BP神经网络在城市湖泊水深反演中的应用  

Application of BP Neural Network in Inversion of Urban Lake Water Depth

作　　者：任中杰 曹久立[1] 万永智 REN Zhongjie;CAO Jiuli;WAN Yongzhi(Xuzhou Branch of Jiangsu Provincial Hydrology and Water Resources Investigation Bureau,Xuzhou 221000,China)

机构地区：[1]江苏省水文水资源勘测局徐州分局,江苏徐州221000

出　　处：《水资源开发与管理》2024年第7期75-78,共4页Water Resources Development and Management

摘　　要：水深遥感反演广泛应用于湖泊和近海水体的水深监测,其中统计回归法是较为常用的一种方法。但水深数据与光谱信息并非单一的线性关系,因此限制了模型的反演精度。本文基于LANDSAT9影像,利用SFIM融合算法提升了影像空间分辨率后,进而建立BP神经网络模型对水深进行反演,并与多元线性回归模型进行对比。结果表明,BP神经网络在城市湖泊水深反演过程中,精度显著优于多元线性回归模型,在一定程度上能应用于实际。Water depth remote sensing inversion is widely used in the monitoring of lake and nearshore water bodies,with statistical regression being a common method.However,the relationship between water depth data and spectral information is not a simple linear one,which limits the accuracy of the model inversion.In this paper,based on LANDSAT 9 imagery,the spatial resolution of the images is enhanced using the SFIM fusion algorithm,and then a BP neural network model is established to invert water depth,which is compared with a multiple linear regression model.The results show that the accuracy of the BP neural network in the inversion of urban lake water depth is significantly better than that of the multiple linear regression model and can be applied to practical situations to some extent.

关 键 词：BP神经网络 LANDSAT9 SFIM 水深遥感反演 

分 类 号：P332[天文地球—水文科学]