Abstract: Using remote sensing data for large area water quality survey is a typical water environment monitoring method. The Chaohu Lake reflectance spectra are measured using ASD field spectrometer and the MODIS data collected during the same period. After the MODIS data are preprocessed, the spectral characteristics of Chaohu Lake are analyzed and the spectral is compared with references. Then the relation between the measured water reflectance and chlorophyll a is fitted. The fitting formula is consistent with previous research results and the correction coefficient is 0.75. But after analyzing the apparent reflectance of the MODIS L1B and the measured concentration of the chlorophyll-a, the ideal inversion lake model can not be obtained because of the influence of the atmosphere and other factors. After a special atmospheric correction for MODIS data, the ground reflectance spectra are obtained in the channels 1—7. The relation between the ground reflectance and the chlorophyll a is fitted and the correction coefficient is improved. At last, the optimum bands combination are chosen after analyzing the relation between the reflectance spectra and the corresponding moment of the concentration of the chlorophyll a, using EOS/MODIS data after atmospheric correction. The chlorophyll a index I_Chla is defined. Therefore the inversion model for chlorophyll a concentration of Chaohu Lake is achieved, and the correction coefficient is 0.5079. This correction coefficient is not as good as the result(0.75)obtained by the hyperspectral data measured by the field spectrometer. There are 3 possibilities that may cause the errors. First, the bands selection of the MODIS data are not cover exactly the spectral sensitive of the concentration of the chlorophyll a(685—715 nm). Second, the narrowband of the field spectrometer and broadband of the MODIS data is not very exact. Third, the atmospheric correction errors above Chaohu Lake may be too big.