Research On Water Quality Monitoring Based On UAV Multispectral Images And IMP-MPP Algorithm

Water resources are an important part and basic guarantee of the ecological environment,so it is very necessary to monitor water quality.Traditional satellite remote sensing is too rough to image small-sized waters.In recent years,UAV remote sensing has been used to estimate water quality parameters,such as suspended solids,turbidity,and chlorophyll-a levels,due to its low cost,convenience,and high resolution.The matching pixel-by-pixel(MPP)algorithm is used to find the optimal regression model for retrieving water quality parameters from UAV images.However,the MPP has a high computational burden and overfitting problems.Based on this,we propose an improved MPP(IMP-MPP)algorithm to solve the above problems.In this study,Qingshan Lake,Hangzhou City,Zhejiang Province,China,is taken as the study area to evaluate water quality based on suspended solids concentration and turbidity indicators,and a total of 45 samples were collected.The experimental results show that:(1)compared with the average method,the determination coefficient of the optimal inversion model of suspended solids concentration based on IMP-MPP algorithm increased from 0.759 5 to 0.825 5,the average relative error reduced from 16.37% to 15.08%,and the comprehensive error reduced from 0.204 7 to 0.198 1.(2)Compared with the average method,the determination coefficient of the best turbidity inversion model based on IMP-MPP algorithm increased from 0.728 7 to 0.831 1,the average relative error decreased from 17.83% to 16.49%,and the comprehensive error decreased from 0.210 1 to 0.203 3.The IMP-MPP algorithm is further applied to the construction of the inversion model of water quality parameters in East Lake waters of East Lake Campus of Zhejiang A&F University,and the changes of overall water quality parameters content and spatial distribution in its small waters are explored.(1)In this paper,the UAV multispectral images and measured water quality parameter data of East Lake in East Lake Campus of Zhejiang A&F University from 2019 to 2020 are obtained,and the inversion models of suspended solids concentration and turbidity are established respectively,with the determination coefficients of 0.694 1 and 0.732 3,and the average relative errors of 12.78%and 8.07%.(2)Then,the spatial distribution map of concentration was made and further analyzed.The research shows that the IMP-MPP method can accurately obtain the distribution of water quality parameters and their interannual changes in East Lake waters.The method can be applied to long-term water quality monitoring in other small waters,and make a more comprehensive and detailed monitoring and characterization of small-sized waters.