Improving river water quality monitoring using satellite data products and a genetic algorithm processing approach

Adequate river quality monitoring is of major importance for riverine environmental sustainability. This study develops a methodology for real-time water quality measurement in a river at 30 m spatial and 1 day temporal scales using the satellite remote sensing technique to support daily water quality monitoring usually done at gauges. Considering the limited spatio-temporal resolutions of all the current satellite products, this study integrates the corrected band-specific Landsat and Moderate-resolution Imaging Spectroradiometer (MODIS) surface reflectance values, identified by a physically-based approach, with the observed pollutant concentrations. A combination of regression analysis and genetic algorithm (GA) based multivariate nonlinear formulations among the Landsat versus MODIS surface reflectances and Landsat surface reflectance versus in-situ pollutant concentration is used to estimate eight water quality parameters. All the possible combinations of the Landsat and MODIS satellite bands containing the spectral signature of pollutants are selected as independent variables. Linear and nonlinear regression analysis is carried out for these combinations using the SPSS software to get the best (significant) correlated relations which are, further, enhanced using the GA. This formulation is applied and tested in the Brahmani River located in eastern India's Odisha state for its real-time application; and water quality mapping is carried out for a typical river reach of the Brahmani River. A Monte-Carlo simulation based uncertainty and sensitivity analysis of the used algorithms reveal that the methods have the potential to be used in ungauged river reaches.