In the present study,a novel retrieval method for estimating chlorophyll-a concentration in case Ⅱ waters based on bio-optical model was proposed and was tested with the data measured in the laboratory.A series of reflectance spectra,with which the concentration of each sample constituent (for example chlorophyll-a,NPSS etc.) was obtained from accurate experiments,were used to calculate the absorption and backscattering coefficients of the constituents of the case Ⅱ waters.Then non-negative least square method was applied to calculate the concentration of chlorophyll-a and non-phytoplankton suspended sediments (NPSS).Green algae was firstly collected from the Kasumigaura lake in Japan and then cultured in the laboratory.The reflectance spectra of waters with different amounts of phytoplankton and NPSS were measured in the dark room using FieldSpec Pro VNIR (Analytical Spectral Devises Inc.,Boulder,CO,USA).In order to validate whether this method can be applied in multispectral data (for example Landsat TM),the spectra measured in the laboratory were resampled with Landsat TM bands 1,2,3 and 4.Different combinations of TM bands were compared to derive the most appropriate wavelength for detecting chlorophyll-a in case Ⅱ water for green algae.The results indicated that the combination of TM bands 2,3 and 4 achieved much better accuracy than other combinations,and the estimated concentration of chlorophyll-a was significantly more accurate than empirical methods.It is expected that this method can be directly applied to the real remotely sensed image because it is based on bio-optical model.