Optimal Band Selection of Landsat-8 Images for Estimation of CDOM of Lakes Using Support Vector Regression

Colored dissolved organic matter (CDOM) is an important measure of water quality. CDOM can reduce the amount of light in water layers, disrupt the biological activity of photosynthesis, and inhibit the growth of phytoplankton populations that are essential for the aquatic food chain. Contrary to conducted research to date, which uses a specific wavelength, in this paper, we first examined the possibility of using visible portion of the spectrum to determine CDOM at 254-443 nm (254, 260, 350, 375, 400, 412, 440, 443 nm) in Landsat 8 . we then selected the most appropriate band ratios to measure CDOM at measurable wavelengths using the SVR algorithm (the parameters of which have been optimized using the genetic algorithm). It is noteworthy that in this study, the ratio of Coastal to red bands (R/R), blue to red (R/R), and the ratio of green to red bands (R/R) were considered for CDOM retrieval. Based on the results, considering the coefficient of determination (R = 0.71) and the amount of errors (MSE = 1.161 m, RMSE = 1.077 m and MAE = 0.946 m), it was concluded that the ratio of green to red bands in Landsat 8 is the most suitable choice for determining the colored dissolved organic matter. Moreover, according to the results from this study, the measurement of CDOM (440) is the most appropriate index for evaluating the quality of lake water resources in terms of their concentrations.