Multi-annual changes of NO_x emissions in megacity regions: nonlinear trend analysis of satellite measurement based estimates

Abstract. Hazardous impact of air pollutant emissions from megacities on atmospheric composition on regional and global scales is currently an important issue in atmospheric research. However, the quantification of emissions and related effects is frequently a difficult task, especially in the case of developing countries, due to the lack of reliable data and information. This study examines possibilities to retrieve multi-annual NO_x emissions changes in megacity regions from satellite measurements of nitrogen dioxide and to quantify them in terms of linear and nonlinear trends. By combining the retrievals of the GOME and SCIAMACHY satellite instrument data with simulations performed by the CHIMERE chemistry transport model, we obtain the time series of NO_x emission estimates for the 12 largest urban agglomerations in Europe and the Middle East in the period from 1996 to 2008. We employ then a novel method allowing estimation of a nonlinear trend in a noisy time series of an observed variable. The method is based on the probabilistic approach and the use of artificial neural networks; it does not involve any quantitative a priori assumptions. As a result, statistically significant nonlinearities in the estimated NO_x emission trends are detected in 5 megacities (Bagdad, Madrid, Milan, Moscow and Paris). Statistically significant upward linear trends are detected in Istanbul and Tehran, while downward linear trends are revealed in Berlin, London and the Ruhr agglomeration. The presence of nonlinearities in NO_x emission changes in Milan, Paris and Madrid is confirmed by comparison of simulated NO_x concentrations with independent air quality monitoring data. A good quantitative agreement between the linear trends in the simulated and measured near surface NO_x concentrations is found in London.