Research on Water Pollution Identification in Gannan Region Based on Multi-Temporal Landsat8 Images
The water resources in southern Jiangxi Province are abundant.Under the influence of human activities,agriculture and resource exploitation,there are some problems of water pollution in this area.The traditional single point analysis method is used to analyze water pollution,which only represents local and typical conditions,while the water quality parameter inversion based on multi-temporal remote sensing can be used for regional high-precision quantitative inversion.Based on Landsat8 OLI satellite images,this paper adopts support vector machine(SVM)method to identify water bodies.Based on the actual situation in southern Jiangxi Province,water quality parameters such as ammonia nitrogen,total phosphorus and chlorophyll a were selected as the research objects,and a response relationship model between remote sensing pop information and measured data was established by statistical regression method.The inversion results showed that the total concentration of ammonia nitrogen was 0.2～0.4 mg/L,total phosphorus was 0.06～0.18 mg/L,and chlorophyll a was 5.01～6.81 µg/L.Determination coefficient(R2)and mean absolute error(MAE) were used to verify the model feasibility and inversion accuracy.The inversion results R2 were 0.840,0.745 and 0.745,respectively.MAE were 0.054,0.049 and 0.069,respectively.The method of inversion model is reliable and can accurately identify sewage outlets,providing reliable technical support for protecting water resources and preventing water pollution.
maximum likelihoodsupport vector machineammonia nitrogentotal phosphoruschlorophyll a
万方数据
维普
