Validation of Aerosol Optical Thickness Product over China with MODIS Data Operated at NSMC

The aerosol optical thickness producing software provided by Cooperative Institute for Meteorological Satellite Studies (CIMSS) of University of Wisconsin has been modified and operationally run at NSMC. And the MODIS/AOT product is shared at National MODIS Data Center.The MODIS aerosol optical thickness product has been validated with AERONET Level 2.0 aerosol optical thickness product so that the product can be improved and popularized.The MODIS/AOT product from January 2005 to May 2007 has been matched with L2.0 AOT product from AERONET stations in east Asia during the same period. The spacial average value of the MODIS/AOT within the 10 km distance from the site of AERONET station have been compared with the temporal average value of the AERONET/AOT within 30 minutes period the satellite passing the station. The validation result show that RMSE of all validation AOT samples over land is near 0.25, and about forty-four percent of the test samples meet the expected uncertainty of±0.05±0.20τ.The precision of MODIS/AOT is different according to seasons and areas.Usually, RMSE of MODIS/AOT is smaller at drought season than that at rain season. Cloud is the main factor impacted on the large RMSE at rain season. At south evergreen vegetable area, MODIS/AOT RMSE is better than that of north seasonal changed area.MODIS/AOT at 658nm has obviously systematical over-retrieval since the surface reflectivity of red band is over-estimated. Absolute error of MODIS/AOT at 466nm retrieved with blue band data has about half of the positive value and RMSE of MODIS/AOT at 466nm is higher than that of 658nm. The cause is that MODIS/AOT obtained with blue band is more sensitive for aerosol model used in retrieval process than it done with red band, though the estimated surface reflectivity of blue band has smaller error than that of red band. In a word, error from aerosol model is main cause of high random error of MODIS/AOT at 466nm.Correlation among the surface reflectivity at 0.47, 0.66 μm and 2.1μm bands are discussed using selected 79 items of vegetation spectrum reflectivity observed in China by LRCVES/CMA, in order to analyze effects of surface reflectivity for retrieval MODIS/AOT and get some ideas that could improve the precision of MODIS/AOT.The vegetation spectrum reflectivity is obtained from different vegetations and from different growing period of the same vegetation.The analysis show that there is high linear correlation (86.43%) between reflectivity at 0.47 μm bands and reflectivity at 0.66 μm bands for the vegetations.And linear correlation between vegetation reflectivity at 2.1 μm and vegetation reflectivity at 0.66 μm is 68.58%. Linear correlation between vegetation reflectivity at 2.1 μm and vegetation reflectivity at 0.47 μm is 59.79%. It can be concluded that the scheme which decides vegetation surface reflectivity of two visible bands in algorithm of Collection 5 MODIS aerosol products agrees well with the statistics, and is instrumental for algorithm of AOT retrieval with other similar satellite sensors.