Remote Sensing Monitoring Technology And Application Of Water Quality In Macrophytic Lake

With the rapid development of China’s economy, urbanization and industrialization, water pollution is becoming more and more serious, especially in inland lakes, water quality compliance rate(I~III) of lakes is less than 30%. Automatic monitoring and manual monitoring of the traditional water quality monitoring method in lake, cannot meet the needs of comprehensive monitoring in lakes, remote sensing technology can overcome the shortcomings of the traditional method and realize the dynamic monitoring of water quality in lakes. In recent years, remote sensing technology is more and more applied to water quality monitoring in lakes, which is mainly used in the monitoring of water quality in algae lakes, represented by the Taihu Lake. By contrast, research that remote sensing technology used in macrophytic lakes and macrophytic-algal lakes is relatively less. There are high proportion of macrophytic lakes and macrophytic-algal lakes in China. Water quality parameters are difficult to monitor based on remote sensing technology because of aquatic macrophytes in macrophytic lakes. So monitoring water quality using remote sensing in macrophytic lakes and macrophytic-algal lakes helps obtain water quality of lakes in China. The chlorophyll a concentration, total suspended solids concentration and turbidity are important parameters to characterize the water quality of the lake. In this study, the chlorophyll a concentration, total suspended solids concentration and turbidity are monitored using remote sensing in Nansi Lake, a macrophytic lake as the study area. Nansi Lake is the important water source in eastern route of South to North Water Diversion.The water quality monitoring in Nansi Lake using remote sensing is of great significance to control the quality of eastern route of South to North Water Diversion.In this paper, the Nansi Lake is divided into aquatic macrophytes area and water area based on the result of regionalization by remote sensing. The phenology characteristics of the submerged vegetation and floating vegetation are identified by MODIS NDVI products, HJ1 A HSI data, HJ1A/1B CCD data and GF-1 WFV data. In the aquatic macrophytes area, the chlorophyll a concentration, total suspended solids concentration and turbidity are lower when the aquatic macrophytes are in growing period. But water quality will be deteriorated due to the decay and apoptosis of aquatic plants. This research identifies the aquatic macrophytes area in Nansi Lake using GF-1 WFV1 image on June 6, 2015, chlorophyll a concentration, total suspended solids concentration and turbidity are monitored indirectly according to the phenology characteristics of the aquatic macrophytes. In water area, the model, such as semi-empirical, semi-analytical model, WT-NDBPSO-PLS model, PSO-SVM model, EW-CM model, SPA-CM model and BMA model are established to retrieve the chlorophyll a concentration, total suspended solids concentration and turbidity in Nansi Lake. And the best retrieval model is chosen based on the retrieval model optimization technology. The main conclusions of this study are as follows:(1)The potamogeton lucens, potamogeton crispus, myriophyllum spicatum and potamogeton pectinatus are major aquatic plants in Nansi Lake. The phenology characteristics of potamogeton lucens, myriophyllum spicatum and potamogeton pectinatus are similar. They began to grow in the spring and grow fast in the summer, and then begin to apoptosis, gradually. Potamogeton crispus begin grow in the early spring, and grow fast in the late spring, and then begin to apoptosis rapidly and no longer grown. Potamogeton lucens, myriophyllum spicatum and potamogeton pectinatus can purify and indicate the water of Nansi Lake. Chlorophyll a concentration, total suspended solids concentration and turbidity of water are under 8ug/L, 15mg/L, 15 NTU, respectively, in the aquatic macrophytes area where the aquatic macrophytes are in growth. But water quality will be deteriorated due to the decay and apoptosis of aquatic plants, which leads to total suspended solids concentration and turbidity increased significantly in Nansi Lake. Water quality of the aquatic macrophytes area in Nansi Lake is poor during the decay of the aquatic macrophytes.(2) The band ratio model is better than the single band model and the first-order differential model in the semi-empirical or semi-analytical retrieval models of the three water quality parameters. The retrieval accuracy of chlorophyll a concentration using R696.2nm/R401.9nm linear model is the highest, comprehensive error is 28.74%. The retrieval accuracy of total suspended solids concentration using R*681.2nm/ R*540.5nm quadratic model is the highest, comprehensive error is 33.83%. The retrieval accuracy of turbidity using R′585.6nm exponential model and R688nm/R568.2nm quadratic model are rather ideal. The retrieval comprehensive errors of total suspended solids concentration and turbidity using are more than 50%, the accuracy of unified model is lower.(3) The retrieval accuracy of chlorophyll a concentration using WT-NDBPSO-PLS has improved compared with the traditional semi-empirical or semi-analytical model, whose comprehensive error is less than 27%. The retrieval accuracy of total suspended solids concentration using WT-NDBPSO-PLS is more than 50%, near 35% for turbidity retrieval. The WT-NDBPSO-PLS model is not suitable for retrieving total suspended solids concentration and turbidity which have high gradient.(4) The retrieval accuracy of three water quality parameters using PSO-SVM is higher than the traditional semi-empirical, semi-analysis model and WT-NDBPSO-PLS model, which will be significantly improved when the water quality parameters and spectral variables have significantly nonlinear relationship. The accuracy of nsr-PSO-SVM model is higher than osr-PSO-SVM model as a whole. The normalized processing of spectral reflectance can improve the accuracy of PSO-SVM model to a certain extent.(5) The combing model based on entropy weights, set pair analysis and BMA can synthesize characteristics of each model and retrieve water quality parameters cooperatively using multi-model. The combing model has a higher modeling and retrieval accuracy as a same time.(6) According to the comprehensive comparison of the average bandwidth, coverage rate and average offset range of each model, the combing retrieval model of chlorophyll a concentration and turbidity can give a more reliable retrieval interval, and the same for total suspended solids concentration retrieval is PSO-SVM.(7) Nsr-PSO-SVM model is the best retrieval model of chlorophyll a concentration, total suspended solids concentration and turbidity in Nansi Lake.In this study, remote sensing monitoring technique for monitoring water quality in the aquatic macrophytes area of Nansi Lake indirectly, can guide remote sensing monitoring of water quality in aquatic macrophytes area of other macrophytic and algal lakes. The efficiency of parameter setting and retrieval accuracy are improved remarkably based on the improvement of PLS model and SVM model of the three water quality patemeters in Nansi Lake. The three water quality parameters in Nansi Lake can be retrieval cooperatively using multi-model and the stability of model is improved.