Water quality monitoring using hyperspectral remote sensing data in Taihu Lake China

Although designed for demonstration purpose on land applications, Hyperion data are being tested for its capabilities on monitoring water quality over high turbid inland water target- Lake Taihu, in East China. The field survey was coincident with EO-1 overpasses, retrieving the concentrations of chlorophyll a (chl-a), suspended sediment (ss). In order to determine the optimal bands and algorithms for the retrieval of the optically active substances, this paper mainly focus on the correlation analysis between the concentrations of the chl-a and suspended sediments and the Hyperion reflectance in three remote sensing algorithms (band ratio (r1/r2), band difference (r1-r2) and NDVI algorithm (r1-r2)/ (r1+r2)). The statistical results on the optimal band determination show that the band difference technique between R(732.07~884.7) and R(1174.77~1194.97) has stronger correlation with the suspended sediments(r>0.70,n=25), while the NDVI algorithm between R(620.15~691.37) and R(721.9~844)has stronger correlation with chlorophyll a (r>0.90,n=25). The ss inversion model was established using the reflectance difference of (R874.53-R1184.87), which has the strongest correlation(r=0.79,n=25) with the suspended sediments (r2=0.65), and the chl-a inversion model was established using the algorithm of(R620.15-R732.07) / (R620.15+R732.07) , which has the strongest correlation(r=0.90,n=25) with the chlorophyll a(r2=0.89). The research on the optimal algorithms determination using the Hyperspectral remote sensing data will facilitate the water quality monitoring using the multi-spectral remote sensing data, like TM, MODIS, etc.