Abstract:

The performance of 4 ACCMIP models (GFDL-AM3, NCAR-CAM3.5, GISS-E2-R and MIROC-CHEM) were evaluated in simulating surface PM2.5 concentration by using observation data, which obtained from MODIS and MISR satellite data. Specifically, the performance of reproducing the spatial distribution and time evolution were evaluated in this study. As shown from the results, ACCMIP models can reproduce the spatial pattern of surface PM2.5 concentration well over eastern China. The performance of GFDL-AM3 in simulating surface PM2.5 spatial pattern is the best, among the four models. The greater inconsistency was found in central Xinjiang and western Inner Mongolia. There is a good agreement in time series between the observational data and the model data. Multimodel ensemble mean can simulate the trends of surface PM2.5 concentration in Northeast China, Central China, coastal region of East China and western Xinjiang.

Key words: ACCMIP, PM2.5 concentration, the performance of evaluation