Comparison of model-simulated tropospheric NO 2 over China with GOME-satellite data

Abstract

Tropospheric NO 2 column densities over China simulated with a regional model using different emission inventory input are compared with Global Ozone Monitoring Experiment (GOME) satellite data. These emission inventories include (i) emission estimates for the year 1995 from the Emission Database for Global Atmospheric Research (EDGAR), (ii) regional emission inventory used in the Transport and Chemical Evolution over the Pacific (TRACE-P) program with emission estimates for the year 2000 and (iii) national emission inventory used in the Chinese Ozone Research Programme (CORP) with emission estimates for the year 1995. Model simulations were performed for a summertime period and the results under clear-sky conditions were selected for comparison with GOME data of the years 1996 and 2000. The model generally reproduces high tropospheric NO 2 column densities in polluted areas of China that have been observed by GOME. However, the model simulations do not agree with the GOME measurements in a quantitative sense for some regions. Region-to-region comparisons show that with all the emission inventories the model underestimates the tropospheric NO 2 column density in remote and rural areas of China. It is found that TRACE-P underestimates the tropospheric NO 2 column density in all the regions with respect to the GOME measurements (by more than 50%). CORP and EDGAR appear to behave well for the model simulations in the North of China (within 15% deviations), but poorer for the model simulations in other regions (within 30–80% deviations). Linear regressions were performed with the NO 2 column densities available from GOME, Y, and the model, X, of each grid cell in selected regions. For all the region of China, comparison statistics are Y = 0.3 6 1 × X + 0.9 3 5 with r 2 = 0.3 6 0 and bias=−41% for EDGAR, Y = 1.1 2 4 × X + 0.9 2 0 with r 2 = 0.4 7 2 and bias=−67% for TRACE-P, and Y = 0.4 3 1 × X + 0.8 3 5 with r 2 = 0.5 0 9 and bias=−31% for CORP.