We perform simulations with the coupled meteorology–chemistry model WRF-Chem, version 3.7.1 . The model domain consists of 170 by 170 cells at 20×20 km2 horizontal resolution covering Europe, centered at 51.98∘ N and 5.66∘ E. Vertically, the domain extends from the Earth's surface up to 50 hPa and consists of 27 layers with 13 layers in the lowermost 1500 m. Chemistry simulations of O3 and its precursor groups (NOx and VOCs) are performed with the Carbon Bond Mechanism Z (CBM-Z) gas-phase chemical mechanism . Simulations of atmospheric chemistry with this mechanism compare well with the European multi-model mean for summer O3 in a gas-phase mechanism comparison study . A complete list of parameterization options adopted in our WRF-Chem setup can be found in Table S1 of the Supplement. Our simulations were performed with a time stepping of 180 s for a period of 38 d (24 June–31 July 2015), allowing a 1-week spin-up to analyze the model output for July. An evaluation of large-scale meteorological performance with ERA-Interim reanalysis fields can be found in Sect. S2 of the Supplement.

We used anthropogenic emissions from the Netherlands Organisation for Applied Scientific Research – Monitoring Atmospheric Composition and Climate (TNO-MACC-III) inventory for 2011, the most recent inventory available when the model experiments were performed. TNO-MACC-III contains anthropogenic emissions for lumped species groups (NOx and VOCs). NOx emissions were partitioned assuming that 97 % is emitted as NO and 3 % as NO2. VOC emissions were divided over 15 emission categories in CBM-Z, following the VOC speciation by . This speciation procedure is further described in Table S3 of the Supplement. Point-source emissions were distributed over the five lowermost model layers following sector-specific emission altitude profiles .

Biogenic emissions of VOCs and soil NOx were calculated online with the Model of Emissions of Gases and Aerosols from Nature (MEGAN) implementation within WRF-Chem . The domain-total biogenic isoprene emissions are 1.82 Tg of isoprene, which is slightly lower than the 9-year spread of 2–4.5 Tg isoprene for July, based on an inverse modeling study using OMI HCHO column measurements for 2005–2013 . We simulate lightning NOx emissions using a parameterization based on cloud-top height , using a flash rate of 80 mol flash-1 based on a recent satellite-based estimate . Simulations with higher flash rates of 500 mol flash-1 and 310 mol flash-1 resulted in overestimated upper-tropospheric contributions to the NO2 columns relative to OMI.

Anthropogenic emissions are the dominant NOx source over Europe in July, with a total monthly emission strength of 304 Gg N (76 %). Minor contributions are associated with lightning (81.4 Gg N; 20 %) and soils (15.0 Gg N; 4 %). We note that especially soil NOx emissions are low compared to previous studies, in which soils, including agricultural areas, have been estimated to contribute 40 % to the total European NOx emission budget .

Meteorological initial and boundary conditions were taken from ERA-Interim reanalysis data . Chemical boundary conditions for O3, NO, NO2, CO and peroxyacetyl nitrate (PAN) are taken from the Copernicus Atmosphere Monitoring Service (CAMS) chemical reanalysis product for Europe retrieved at http://apps.ecmwf.int/datasets/data/cams-nrealtime/levtype=sfc/ (last access: 18 September 2019). Upper boundary conditions for ozone were prescribed with climatological values (retrieved at https://www2.acom.ucar.edu/wrf-chem/wrf-chem-tools-community, last access: 17 September 2019).

Surface measurements are taken from the European Air Quality Data Portal operated by the European Environment Agency, hereafter referred to as AirBase . We used all data at rural background stations from the validated E1a data stream. The large availability of the data allows us to make a strict selection on data availability. For monthly averages, we discard stations if data are missing for more than 24 h. Stations used for the evaluation of monthly averages at 12:00 UTC may have a maximum data gap of one data point. This resulted in a final selection of 184–397 stations, depending on the performance metric (see Table ). In our analysis of O3 and NO2, we evaluate monthly time series and midday (12:00 UTC) concentrations (denoted as [O3]12h and [NO2]12h, respectively). We additionally calculate the maximum daily 8 h mean ozone concentration (MDA8 O3), a widely applied metric for O3 health impacts.

We use tropospheric NO2 columns from OMI aboard NASA's Earth Observing System (EOS) Aura mission . The polar-orbiting instrument detects radiation backscattered from the Earth's atmosphere. Retrieval of tropospheric vertical column densities (VCDs) from space follows a three-step procedure . First, total slant columns (SCDs; i.e., columns along the average light path through the atmosphere) are obtained from a spectral fit to the OMI-measured reflectance spectra in the visible wavelength range using the differential optical absorption spectroscopy (DOAS) method. Then, the stratospheric contribution component is separated from the total NO2 column via data assimilation into the TM5 global chemistry transport model . The final step is to obtain tropospheric VCDs by dividing the SCDs by a tropospheric air mass factor (AMF) that describes the vertical sensitivity of the instrument to atmospheric NO2 . This is a function of satellite viewing geometry, surface albedo, terrain height, cloud properties and a priori NO2 profile.

The recent EU FP7 QA4ECV project has led to the development of a new OMI NO2 data product . The underlying consortium retrieval algorithm is based on the NO2 column retrieval principles described in but with improvements in the three aforementioned steps . described how better wavelength calibration and inclusion of liquid water absorption and an intensity offset correction reduced uncertainties in NO2 SCDs to 0.7-0.8×1015 molec. cm-2 (up to ±35 %). improved the AMF calculation method via the extension of the AMF look-up table with more reference points and a correction for the sphericity of the atmosphere. The ancillary data for the AMF calculation have also improved relative to earlier algorithms such as DOMINO v2 : surface albedo from the 5-year OMI albedo climatology , cloud information from the improved OMI O2–O2 algorithm and a priori NO2 profiles from TM5-MP at 1∘ × 1∘ . The study by also showed that substantial differences between AMFs arise when different a priori NO2 profiles (as well as surface albedo and cloud properties) are used in the retrieval. This underlines that a recalculation of the tropospheric AMFs based on simulated WRF-Chem NO2 profiles at 20×20 km2, replacing the coarse TM5-MP 1∘ × 1∘ NO2 profiles, may help to reduce model–satellite differences , and we will explore this further below.

We take care to remove inconsistencies in the model–satellite comparison introduced by different assumptions about the vertical NO2 profile in the satellite product compared to the model. The AMF calculation requires assumptions about the vertical profile of NO2 to convert slant columns into vertical columns. We replace the a priori TM5-MP NO2 profiles (at 1∘ × 1∘) by WRF-Chem NO2 profiles at a 20×20 km2 resolution. This has two advantages: (1) model–satellite comparisons are no longer affected by differences in model assumptions between WRF-Chem and TM5-MP that lead to different vertical NO2 profiles, and (2) the higher-resolution WRF-Chem setup resolves spatial gradients in the a priori profile that are not appropriately captured in TM5-MP due to the coarser model resolution. Single-orbit results indicate that recalculation of the AMFs leads to retrieved columns that are 1×1015 molec. cm-2 higher in densely populated areas and lower or unaffected in surrounding non-urban regions. This effect has been seen before in earlier studies .

We apply the method described by and to replace the TM5-MP vertical NO2 profile by the WRF-Chem profile in the calculation of the AMF: 1Mtrop,WC=Mtrop,TM5×∑l=1LAtrop,lxl,WC∑l=1Lxl,WC, where Mtrop is the tropospheric AMF based on an assumed profile from WRF-Chem (WC) or TM5, Atrop,l is the tropospheric averaging kernel element for layer l, xl,WRF-Chem is the NO2 column density in model layer l, and L is the uppermost TM5-MP layer in the troposphere. The tropospheric averaging kernel in Eq. () is defined as follows : Atrop=A×MMtrop, where M and Mtrop refer to the AMF and the tropospheric AMF, respectively. Note that the WRF-Chem vertical NO2 profile has been sampled at the TM5-MP vertical layer structure, so l refers to TM5-MP model layers.

We start our evaluation of O3 chemistry in WRF-Chem (with bottom-up NOx emissions, i.e., not yet based on the OMI-inferred NOx emissions) by a comparison of monthly averaged 24 h mean surface ozone simulations with AirBase observations (Fig. a and b; Table ). WRF-Chem reproduces the spatial distribution of surface ozone satisfactorily, with an increase in surface O3 concentrations from north to south, as reported elsewhere e.g.,. Highest concentrations are found around the Mediterranean basin. O3 concentrations over central and southern Europe are underestimated in WRF-Chem. Simulated monthly averaged concentrations do not exceed 110 µg m-3, while higher concentrations were observed at several stations in the southern part of the domain. Most notably, WRF-Chem does not capture observed high concentrations of ±130 µg m-3 in northern Italy. The good agreement between WRF-Chem and in situ data in the western part of the domain close to the model boundaries with a prevailing westerly circulation indicates that the model boundary conditions describe inflow of long-lived compounds such as O3 from the western boundary well.

Monthly averaged ozone concentrations are an important and widely used metric to evaluate model skill but are not necessarily indicative of the peak ozone concentrations that typically occur in the afternoon. These monthly averages include the nocturnal conditions with generally the presence of stable boundary layers, in which the titration of ozone in the NOx-saturated regions is difficult to model e.g.,. The simulated and observed monthly averaged ozone concentrations at 12:00 UTC (Fig. c, d) demonstrate a similar geographical distribution compared to the monthly average but with higher values because photochemical ozone production generally peaks during daytime. This figure demonstrates that peak ozone values occur around the Mediterranean basin, most prominently in northern Italy and Spain, where the levels of sunlight and ozone precursor concentrations are high. WRF-Chem shows elevated ozone with respect to adjacent areas, but maximum simulated ozone levels do not exceed 120 µg m-3. This underestimation of peak ozone concentrations is also apparent from in Fig. b (discussed in more detail in Sect. ), which shows the simulated versus the observed 12:00 UTC ozone concentrations.

Our results are in agreement with previous regional chemistry model evaluations for Europe. Such studies typically focus on seasonal variability; we compare our results with the results for European summer (JJA) from those studies. found that a model ensemble underestimates the daytime maximum O3 concentration for sites where observed O3 concentrations exceed 120–140 µg m-3, which agrees with our results. In that study, the ensemble mean model bias tends to become more negative for observed concentrations above 80 µg m-3 . The two ensemble members that use CBM-Z chemistry, similar to our WRF-Chem model setup, are qualitatively in line with the ensemble mean, lending support to the use of CBM-Z in this study. compared two chemical mechanisms in a WRF-Chem evaluation study over Europe and reported large differences in the representation of peak summer (JJA) ozone: one chemistry model (MOZART) overestimates mean and MDA8 ozone, while simulations with the other chemistry scheme (RADM2) shows underestimations of peak ozone that are in line with our findings. We will discuss the dependence of ozone simulation on the chemical mechanism choice in detail in Sect. . The ensemble model mean daytime ozone concentration in is underestimated by 10–30 µg m-3 in four subregions of the European continent. analyzed WRF-Chem O3 concentrations for 2007 and found that yearly averaged midday ozone is underestimated by approximately 10 µg m-3. The model performance in the aforementioned studies is qualitatively similar to our findings and the magnitude compares well. Overall, most studies consistently show underestimated daytime O3, regardless of the chemical mechanism, model resolution and other model assumptions. To further explore the potential role of a model misrepresentation of NO2 concentrations in explaining this model O3 bias, the next sections will focus on a model comparison with in situ and remote sensing data for NO2.

Figure a and b present a comparison of monthly averaged surface concentrations of NO2 between WRF-Chem and AirBase (note the logarithmic scale). Performance statistics are shown in Table . We find that WRF-Chem reproduces the spatial distribution well, with peak NO2 occurring in northwest Europe and northern Italy. In these regions with high NOx emissions, average WRF-Chem-simulated concentrations are however underestimated by up to 10 µg m-3 compared to observations. AirBase concentrations show a region with elevated NO2 concentrations in southwest Germany. WRF-Chem also shows elevated NO2 concentrations in this region but does not reach such elevated concentrations. Overall, WRF-Chem shows more spatial heterogeneity in surface NO2 concentrations than is apparent from the observations. Observed NO2 concentrations in background areas in Spain, France and eastern Europe are 2–5 µg m-3 or higher, while the model consistently simulates values <2 µg m-3 in these regions. This overall underestimation is also seen in Fig. , where the simulated daily mean NO2 concentration is shown against AirBase observations. The model performance of our WRF-Chem setup is in line with previous WRF-Chem studies. found small overestimations (0.67–2.96 µg m-3) in mean NO2. Another study found an annual average mean bias of -0.9 µg m-3, caused by underestimations of peak NO2 in WRF-Chem .

A comparison between WRF-Chem and AirBase monthly averaged 12:00 UTC NO2 concentrations is presented in Fig. c and d and Table . We find that WRF-Chem on average strongly underestimates midday NO2 concentrations by 2.96 µg m-3 (38.5 %).

Before we perform a comparison between NO2 VCDs from WRF-Chem and OMI, we first discuss the effect of the NO2 profile shape on the OMI-retrieved columns. Figure  shows the change in the monthly averaged OMI NO2 column density after replacing TM5-MP NO2 profiles by WRF-Chem profiles using the procedure described in Sect. . The OMI NO2 VCDs change most prominently over urban/industrial areas such as the Netherlands, Paris, Berlin, Madrid, Milan and Rome. The background areas are largely unaffected or show small (±0.2×1015 molec. cm-2) NO2 VCD increases (e.g., Spain) or decreases (regions in France, Germany, Poland, Ukraine and Romania). The vertical NO2 profile over sea regions in western Europe strongly peaks at the surface, because shipping NOx in WRF-Chem is emitted in the lowermost model layer. Overall, the average NO2 column change over non-land regions is small (<2 %).

We subsequently compare WRF-Chem to this modified OMI product. The monthly averaged NO2 vertical column densities from WRF-Chem and OMI are displayed in Fig. . The model is sampled at 12:00 UTC, close to the OMI overpass time of ± 13:30 LT, and is co-sampled with valid satellite observations. There is good agreement in the spatial distribution of monthly averaged NO2 VCDs (r2=0.68). NO2 columns are underestimated by 0.3×1015 molec. cm-2 on average, with strong underestimations of up to 2×1015 molec. cm-2 in urban and industrial northwestern Europe. WRF-Chem overestimates NO2 columns in some isolated urban areas with high NOx emissions such as London, Madrid, Rome and in parts of eastern Europe.

We note that Fig.  shows small underestimations of the simulated NO2 VCD compared to OMI (±0.2×1015 molec. cm-2) in background regions (e.g., the Alps, rural Spain and France, Scandinavia) and over the oceans. Simulated NO2 columns therefore show stronger spatial gradients than OMI-retrieved columns, which is in line with . Other distinct underestimations in the simulated NO2 columns compared to OMI indicate a misrepresentation of emissions. For example, the simulated NO2 column in northwestern Spain is underestimated by 2×1015 molec. cm-2 compared to OMI. The enhanced NO2 columns in this region mainly reflect the contribution to atmospheric NOx by power plant emissions. Although emissions from power plants should have decreased in recent years in this region , these emissions seem to be underestimated in WRF-Chem. However, since these results are only representative of July 2015, a more dedicated analysis is needed to further corroborate this hypothesis.

We have shown that our WRF-Chem setup with bottom-up emissions underestimates NO2 with respect to both surface and column measurements. To combine these model comparisons against different data sources, we already discuss parts of Fig. , which compares the agreement between simulations with bottom-up and top-down emissions. Figure a shows the relative difference of WRF-Chem against AirBase and OMI NO2 binned as a function of bottom-up anthropogenic emission strength. This shows an overall underestimation of WRF-Chem at the surface and in the troposphere, except for regions with strongest emissions. There is a relatively larger model underestimation of surface NO2 than of the NO2 VCD in regions with comparatively low emissions. Given that the surface NO2 mixing ratios are more sensitive to surface emissions than the NO2 column , this suggests that emissions are generally too low in WRF-Chem, but especially that emissions in rural background regions are underestimated. This, in turn, suggests that the representation of surface NOx emissions in WRF-Chem (anthropogenic emissions for 2011 and online calculated natural soil emissions) are too low to explain the observations in July 2015. In the following section, we will derive satellite-constrained NOx emissions and discuss potential reasons for this mismatch.

We derive top-down NOx emissions using the method described in Sect. . Figure  shows the July total bottom-up and top-down surface NOx emissions and their difference. Top-down NOx emissions amount to 498 Gg N, which is 56 % higher than the bottom-up inventory, and increases occur across the domain (Fig. c). NOx emissions are reduced in several isolated grid cells that generally correspond to urban areas. The difference between top-down and bottom-up emissions is larger than the 16 % increase reported by , although that study found strong (40 %–67 %) local increases in areas with high NOx emissions such as Belgium, western Germany and northern Italy.

Our top-down emissions are much higher than the bottom-up emissions over Germany and Poland. Over Belgium and the Netherlands, the difference between top-down and bottom-up emissions is also substantial but notably smaller despite larger differences between OMI and WRF-Chem NO2 columns over the low countries (Fig. c). This reflects the chemical regime with very high bottom-up NOx emissions in this region, resulting in suppressed midday OH concentrations and consequently longer NO2 lifetimes (as diagnosed by low beta values over northwestern Europe in Fig. S1 of the Supplement).

We subsequently replace bottom-up emissions with our observation-constrained top-down NOx emissions and perform a new WRF-Chem simulation. As expected, the new NO2 columns agree much better with the OMI NO2 columns than those from the simulation with bottom-up emissions (Fig. ). WRF-Chem with bottom-up emissions generally underestimates OMI NO2 columns by 23.4 %. As expected, the simulations with the top-down emissions agree better with OMI, and the slope of 0.98 between the new WRF-Chem and OMI NO2 columns (Fig. b) suggests that the systematic underestimation in the model is effectively resolved by applying the top-down emissions. The mean relative error is reduced to -7.5 %, and the spatial correlation coefficient between WRF-Chem and OMI NO2 also improves considerably (from 0.68 to 0.84).

Figure  shows the bottom-up and top-down NOx emissions as a function of the bottom-up anthropogenic emission strength. This comparison demonstrates that top-down NOx emissions are higher than bottom-up emissions regardless of the emission strength. However, top-down emissions are 50 %–100 % higher than bottom-up estimates for relatively weak emissions between 0.5 and 50 Mg N month-1 cell-1 and only up to 20 % higher for some urban and industrial hotspots (Fig. b). This 0.5–50 Mg N month-1 range is dominated by WRF-Chem grid cells located in the rural areas of Europe, excluding the largest urban agglomerations as well as low-emission regions such as mountainous areas. Our substantially larger top-down emissions partly reflect a required increase in NOx emissions in areas where soil NO emissions are expected to be a dominant NOx source. Soil NO emissions are simulated in WRF-Chem using an implementation of the MEGAN biogenic emission model. The observed discrepancy between the WRF-Chem-simulated and OMI-observed NO2 VCD triggers to assess how much of this discrepancy can be attributed to this model's representation of soil NO emissions.

To separate the soil NOx contribution from the anthropogenic emission updates, we perform a simple budget calculation as a first-order constraint on the partitioning of the top-down emissions between their anthropogenic and soil-based sources. We assume that the relative difference in anthropogenic sources is uniform over the emission bins in Fig. . This factor is calculated as the median of the relative change in emissions for the three highest bins (>50 Mg N cell-1 for July; see Fig. ) and amounts to 0.22. This allows us to attribute the remaining emission difference to soils. Based on this crude first estimate, we derive top-down soil NOx emissions to be 112 Gg N month-1 versus WRF-Chem-/MEGAN-simulated bottom-up soil NO emissions of only 15 Gg N month-1. The anthropogenic enhancement factor is relatively uncertain but does not strongly impact our derived posterior soil NOx emission estimate: if, instead of the median (m=0.22), we use the mean relative change in emissions for the three highest bins (μ=0.41), our soil contribution is still a factor >4 larger (69.0 Gg N month-1) compared to WRF-Chem's simulated bottom-up soil NO source. Therefore, this first-order estimation suggests that a substantial fraction (43 %–69 %) of the NOx emission increment after optimization can be attributed to soils.

To evaluate the derived total soil NOx emissions, we perform a comparison with literature-based estimates in Table . We find that bottom-up soil NOx emissions are underestimated by a factor of 5–7 compared to previous studies. In some of those studies e.g.,, land use management practices (fertilizer and manure application) provide a substantial contribution to European soil NO emissions, a feature that appears to be missing in the representation of soil NO emissions in WRF-Chem. This supports our hypothesis that a substantial fraction of the increase in surface NOx emissions may be attributed to soils. We will discuss this further in Sect. .

Table  summarizes the model performance of our bottom-up and top-down WRF-Chem simulations against a large number of AirBase NO2 observations throughout Europe in July 2015. The simulation with top-down emissions improves upon the a priori run in all metrics. Most notably, the model index of agreement (d) improves by 0.10 (14 %). The modified model setup still slightly underestimates the monthly averaged observed NO2 observations, as indicated by a slope of 0.89. However, the low bias in WRF-Chem surface NO2 concentrations with respect to AirBase improves from -2.5 to -1.1 µg m-3.

Compared to the monthly average, we find little improvement in WRF-Chem's skill to predict surface NO2 at 12:00 UTC. The model's low bias in NO2 reduces from -3.0 to -2.6 µg m-3 and the index of agreement improves by only 0.02 (4 %). This more modest improvement in performance can be understood from midday surface NO2 concentrations being more strongly driven by photochemical removal processes and boundary layer development than the 24 h mean NO2 levels, which are more sensitive to NOx emissions due to strongly reduced mixing and photochemistry at night. Figure  displays WRF-Chem monthly 24 h mean NO2 concentrations against AirBase observations for the bottom-up (black) and top-down (red) simulations. The model orthogonal distance regression (ODR) slope improves considerably, while the explained variance of the model improves slightly to 0.46.

Figure  shows the relative biases between WRF-Chem and observed NO2 as a function of (binned) bottom-up anthropogenic NO emission strength. Both the WRF-Chem simulations with bottom-up emissions (Fig. a) as well as the simulation with top-down emissions (Fig. b) show a low bias against OMI and AirBase for regions with low emissions and a positive relative bias in regions with stronger emissions. The relative bias is however considerably reduced in the simulation with top-down NOx emissions, both at the surface and in the column. However, WRF-Chem still displays a stronger relative bias compared to AirBase than compared to OMI. This feature can likely be attributed to a difference in spatial scales between the 20×20 km2 resolution model and the footprint area of local AirBase measurements, which can be easily influenced by a nearby NOx source that is less well captured in the model, due to instantaneous mixing over a larger volume. Another potential explanation for lower relative bias of WRF-Chem compared to AirBase than compared to OMI is interference of in situ measurements with molybdenum converters (see Sect. ). This is in line with our previous finding that the slope of the top-down NO2 column regression fit approaches 1, while the slope of the fit for in situ NO2 observations is still below 1. We also note that the spread in the relative bias compared to AirBase increased for the top-down simulation, with more positive relative bias values for all bins. Nonetheless, the results shown in Fig.  provide confidence regarding application of the model as a tool to reconcile local-scale bottom-up emissions and concentrations with larger-scale remote-sensing-based NO2 measurements.

Next, we address our main question whether the improved simulation of NO2 leads to better model performance for surface ozone simulations. We find that WRF-Chem with top-down emissions improves upon the bottom-up simulation for both the 24 h mean, as well as the 12:00 UTC and MDA8 ozone metrics. The model index of agreement improves by 0.08–0.11 (13 %–17 %; Table ). However, the top-down model still simulates too-low surface O3, especially over southern, eastern and central Europe, where observed surface O3 exceeds 80 µg m-3 at 12:00 UTC (see Fig. ).

A comparison between monthly averaged midday O3 concentrations from the bottom-up and top-down simulations (Fig. a and b, respectively) shows that ozone increases across the model domain. This particularly improves the WRF-Chem–AirBase agreement in large parts of western and central Europe. The simulated ozone values in northern Italy remain underestimated.

Surface ozone concentrations display a strong increase due to the use of top-down NOx emissions (Fig. ). The areas where ozone concentrations increase by >10 µg m-3 largely coincide with regions where top-down NOx emissions are much higher than the bottom-up emissions (Fig. c), such as in northern Spain, southern Germany, southern Poland, Croatia, Serbia, western Greece and southern Romania. There are also strong simulated ozone increases in central France and over the Adriatic Sea. These regions are all characterized as (rural) background areas, where ozone formation is strongly sensitive to the increases introduced in the NOx emissions for the relatively low bottom-up anthropogenic and soil emissions. We find decreases in ozone around the main shipping lanes, where the higher NOx emissions further enhance ozone titration. The enhanced titration also reduced simulated surface ozone around urban regions such as Barcelona, Rome and Paris. The increases in surface NOx emissions in the BeNeLux region and western Germany slightly increase simulated midday surface ozone. Ozone production is less sensitive to NOx emissions in these high NOx-emitting regions compared to the unpolluted background .

Figure  shows that O3 simulations with the higher top-down NOx emissions lead to a somewhat better match between modeled and observed surface O3, with an improvement in spatial correlation coefficient from 0.43 to 0.57, and an increase in slope from 0.33 to 0.41. Overall, the model low bias has reduced from -15 to -8 µg m-3, which indicates that the use of OMI NO2 VCD data to constrain WRF-Chem surface NOx emissions results in a considerable improvement regarding simulation of surface layer O3 concentrations.

We additionally analyzed changes in the temporal evolution of ozone concentrations resulting from NOx emission changes (Fig. ). Daytime median O3 concentrations are better captured in the Po Valley, central Spain and Poland. The NOx emission changes lead to a model overestimation of surface O3 concentrations for central France and south Germany, while concentrations change only slightly in the BeNeLux and Ruhr areas. In those regions, the mean bias error increases, while the hourly correlation coefficient and RMSE values improve for all regions (Table S4 of the Supplement). In all areas, changes in NOx emissions led to increased ozone concentrations particularly during daytime. Enhancements in simulated nighttime concentrations are only observed in central Spain. In other areas, nighttime O3 concentrations are overestimated in both simulations. Peak daytime O3 concentrations are better captured in all areas, as evidenced by the increase of the 75th percentile of simulated O3 concentrations with top-down emissions. However, peak O3 concentrations remain underestimated in the Po Valley, central Spain and southern Germany. Additionally, nighttime O3 concentration overestimations remain, likely due to issues related to model resolution and vertical mixing. Overall, the NOx emission changes most effectively increase O3 concentrations during periods with elevated ozone (Fig. S3 of the Supplement), which coincide with high solar radiation and temperatures and thus have a strongly NOx-dependent O3 formation.