Sensitivity of air quality model responses to emission changes: Comparison of results based on four EU inventories through FAIRMODE benchmarking methodology

Abstract

Despite the application of an increasingly strict EU air quality legislation, air quality remains problematic in large parts of Europe. To support the abatement of these remaining problems, a better understanding of the potential impacts of emission abatement measures on air quality is required, and air chemistry transport models (CTMs) are the main instrument to perform emission reduction scenarios. In this study, we study the robustness of the model responses to emission reductions when emission input is changed. We investigate how inconsistencies in emissions impact the modelling responses in the case of emission reduction scenarios. Based on EMEP simulations over Europe fed by four emission inventories-EDGAR 5.0, EMEP-GNFR, CAMS 2.2.1 and CAMS version 4.2 (including condensables)-we reduce anthropogenic emissions in six cities (Brussels, Madrid, Rome, Bucharest, Berlin and Stockholm) and two regions (Po Valley in Italy and Malopolska in Poland) and study the variability in the concentration reductions obtained with these four emission inventories. Our study reveals that the impact of reducing aerosol precursors on PM10 concentrations result in different potentials and potencies, differences that are mainly explained by differences in emission quantities, differences in their spatial distributions as well as in their sector allocation. In general, the variability among models is larger for concentration changes (potentials) than for absolute concentrations. Similar total precursor emissions can, however, hide large variations in sectorial allocation that can lead to large impacts on potency given their different vertical distribution. Primary particulate matter (PPM) appears to be the precursor leading to the major differences in terms of potentials. From an emission inventory viewpoint, this work indicates that the most efficient actions to improve the robustness of the modelling responses to emission changes would be to better assess the sectorial share and total quantities of PPM emissions. From a modelling point of view, NOx responses are the more challenging and require caution because of their non-linearity. For O3, we find that the relationship between emission reduction and O3 concentration change shows the largest non-linearity for NOx (concentration increase) and a quasi-linear behaviour for volatile organic compounds (concentration decrease). We also emphasise the importance of accurate ratios of emitted precursors since these lead to changes in chemical regimes, directly affecting the responses of O3 or PM10 concentrations to emission reductions.