Air-stagnation episodes based on regional climate models part I: evaluation over Europe

Estimating the impact of climate change and emission scenarios on air pollution can be done using regional climate models (RCMs). Climate uncertainties are commonly estimated using RCM ensembles such as provided by EURO-CORDEX. Despite the strong relations between the weather and air pollutants, interactions are usually complex and require meteorological parameters that are not commonly available for the RCM ensembles. Pollution peaks, however, often coincide with stagnant atmospheric conditions that can be captured with widely-available RCM data. We first show that a commonly-used atmospheric stability index that uses rainfall, near-surface and 500 hPa wind speed, relates well to average and extreme air pollutant concentrations over Europe using Copernicus Atmosphere Monitoring Service (CAMS) data. We then provide an in-depth validation of 25 RCMs to reproduce the spatio-temporal features of air stagnation by comparison with ERA5. Overall the models were found to reproduce stagnant episodes fairly well, especially after bias correction. The systematic underestimation of stagnation frequency and duration is traced back to overestimated near-surface wind speed for a large group of models at high-elevation regions where the temporal correlations are also low. Regardless of the reference dataset, two model groups are identified that, independent on their resolution, give strongly different results in terms of orographic dependence of surface wind speed. These strong discrepancies underscore the need for bias correction when using RCM data for analysis of stagnation episodes.