Extraction of chlorophyll-a concentration based on spectral unmixing model using field hyperspectral data in Taihu Lake

In China, one of the most common ecological problems of inland water bodies is represented by the eutrophication which diminishes water quality. And the chlorophyll-laden water becomes an obvious sign. Chlorophyll-a concentration measurement is usually, used for assessing tropic status of lakes. The development of spectral resolution enables hyperspectral technology possible to monitor water quality successfully, which is based on developing relationships between radiance/reflectance in single band or band ratios and chlorophyll concentration. In this paper, a spectral unmixing model was established based on single-phase field hyperspectral data. And three data types were supported for this model: original data, normalization data and differential data. And selected end-member from known reflectance spectrum, we retrieved chlorophyll-a concentration. The result shows the spectral unmixing model based on differential data gives the best result. Validated this model and shows a good precision and stabilization. Finally, three-phase field hypersptral datum were processed and chlorophyll-a concentration was extracted using the best model. The result shows that spectral unmxing model is a feasible model in the practical application of remote sensing water quality monitoring.