Research On Water Quality Parameters Estimation Of Baiyangdian Lake Based On Multi-source Remote Sensing Data

Under the influence of human activities and industrial development,lakes face severe water pollution problems,which directly affect the ecological environment of lakes and human production and life.Lake monitoring is an important basis for the prevention and control of water pollution.Although traditional monitoring methods can analyze water quality parameters in detail,it is time-consuming,labor-intensive,and costly.It is easy to be restricted by weather and hydrological conditions,and it is difficult to complete timely and large-scale monitoring.Remote sensing technology can quickly and widely reflect the spatial and temporal distribution of water quality.Diverse remote sensing data and continuously updated water quality remote sensing inversion models have made water quality monitoring a real-time and convenient road,and promoted the development of remote sensing technology in water quality monitoring applications.In this paper,Baiyangdian Lake is used as the research area,and the concentrations of chlorophyll and chemical oxygen demand were estimated from various remote sensing data.Only the Landsat-7 satellite data synchronized with the on-site sampling is more suitable,but its lower spatial resolution brings a large error to the estimation of the concentration of water quality parameters to a certain extent.Therefore,this paper makes up for the research by means of spatiotemporal fusion based on multi-source remote sensing data.The problem of lack of high-resolution data during the period to ensure the accuracy of lake water monitoring.The results showed that:(1)Aiming at the problem of low resolution of geostationary satellite data,using the STNLFFM algorithm to conduct spatiotemporal fusion experiments using GF-4 and MODIS data and Sentinel-2 data,respectively.The Landsat-7 data is selected for qualitative and quantitative evaluation of the fusion data.It was shown that the spatiotemporal fusion effect of GF-4 and Sentinel-2 is better,and the correlation,root mean square error and average difference all meet the research needs.(2)By analyzing the spectral characteristics of chlorophyll a in the water body,a three-band algorithm based on a semi-analytical model was used to invert the chlorophyll a concentration in the Shaogchedian lake.The inverse chlorophyll concentration was quantitatively analyzed.The R2 is 0.67,RMSE is 1.42ug/L and MRE is 10%,which ensures the accuracy of water quality monitoring(3)Retrieve chlorophyll and COD concentrations of Baiyangdian Lake using BP neural network model based on Landsat-7 data and Sentinel-2 data of spatiotemporal fusion.The accuracy of the inversion results of two water quality parameters was evaluated using 12 chlorophyll measured verification points and 8 COD measured verification points.The analysis found that the Sentinel-2 data generated by spatiotemporal fusion showed good accuracy in calculating the concentration of two water quality parameters,which can better reflect the spatial characteristic distribution of baiyangdian chlorophyll a concentration and COD concentration.