利用Landsat7 ETM+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,对缺失的1995年的水下地形资料进行了插补,并利用插补后的1973—2003年的水下地形资料计算了南港一南槽河段的泥沙冲淤量,分析了南港-南槽河段等深线和横断面的变化,结果表明:①BP人工神经网络水深反演模型能反演出研究区的水深分布情况,特别是对水深浅于-10m区域,模型反演效果较好;②南港-南槽河段在1973—2003年间整体为淤积,泥沙淤积量为1.39亿m^3;③河道南岸-2m等深线已基本稳定;由于发生了江亚南沙切滩和南槽上段改道,河道南岸-5m等深线变化较大,加之江亚南沙被切割的部分并入九段沙上沙头。因而河道北岸-2m和-5m等深线均上移约9km。
A BP neural network model (BPNNM) was constructed to retrieve the water depth information of 1995 for the South Channel- the South passage of the Yangtze River Estuary by using the relationship between reflectance derived from Landsat7 ETM + satellite data and water depth information, The amount of deposition and erosion was calculated and the change of isobaths and cross section were anatyzed by using the under water topographical data from 1973 to 2003. Results show that the BPNNM allows the water depth information in the study area to be retrieved at a relatively high accuracy level for the water depth of less than 10 meters ; from 1973 to 2003, the South Channel -the South passage is generally characterized by deposition, and the total deposition volume is 1.39 × 10^8m^3 ; the -2m isobaths of south bank is steady, but the -5m isobaths of the south bank change greatly due to the Jiangyanansha being cut up and the upside of North passage Jiangyanansha being cut and merged into the Jiuduansha made the ahead about 9km. changing its course, in addition, the part of 2m and - 5m isobaths of north bank all move