高光谱遥感与LSTM网络耦合的水体富营养化污染区域检测OA

A detection method for eutrophic pollution areas in water bodies coupled with hyperspectral remote sensing and LSTM networks was proposed.Band ranges were selected according to the optical characteristics of water bodies,and hyperspectral remote sensing images of water bodies meeting quality requirements were acquired.The acquired remote sensing images were imported into the LSTM network,results containing dynamic and spectral features were extracted,and water quality parameters including nutrient concentrations and water color indices were inverted.Pollution criteria were established based on the formation mechanism and manifestation of water eutrophication.The inverted water quality parameters were compared with the established criteria,and the detection results of water eutrophic pollution areas were obtained.Compared with traditional detection methods,smaller detection errors of water nutrient concentrations were achieved by the optimized method,and the detection results of eutrophic status and pollution areas were more consistent with the actual conditions of polluted areas.It was verified that the optimized method exhibited superior detection performance.