A full year reanalysis of European air quality in 2016 focusing on the evaluation of anthropogenic emissions by applying advanced spatio-temporal inversion

Emission data as main input to air quality forecast models introduce large uncertainties. These data originate from emission inventories that provide estimates of spatially distributed emissions in general as an annual total amount. The temporal variations of the emission data follow prefixed time profiles in models. Yet, the impact of variable societal behaviour and meteorological implications are rarely considered. Furthermore, environmental agencies that raise the data input for emission inventories depend on rough emission estimates from bottom-up and top-down strategies.

To evaluate the annual emission totals of European and in particular German inventories, we perform a full year reanalysis of European air quality in 2016 applying chemistry four-dimensional variational (4D-var) data assimilation with the European Air pollution Dispersion – Inverse Model (EURAD-IM). Assimilating ground-based, airborne, as well as satellite observation data within 24 hours assimilation windows, we successively assess initial value optimisations and emission correction factors for anthropogenic emissions of nitrogen oxides (NO_x), carbon monoxide (CO), particulate matter (PM), sulphur oxides (SO_x), ammonia (NH₃), and non-methane volatile organic compounds (NMVOCs), achieving consistency with observations. The analysis is performed on different model grids of 15 km horizontal resolution for Europe, 5 km for Central Europe, and 1 km for three selected regions in Germany.

Analysing the inferred temporal evolution of emission correction factors reveals that the total NO_x emissions are underestimated in Germany, while NH₃ emissions are found too high leading to an overestimation of modelled NH₃ concentrations using standard emission data. Other emission species show clear seasonal dependence in the correction factors. Comparing the emission correction factors of the different European countries, we find a significant discrepancy of correction strengths between north-western and south and eastern European countries. Analysis results for different model nests vary not only due to finer structures but also in the strength or sometimes even the direction of corrections. Spatially, the distribution of correction factors is driven by areas characterized by high emissions while care must be taken that the deployment of assimilated observation stations still matters. In this context, we also discuss the limits of our analysis technique regarding the observation network configuration and the statistical method of the assimilation technique.