Atmospheric aerosols modify the radiative budget of the Earth due to their optical, microphysical and chemical properties, and are considered one of the most uncertain climate forcing agents. In order to characterise the uncertainties associated with satellite and modelling approaches to represent aerosol optical properties, mainly aerosol optical depth (AOD) and Ångström exponent (AE), their representation by different remote-sensing sensors and regional online coupled chemistry–climate models over Europe are evaluated. This work also characterises whether the inclusion of aerosol–radiation (ARI) or/and aerosol–cloud interactions (ACI) help improve the skills of modelling outputs.Two case studies were selected within the EuMetChem COST Action ES1004 framework when important aerosol episodes in 2010 all over Europe took place: a Russian wildfire episode and a Saharan desert dust outbreak that covered most of the Mediterranean Sea. The model data came from different regional air-quality–climate simulations performed by working group 2 of EuMetChem, which differed according to whether ARI or ACI was included or not. The remote-sensing data came from three different sensors: MODIS, OMI and SeaWIFS. The evaluation used classical statistical metrics to first compare satellite data versus the ground-based instrument network (AERONET) and then to evaluate model versus the observational data (both satellite and ground-based data).Regarding the uncertainty in the satellite representation of AOD, MODIS presented the best agreement with the AERONET observations compared to other satellite AOD observations. The differences found between remote-sensing sensors highlighted the uncertainty in the observations, which have to be taken into account when evaluating models. When modelling results were considered, a common trend for underestimating high AOD levels was observed. For the AE, models tended to underestimate its variability, except when considering a sectional approach in the aerosol representation. The modelling results showed better skills when ARI+ACI interactions were included; hence this improvement in the representation of AOD (above 30 % in the model error) and AE (between 20 and 75 %) is important to provide a better description of aerosol–radiation–cloud interactions in regional climate models.
