It has been shown that the laboratory calibration coefficient of charge-coupled device camera (CCD camera) on China-Brazil Earth Resources Satellite-02B (CBERS-02B) is not fit for the water component retrieval. As a result, the CCD was cross-calibrated by Aqua MODIS with higher calibration accuracy, which was based on the total radiance at the top of atmosphere calculated from normalized water-leaving radiance and aerosol parameters of Aqua MODIS. The analysis about the error related to the parameters used in the cross-calibration showed that the error of the calibration coefficients can be less than 5%, which was mostly determined by the accuracy of aerosol scattering of CCD band 830nm. Using the calibration coefficients, chlorophyll concentration was retrieved from CCD imagery and was compared with that from Aqua MODIS. The comparison showed that the calibration result worked better than the laboratory coefficient.
Application of MODIS in ocean color is mainly based on bands 8-16, whose spatial resolution is 1km. This spatial
resolution can't meet the demand of inland waters with small area. Then, taking TaiHu lake in China as an example, we
put forward an atmospheric correction algorithm for bands 1 and 2 whose spatial resolution is 250m. Firstly, we choose
one pixel whose digital number of band 16 is the smallest in Taihu lake as the clear pixel. The aerosol parameters of the
clear pixel are calculated by the standard atmospheric correction algorithm for Case 1 waters. Secondly, we can calculate
the aerosol scattering radiance of bands 1, 2 of other pixels with assumption that the aerosol type and optical thickness
keep the same over Taihu lake. This algorithm combines the advantage of bands 8-16 in ocean color atmospheric
correction with the high spatial resolution of bands 1 and 2. In order to test the precision of this algorithm, we choose an
MODIS-Aqua image which are covering Taihu lake and are acquired in the time of 2004 Taihu autumn cruise. We use
our atmospheric correction algorithm to process the selected image and compare the retrieved remote sensing reflectance
(Rrs) with measured Rrs. The average relative of bands 1 and 2 are respectively 24.85% and 41.44%, demonstrating that
this algorithm has the potential of application in the atmospheric correction of inland waters.