欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
利用Landsat-7ETM＋遥感影像反射率和实测水深值之间的相关性可以探测水深。该文介绍单波段、双波段比值和多波段3种线性回归模型以及动量BP人工神经网络水深反演模型。选择长江口北港河道上段作为研究区，利用上述模型，分两种情况进行水深反演：一是以河道全部历史样本建模；二是将河道按自然水深划分为浅水区和深水区分别建模。结果表明：神经网络模型预测精度高于线性回归模型；水深分区后线性回归和神经网络模型预测误差均有所减小。
英文摘要：
Based on the relationship between reflectance derived from Land.sat ETM ＋ satellite data and measured water depth data, water depth information can be retrieved. This paper simply introduces three kinds of linear regression models,which are single band, ratio by two bands and multi - band bathymetry models, then to discuss momentum BP neural network model（MBPNNM）. The North Channel of the Yangtze River Estuary is chosen as study object. The paper compares and analyzes the average relative error of different models under two conditions,one is modeling using all historical samples,another is to divide the whole channel into shallow water area and deep water area according to natural water depth. The results show that the effect of the MBPNNM is better than linear regression models under two conditions,and the average relative error reduces after sub-area. There are several shortages exist in the study,one is the difference of time between image data and actual water depth data, which increases the error;another is that the image has shortage in spatial resolution and spectral resolution, which decreases the precision because water depth information is not enough from spectrum. Under the constant advancement of remote sensing technology,mone and more high spatial and spectral resolution and high time frequency data are becoming available. So to retrieve water depth using remote sensing has great potential in prospect.
同期刊论文项目
同项目期刊论文
期刊信息
位置：