Water Quality Prediction Methods Based On Remote Sensing Inversion And Its Application

With the development of human activities and industries,the pollution of water environment resources has risen sharply,which endangers the lives and health of residents.Accurate detection of water quality parameters in a river basin and the reasonable prediction of future changes in water quality parameters are necessary prerequisites for scientific planning of water resources.The use of remote sensing geographic information technology can provide a macroscopic assessment of the water quality of various waters in large-scale spaces and is an effective complement to traditional water quality detection methods.In this paper,we realized water quality prediction and remote sensing inversion of the river basin based on the recursive neural network and support vector machine algorithms,and the correctness of the algorithms are verified by water quality prediction and water quality inversion of a reservoir in Zhejiang Province.First,we proposed a water quality index prediction method based on sparse recursive neural network.Secondly,on the basis of using sparse recursive neural networks to preprocess historical water quality of reservoirs,we constructed a variety of remote sensing inversion models.Finally,we designed and implemented a river basin water quality information management system,and the effectiveness of the above two algorithms were verified in the system.The main research contents of this article include:First,we propose a water quality index and grade prediction method based on sparse recursive neural network(SRNN).The learning algorithm of the network is designed based on the principle of least mean square recursive error,and the neural network is used to construct a prediction model for predicting water quality indicators and levels.We have validated the effectiveness of the model by predicting the water quality parameters and water quality of a reservoir in Zhejiang Province.Second,we constructed a remote sensing inversion model based on support vector machines.Using the OLI sensor data of Landsat8 satellites from 2013 to 2017,we modeled the chlorophyll a and total phosphorus water indicators.The experimental results show that compared with thelinear regression model,the support vector machine method can be used to obtain more accurate inversion results.Third,We have designed and implemented a river basin water quality management system that is oriented toward the river managers.The system is based on the Django framework,which realizes the management of water quality data,remote sensing data,and basin monitoring video data.The effectiveness of the water quality prediction algorithm and water quality inversion algorithm are verified in this system.