| The water quality of Baiyangdian Lake is affected by human activities and natural factors,and it has great changes in different time and space.It is difficult to obtain the information of water quality status comprehensively by traditional methods,such as field sampling analysis,However,The application of remote sensing technology to near-shore water quality monitoring can provide a faster,more macroscopic and accurate response about water quality,which is a development trend of water quality monitoring.In this paper,Baiyangdian is used as the research area,combining with measured spectra,Landsat8 images and chemical oxygen demand(COD)measured values,Using SPSS software to construct statistical regression model and using MATLAB software to construct BP neural network model for COD inversion,and Using BP neural network to estimate the COD concentration in the entire waters of Baiyangdian Lake.The results show that the spectral reflectance can distinguish the degree of COD pollution in the Baiyangdian water body.The average relative error between the predicted and measured values of the BP neural network model is 16.5%,and the model accuracy is good.Based on the model inversion of the spatial distribution of COD concentration in the Baiyangdian waters on October 30,2017,the Baiyangdian water body has a certain degree of organic pollution,and some water bodies reached the category of inferior V.The pollution center is mainly located in the southeastern villages and tourist attractions.The living and Production of sewages may be the main reasons for the increase of COD. |