Development of a tool for the assessment of water quality from visible satellite imagery taken over turbid inland waters (with Lake Michigan as an example)

Based on the previously developed hydro-optical models and a combination of the Levenberg-Marquardt multivariate optimization method and neural network emulation technique, a fast operating tool has been developed to process visible satellite imagery taken over turbid waters. Assessed is the efficiency of the developed tool under conditions of noise-contaminated input data. Determined are the number of sensor's spectral channels in the visible and the ranges of concentrations of major water quality constituents assuring the admissible errors of retrieval results. The processed MODIS image of Lake Michigan given herein exemplifies the feasibility/efficiency of the developed tool.