摘要
以渭河陕西段水域为研究对象,在获取了实地监测数据和SPOT5遥感影像的基础上,对遥感数据进行预处理,建立了BP神经网络水质反演模型和RBF神经网络水质反演模型。并对水质参数CODcr、NH3-N、DO、CODmn进行反演。研究结果表明,利用神经网络模型反演水质参数是可行的,由于是非线性模型,其反演结果明显好于线性回归模型的结果。
This paper is focused on Weihe River in Shaanxi Province. With conventional monitoring data of Weihe River in Shaanxi and SPOT-5 data in the same period obtained, the remote sensing data are prepro- cessed to set up a BP neural network model and RBF neural network model to reverse concentrations of CODcr,NH3-N,DO,CODmm. The result of experiment shows that artificial neural network used for inver- sing water quality parameters was effective, because artificial neural network is nonlinear, its retrieving re- sults were much better than linear regression model.
出处
《遥感技术与应用》
CSCD
北大核心
2009年第1期63-67,共5页
Remote Sensing Technology and Application
基金
国家自然科学基金项目(40671133)
关键词
遥感影像
人工神经网络
水质反演
渭河
Remote sensing image
ANN
Water quality retrieving
Weihe River
