Remote Sensing Monitoring Of Permanganate Index Based On Spectral Classification

According to monitoring results of national control section of surface water in 2019,water quality of the rivers in some forest cover areas of Heilongjiang Province has exceeded the standard all the year round,and the main index was permanganate index(COD_Mn).Aiming at solving the problem of COD_Mnexceeding the standard of surface water in Heilongjiang Province,two typical rivers,namely Tangwang River and Songhua River,were designed and carried out in the study area in 2020 and 2021 respectively.38 groups of measured hyperspectral data and water quality parameter data were obtained.At the same time,105 Sentinel-2 images covering Tangwang River and 58 Sentinel-2 images covering Songhua River were downloaded and processed.Main components of two kinds of water bodies were determined by analyzing the spectral characteristics and studying optical properties.In this study,two methods were used to establish retrieval model of COD_Mnin the study area.The first was to analyze the correlation between main components of water body and COD_Mnto establish indirect retrieval model of COD_Mn,and the second was to analyze the correlation between measured remote sensing reflectance and COD_Mnto establish direct retrieval model of COD_Mn.Comparing the accuracy of COD_Mnretrieval model established by two methods,the optimal model was selected to be applied to Sentinel-2 images of Tangwang River and Songhua River from 2017 to 2021respectively to achieve high-frequency,large-scale and high-precision COD_Mnconcentration monitoring of rivers.The main conclusions were as follows:1.According to difference in the shape of measured surface spectral reflectance,water body was divided into two types.Tangwang River was the first kind of water body,and Songhua River was the second kind of water body.The absorption coefficients of Tangwang River and Songhua River were analyzed from three major components that characterized optical characteristics of water body（CDOM,chlorophyll-a and total suspended matter）.It was determined that water body of Tangwang River was the dominant type of CDOM,and Songhua River was the dominant type of TSM concentration.2.COD_Mnretrieval model in Tangwang River was established by two methods.Firstly,the indirect retrieval model of COD_Mnwas established according to CDOM in Tangwang River.Secondly,the correlation between remote sensing reflectance of each band and COD_Mnwas analyzed,and the direct retrieval model of COD_Mnwas established.Finally,the accuracy of retrieval models established by two methods were compared,and the semi-empirical model established directly by remote sensing reflectance was applied to Sentinel-2 image of Tangwang River from 2017 to 2021.3.COD_Mnretrieval model in Songhua River was established by two methods.Firstly,the indirect retrieval model of COD_Mnwas established according to TSM concentration in Songhua River.Secondly,the correlation between remote sensing reflectance of each band and COD_Mnwas analyzed,and the direct retrieval model of COD_Mnwas established.Finally,the accuracy of retrieval models established by two methods were compared,and the semi-empirical model established directly by remote sensing reflectance was applied to Sentinel-2 image of Songhua River from 2017 to 2021.4.In terms of spatial distribution,COD_Mnconcentration in Tangwang River decreased gradually from upstream to downstream.According to statistical results in each section of Tangwang River,average COD_Mnconcentration in five sections was 3.78-37.28 mg/L,most of which exceeded the surface water class III limit（6mg/L）.In some periods,average COD_Mnconcentration in upper and lower reaches 15-35 mg/L,which far exceeded the surface water class V limit（15mg/L）.On the interannual scale,average COD_Mnconcentration from 2017 to2021 showed a trend of first decreasing-then increasing-then decreasing-then increasing.On the seasonal scale,COD_Mnconcentration varied greatly in different periods of each year,and there was no obvious changeable regularity.5.In terms of spatial distribution,COD_Mnconcentration in the lower reaches of Songhua River was higher than that in the upper reaches.Average COD_Mnconcentration in the upper reaches was stable at 5.27-6.89 mg/L from 2017 to 2021,which remained within the limits of III and IV.Average COD_Mnconcentration in the lower reaches was 6.78-11.24 mg/L from 2017 to2021,which exceeded the surface water class III limit.On the interannual scale,variation range of COD_Mnconcentration was small from year to year,and water quality was relatively stable.On the seasonal scale,COD_Mnconcentration varied greatly from year to year,and there was no obvious changeable regularity.There are 50 pictures,4 tables and 97 references in this thesis.