Evaluation of the WRF-Chem performance for the air pollutants over the United Arab Emirates

Abstract. This study presents a comprehensive evaluation of the Weather Research and Forecasting model coupled with chemistry (WRF-Chem) in simulating meteorological parameters and concentrations of air pollutants across the United Arab Emirates (UAE) for June and December 2022, representing the contrasting summer and winter climatic conditions. The assessment of WRF-Chem performance involves comparisons with ground-based observations for meteorological parameters and satellite retrievals from the TROPOspheric Monitoring Instrument (TROPOMI) for gaseous pollutants and the Moderate Resolution Imaging Spectroradiometer (MODIS) for aerosols. The comparison with TROPOMI column concentrations demonstrates that WRF-Chem performs well in simulating the spatio-temporal patterns of total column CO and tropospheric column NO2 and O3, despite certain deficiencies in modelling tropospheric NO2 column concentrations. In particular, WRF-Chem shows a strong correlation with TROPOMI retrievals, with correlation coefficients ranging from 0.53 to 0.82 during summer and 0.40 to 0.69 during winter for these gaseous pollutants. The model tends to overestimate NO2 levels, with a higher discrepancy observed in summer (0.50 × 1015 molecules cm−2) compared to winter (0.18 × 1015 molecules cm−2). In comparison with TROPOMI-CO data, the discrepancies are more pronounced in winter, with an underestimation of 0.12 × 1018 molecules cm−2. Additionally, WRF-Chem consistently overestimates ozone levels in both seasons. WRF-Chem also exhibits a moderate correlation with both AERONET and MODIS aerosol optical depth (AOD) measurements. The correlation at Mezaira is 0.60, while a correlation of 0.65 is observed with MODIS AOD. However, the model tends to overestimate AOD, with a bias of 0.46 at Mezaira and 0.35 compared to MODIS AOD. Meteorological evaluations reveal that the model generally overestimated air temperature at 2 m above ground (T2m) in summer (≤ 0.2 °C) and underestimated it in winter (∼ 3 °C), with correlation coefficients between 0.7 and 0.85. Temperature biases are linked to surface property representation and model physics. For wind speed at 10 m (WS10m), biases were within ± 0.5 m s−1, indicating good agreement, although overestimations suggest deficiencies in surface drag parameterization. The dry bias observed was consistent with other studies due to dry soil, inaccurate mesoscale circulation representation, and bias in forcing data. The model also overestimated incoming shortwave radiation by ∼ 30 W m−2 in December due to reduced cloud cover. Night-time cold and dry biases were observed due to more substantial wind speeds and cooler air advection. Comparisons with ERA5 reanalysis showed regional T2m variations with high correlation coefficients (0.97 in summer, 0.92 in winter). Both WRF-Chem and ERA5 displayed consistent seasonal patterns in the planetary boundary layer, correlating with temperature changes and indicating good overall model performance.