The impact of urban land-surface on extreme air pollution over central Europe

Abstract

This paper deals with the urban land-surface impact (i.e., the urban canopy meteorological forcing; UCMF) on extreme air pollution for selected central European cities for present-day climate conditions (2015–2016) using three regional climate-chemistry models: the regional climate models RegCM and WRF-Chem (its meteorological part), the chemistry transport model CAMx coupled to either RegCM and WRF and the “chemical” component of WRF-Chem. Most of the studies dealing with the urban canopy meteorological forcing on air pollution focused on change in average conditions or only on a selected winter and/or summer air pollution episode. Here we extend these studies by focusing on long-term extreme air pollution levels by looking at not only the change in average values, but also their high (and low) percentile values, and we combine the analysis with investigating selected high-pollution episodes too. As extreme air pollution is often linked to extreme values of meteorological variables (e.g., low planetary boundary layer height, low winds, high temperatures), the urbanization-induced extreme meteorological modifications will be analyzed too. The validation of model results show reasonable model performance for regional-scale temperature and precipitation. Ozone is overestimated by about 10–20 µg m−3 (50 %–100 %); on the other hand, extreme summertime ozone values are underestimated by all models. Modeled nitrogen dioxide (NO2) concentrations are well correlated with observations, but results are marked by a systematic underestimation up to 20 µg m−3 (−50 %). PM2.5 (particles with diameter ≤2.5 µm) are systematically underestimated in most of the models by around 5 µg m−3 (50 %–70 %). Our results show that the impact on extreme values of meteorological variables can be substantially different from that of the impact on average ones: low (5th percentile) temperature in winter responds to UCMF much more than average values, while in summer, 95th percentiles increase more than averages. The impact on boundary layer height (PBLH), i.e., its increase is stronger for thicker PBLs and wind speed, is reduced much more for strong winds compared to average ones. The modeled changes in ozone (O3), NO2 and PM2.5 show the expected pattern, i.e., increase in average 8 h O3 up to 2–3 ppbv, decrease in daily average NO2 by around 2–4 ppbv and decrease in daily average PM2.5 by around −2 µg m−3. Regarding the impact on extreme (95th percentile) values of these pollutants, the impact on ozone at the high end of the distribution is rather similar to the impact on average 8 h values. A different picture is obtained however for extreme values of NO2 and PM2.5. The impact on the 95th percentile values is almost 2 times larger than the impact on the daily averages for both pollutants. The simulated impact on extreme values further well corresponds to the UCMF impact simulated for the selected high-pollution episodes. Our results bring light to the principal question: whether extreme air quality is modified by urban land surface with a different magnitude compared to the impact on average air pollution. We showed that this is indeed true for NO2 and PM2.5, while in the case of ozone, our results did not show substantial differences between the impact on mean and extreme values.