基于BP人工神经网络的水体遥感测深方法研究

利用Landsat7ETM＋遥感图像反射率和实测水深值之间的相关性,建立了动量BP人工神经网络水深反演模型,并对长江口南港河段水深进行了反演.结果表明：具有较强非线性映射能力的动量BP神经网络模型能较好地反演出长江口南港河段的水深分布情况;由于受长江口水体高含沙量的影响,模型对小于5 m的水深值反演精度较高,而对大于10 m的水深值反演精度较低.

Study on remote sensing of water depth based on BP artificial neural networks

A momentum BP neural network model(MBPNNM) was constructed to retrieve the water depth information for the South Channel of the Yangtze River Estuary using the relationship between reflectance derived from Landsat ETM+ satellite data and water depth information.Results show that MBPNNM,which exhibits a strong capability of non-linear mapping,allows the water depth information in the study area to be retrieved at a relatively high accuracy level.Affected by the sediment concentration of water in the estuary,MBPNNM enables the retrieval of water depth of less than 5 meters accurately.However,the accuracy is not ideal for the water depth of more than 10 meters.