Characterization and modeling of bio-optical properties of water in a lentic ecosystem using in situ Hyperspectral remote sensing

Hyperspectral remote sensing has shown great promise in characterizing and monitoring of optical properties of water. This study aims at characterizing the spectral reflectance and to develop hyperspectral algorithms for retrieval of bio-optical properties of Bhindawas wetland, a man-made lake in Haryana, India. The spectral reflectance of the lake was measured using SVC GER 1500 Spectroradiometer and water samples were collected from different sampling sites within the lake during three different field surveys in 2014. Characterization of spectral responses was carried out using principal component analysis and Canonical Correspondence Analysis (CCA). It revealed that the dataset was typical of Case II waters by extracting two principal components that explained around 99% of the variation, and CCA identified that different optical parameters such as TSS, TOC, Chla and turbidity delineate its spectral response. Water quality results were correlated with reflectance to determine their relationships. Furthermore, multiple linear regression was used to derive the two and three band model for retrieval of TSS, Chla and Turbidity concentration for lake. Retrieval algorithms with significant accuracy were developed for Chla (R-2=0.80, RMSE=0.12 mu g/l), TSS (R-2=0.86, RMSE=59.1mg/l) and Turbidity (R-2=0.84, RMSE=13NTU).