Weihe River Water Quality Based On Neural Network Quantitative Remote Sensing Research

With more and more serioous problem on water pollution,in particular,island water quality monitoring plays an essential role in developments of society and economic.It is significant to monitor the water quality and get to know the trend of the water quality change for water resource manafement and water environment protection.The traditionnal approachs of monitoring water quality are by collecting water samples and analysing these samples,those approaches have to get a large number of water samples,and they have a local significance.The shortcomings of them are not only heavy workload,but also high cost and long cycle.Water monitoring based on remote sensing has the advantages of large scale,speediness,cost-effective and capable of making long-term synamic water quality monitoring.It has a enormous potential on inland water monitorin.In recent years,in home and abroad,scholars have applied it to inland water monitoring,but it is very little research about Wei River.So the paper is focused on Wei Rever Shaanxi Province by using of SPOT-5 data.The main works in this thesis are as follows:In the paper,the Wei River in Shaanxi Province for the study area uses remote sensing image SPOT-5.Under the existing water quality standards and combinating with Wei River quality published Shaanxi Province of environment bulletin.Wei River pollution is mainly to organic pollution,and referencing on stations of the monitoring data.The paper selects CODmn, NH₃-N,DO,COD to inversion.The water of Wei River is typical inland water with complicated spectral characteristic.It is difficult to presnet the relationship between band reflectance and water paramenter only using regression methods.Artificial Neural Network is suitable to simulate such complicated relationships.We have given some application of the Artificial Neural Networks in water resource quzlity.The significance of its application in water environmental system has been summarized.In the paper,BP,RBF,GRNN and improved GRNN networks.have heen introduced and Networks have been used to set up inversion model.Then,different remote sensing inversion models have been built.Finally,the inversion results of these models have been compared with the results by regression mothods.The result of experiment shows that aitificial neural network used for inversing water quality parameters was effective,because artificial neural network is nonlinear,its retrieving results were much better than linear regression model.