Eurodelta-Trends, a Multi-Model Experiment of Air Quality Hindcast in Europe over 1990-2010. Experiment Design and Key Findings
Abstract
The Eurodelta-Trends multi-model chemistry-transport experiment has been designed to better understand the evolution of air pollution and its drivers for the period 1990-2010 in Europe. The main objective of the experiment is to assess the efficiency of air pollutant emissions mitigation measures in improving regional scale air quality. The experiment is designed in three tiers with increasing degree of computational demand in order to facilitate the participation of as many modelling teams as possible. The basic experiment consists of simulations for the years 1990, 2000 and 2010. Sensitivity analysis for the same three years using various combinations of (i) anthropogenic emissions, (ii) chemical boundary conditions and (iii) meteorology complements it. The most demanding tier consists in two complete time series from 1990 to 2010, simulated using either time varying emissions for corresponding years or constant emissions. Eight chemistry-transport models have contributed with calculation results to at least one experiment tier, and six models have completed the 21-year trend simulations. The modelling results are publicly available for further use by the scientific community. We assess the skill of the models in capturing observed air pollution trends for the 1990-2010 time period. The average particulate matter relative trends are well captured by the models, even if they display the usual lower bias in reproducing absolute levels. Ozone trends are also well reproduced, yet slightly overestimated in the 1990s. The attribution study emphasizes the efficiency of mitigation measures in reducing air pollution over Europe, although a strong impact of long range transport is pointed out for ozone trends. Meteorological variability is also an important factor in some regions of Europe. The results of the first health and ecosystem impact studies impacts building upon a regional scale multi-model ensemble over a 20yr time period will also be presented.