't link back to the same quantified emissions either. For example in the regional 30 application used this study (REG) the sectoral ECLIPSE raw emissions quantified in SNAP level 31 are treated with the respective sectoral coefficients that stems from the EMEP inventory having a 1 very different synthesis of sub-SNAP sources from that of ECLIPSE. Therefore when we 2 compare ANN with POST we consider that what we observe is the bias of this inconsistency in 3 regional modeling. The question raised is: what is the benefit of adopting a bottom-up post-4 processing for regional scale air-quality modeling?. 5 The effect on ozone concentrations over the urban area is considered moderate (|∆c|=1.9ppb or 6 6.4%) (Fig. 9a and Table 3). Model bias is reduced from +4.5ppb in POST to +2.6ppb in ANN. 7 Ozone sensitivity in this case, is twice as high as the sensitivity to climate model driven 8 meteorology and even higher compared to the impact of annual totals. The ANN simulation is 9 able to increase the skill score by 14% and reduce MNB by 26%. The low Ox sensitivity suggests 10 that discrepancies are mainly due to a better representation of ozone titration. shows the highest sensitivity in the emission post-treatment among all the presented cases 31 (|∆c|=2.2ppb). This is consistent with Menut et al. (2003) who also found that the afternoon peak 1 concentrations at a typical summertime episode in Paris are very sensitive to the NO emissions 2 change. In the evening (after 15:00LT) ANN deviates from the observations faster than POST 3 because the afternoon peak in traffic emissions is more pronounced in the AIRPARIF diurnal 4 profile compared to that used in the ECLIPSE processing which represents an average situation 5 of anthropogenic sources hence a smoother variation. These results indicate that the diurnal 6 variability of modeled ozone over the urban area is very sensitive to the choice of the diurnal 7 profile. But in the climate concept where hourly values are timely too short to take into account, 8 the sensitivity is considered moderate as seen in Table 3. 9 Modeled PM2.5 sensitivity is significant for both summer and wintertime (|∆c|=3.4μg/m 3 or 10 24.8% and 4.6μg/m 3 or 18.3% respectively) ( Table 3). POST wintertime bias is almost two times 11 higher than ANN (Fig. 9b). This is because the coarse resolution annual post-processing 12 coefficients weight towards allocating more of the annual emissions into the winter period 13 significantly influenced by the residential sector emissions which are overstated in the ECLIPSE 14 inventory. A late afternoon peak is modeled with ANN accounting for the traffic emissions, 15 whereas PM2.5 evening levels modeled with the POST run (after 20:00LT) are related to the 16 residential heating activity (Fig. 10b). 17 What we can conclude is that in a climate forcedair quality framework the model response for 18 daily average ozone by 6.2% is rather small considering the significant differences that the two 19 post-processing approaches prescribe for the vertical distribution of emissions and their diurnal 20 variation. Fine particle concentrations are much more sensitive to the applied emission post-21 processing technique. We note here, that recent work has pointed out that the sensitivity of

Sensitivity to the emission inventory resolution (POST vs. AVER) 26
Here, we quantify the effect of the resolution of the emission input. Results show that in the 27 urban areas this sensitivity is the most influential amongst all tests presented in this paper with 28 ozone changes reaching 2.8ppb or 8.3% (Fig. 11a). The change in daily average Ox is smaller but 29 comparable (|∆c|=1.2ppb or 2.9%) suggesting that ozone titration is not the only model process 30 that is affected by the increase in the resolution of the emission dataset. The skill score and MNB 31 improve significantly in the POST run (Table 3). Ozone precursors' emissions from urban 1 sources are mixed with the lower emissions from the surrounding suburban and rural areas inside 2 the large cells of the coarse mesh-grid (AVER). This leads to lower titration rates and therefore, 3 higher ozone levels. Therefore the increase in the resolution of the emission input leads to a 4 reduced positive bias from +7.3ppb (AVER) to +4.5ppb (POST). AVER overestimates ozone 5 peaks by 0.8ppb while POST underestimates them by -1.2ppb. The sensitivity of ozone 6 concentration at the hour of the afternoon peak is linked to NOx concentration at the same hour, 7 which reaches a local maximum due to the evening rush hour (see also Sect. 4.5). Suburban and 8 rural ozone is less sensitive than urban (|∆c|=0.7ppb), with scores practically unchanged (Table  9 3). 10 Fine particle concentrations are also very sensitive to the resolution of the emission input, 11 especially in wintertime (|∆c|=7.1μg/m 3 or 30%), with higher concentrations modeled with the 12 refined emission inventory in POST (Table 3). Similarly to ozone this is because in the coarser 13 inventory represented here by AVER, emissions in the high emitting areas in the city are 14 smoothed down and diluted when averaged with emissions of the less polluted outer areas. 15 We conclude that the resolution of the emission input is the most influential factor from all the 16 studied cases, even more than model resolution itself. PM2.5 showed higher sensitivity than 17 ozone concentrations. The non-linear nature of ozone chemistry suggests that it is important for 18 the ozone precursor emissions to be concentrated correctly to the high emitting areas such as the 19 urban centres. 20 21

Sensitivity to model horizontal resolution (AVER vs. REG) 22
Here, we study the sensitivity of ozone and PM2.5 concentrations to the CTM's horizontal 23 resolution. We compare the simulations of two different spatial resolutions, the AVER run 24 (averaged over the grid-cells of the coarser grid) and the REG simulation on a grid of 0.5º 25 resolution (Fig. 12). REG, models higher ozone concentrations than AVER over the urban area 26 (|∆c|=1.7ppb or 4.7%). As discussed above, NOx emissions in the REG simulation are lower than 27 in REF due to dilution in the coarser grid cells leading to lower ozone titration rates. Suburban 28 and rural ozone has low sensitivity to model resolution (|∆c|=0.5ppb or 1.4% and 0.2ppb or 0.5% 29 respectively) because photochemical build-up occurs at larger time and space scales compared to 30 titration and the refinement of the model grid does not increase performance. This confirms the 31 results in Markakis et al. (2014). The effect on modeled PM2.5 is very small with concentrations 1 slightly higher over the finer mesh grid as a result of the lower primary emissions in REG. 2 We may conclude that the benefit of increasing the CTM's resolution is insignificant for both 3 ozone and PM2.5 especially taking into account the large refinement attempted here (0.5 o to 4 4km). 5 6 5 Sources of error in regional climate forced atmospheric composition 7 modeling 8 In this paper we utilize simulations at two spatial scales: at urban scale over a grid of 4km 9 resolution using the AIRPARIF bottom-up inventory of anthropogenic emissions (REF) and a 10 regional scale run at 0.5º resolution where emissions stem from the ECLIPSE top-down 11 inventory (REG). Both realizations implement identical climate driven meteorology (at 0.44 o 12 resolution) and an 8-layer vertical mesh therefore are susceptible to the same sources of error due 13 to climate model driven meteorology, the resolution of the meteorological input and the 14 resolution of the CTM's vertical grid. However the remaining biases presented in Table 3 over  15 urban areas e.g., the emissions resolution, the model horizontal resolution, the annual quantified 16 fluxes and the post-processing method concern mainly the REG run. As regards ozone REG has 17 a positive bias of 9ppb over the city of Paris while the bias of REF is only +1.8ppb (Fig. 13a). 18 The question we raise is "what are the main sources of uncertainty in regional scale climate 19 driven air-quality simulations and how these could be eliminated or at least reduced?". 20 With this study we are able to identify the source of the excess of |∆c|=7.2ppb of ozone modeled 21 with the REG run compared to REF (Table 4) Considering the discrepancies in the inventorying methodologies used to compile the ECLIPSE 27 and the AIRPARIF datasets (top-down vs. bottom-up), it is very interesting that the least 28 influential factor to the urban ozone response is the annual emissions totals. It seems that the 29 regional simulation would not benefit much from the integration of the local annual totals alone 30 but a more important gain would stem from the application of the AIRPARIF post-processing 31 methodology. The added value from both these factors would reduce the positive bias of REG by 1 2.7ppb. Even largest improvement comes through the better spatial representation of ozone 2 precursors emissions in the local emission inventory (|∆c|=2.8ppb) leading to more faithful 3 titration process; Ox levels are very close in REF and REG (Fig. 13a). It could therefore argued 4 that without increasing model resolution of which the gain would reach only 1.7ppb, the REG 5 simulation would benefit significantly by simply integrating the aforementioned local scale 6 information. 7 The difference in modeled ozone between REF and REG is much smaller over the suburban area 8 (|∆c|=2.4ppb) and the most influential factor to this difference is the annual emission totals 9 covering 45.8% of this difference. Finally as regards ozone one important result of this study is 10 that in the climate-air quality framework modeled concentrations from a coarse resolution run, 11 well agree with the much more intensive (in terms of computational time) fine resolution run and 12 the bias is considered of small magnitude (Fig. 13a). This is because the formation of rural ozone 13 is a slower process than in urban areas and comparable to the characteristic transport time of 14 precursor's pollutants to the coarse grid cell. In the present paper we assess the sensitivity of ozone and fine particle concentrations with 3 respect to emission and meteorological input with a 10yr long climate forced atmospheric 4 composition simulation at fine resolution over the city of Paris. 5 As a general observation our study shows that overall ozone response is considered low to 6 moderate while PM2.5 concentrations were generally very sensitive for the presented cases. The 7 largest sensitivity in modeling the average daily ozone concentrations was observed in the urban 8 areas primarily due to the resolution of the emission inventory (|∆c|=2.8ppb or 8.3%) and 9 secondly to the post-processing methodology applied on the annual emission totals (|∆c|=1.9ppb 10 or 6.2%). These sensitivities are attributed to changes in the titration process. When post-11 processing coefficients were derived from the bottom-up AIRPARIF inventory instead of EMEP, 12 too much ozone titration takes place at the hour of the ozone peak and the sensitivity of daily 13 maximum reached its highest value among all the studied cases (|∆c|=2.2ppb or 5.8%). It is 14 interesting that despite the fact that ozone precursor's emissions are very different between the 15 bottom-up and the top-down inventories, ozone sensitivity to the annual totals was shown to be 16 very small (|∆c|=0.8ppb or 2.5%). Also modeled ozone is fairly insensitive to the use of climate 17 model or reanalysis meteorology. Finally all cases of suburban and rural ozone both for average 18 and maximum concentrations showed a sensitivity of less than 5%. 19 Regarding PM2.5 concentrations, amongst all the presented factors, the emissions related were 20 those shown to be the most influential. The corresponding sensitivity to the use of annual 21 emission totals from a top-down and a bottom-up inventory reached 33% in summer, 33.8% in 22 winter and 31.9% for the daily average concentrations. This is connected to the downscaling 23 methodology applied in the regional-scale totals of the ECLIPSE inventory; using population as 24 proxy for their spatial allocation, leads to overestimation of particle emissions from wood-25 burning over the Paris area. Large sensitivity was also shown due to the resolution of the 26 emission inventory (20.3% in the summer, 30% in the winter and 24.2% in annual basis) because 27 the coarser inventory smoothens the sharp emission gradients over the urban area leading to less 28 primary emissions. Fine particle concentrations were also sensitive to the applied emission post-29 processing technique (22.1% in summer and 16.7% in winter). Only wintertime PM2.5 30 concentrations were significantly affected by the meteorological related sensitivities; by 17.6% 31 due to the use of meteorology from reanalysis instead of climate (mainly because the prescribed 1 changes in modeled precipitation) and by 6.8% due to refinement of the meteorological grid. 2 Both ozone and PM2.5 are little sensitive to the CTM's vertical resolution (changes of less than 3 2.2%). Nevertheless we provide evidence that this low sensitivity may be the result of 4 counteracting factors such as ozone titration, dry deposition and vertical mixing, too much 5 dependent on local topography to be able to generalize for other regions. We also note the weak 6 sensitivity of modeled concentrations to the increase in the CTM's and the meteorological 7 model's horizontal resolution at least for the area and the range of resolutions studied here. 8 Excluding the sensitivities having the smallest impact (roughly less than 2%, see Table 3)  To fill the gap between regional and city-scale assessments we have to combine in a single 21 application the advantages of regional and local scale applications; the low resolution (but high 22 spatial coverage) from one hand and the good representation of emissions (but limited area of 23 coverage) on the other. The results of this study move towards that goal and can be used in order 24 to identify the main sources of error in regional scale climate forced air-quality modeling over 25 the urban areas. These biases could be taken into account in policy relevant assessments. 26 The difference in modeled daily average ozone between the local and regional application over 27 the urban areas (|∆c|=7.2ppb) is attributed to several sources of error: 38.9% is related to the 28 resolution of the emission inventory, 26.4% stems from the post-processing of national annual 29 emission totals, 23.6% is due to model resolution (4km or 0.5 o ) and 11.1% is associated to the 30 annual emissions used as starting point for the compilation of the anthropogenic emission 31 dataset. Although the greatest benefit in the regional-scale modeling seems to come through the 1 increase in the resolution of the emission inventory, simpler actions may be also meaningful, 2 such as the integration of the locally developed annual totals and the downscaling coefficients 3 derived from the existing bottom-up modeling systems which combined could reduce the bias of 4 the regional application by 37.5%. We note here that PM2.5 levels in the urban regions are likely 5 mostly controlled by primary emissions; increasing the emissions inventory resolution will 6 concentrate the PM2.5 emissions into a smaller spatial extent of the urban area (the reverse side of 7 the artificial dilution issue taking place at coarse resolution); if the emissions totals are 8 themselves biased high, then the resulting error will only become apparent at higher resolution. 9 Therefore, the emissions resolution may be showing that the emissions totals are too high, and 10 this only becomes apparent at high resolutions. 11 As regards PM2.5 modeling our study shows that the regional realization cannot selectively 12 incorporate any combination of local-scale features in order to improve performance as in the 13 case of ozone. The simulation at regional scale (REG) predicts an excess of 3.6μg/m 3 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35   36  37  38  39  40 6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25