基于神经网络的黄东海春季二类水体三要素浓度反演方法  

A Retrieval Method on the chlorophyll,total suspended matter,and gelbstoff concentrations of Case II waters in Yellow Sea and East China Sea Based On Artifi cial Neural Network

作　　者：仲京臣[1] 汪小勇[1] 陈清莲[1] 

机构地区：[1]国家海洋技术中心,天津300111

出　　处：《海洋技术》2005年第4期118-122,共5页Ocean Technology

摘　　要：介绍了一种基于人工神经网络的二类水体海域的三要素浓度反演方法。根据2003年春季黄东海试验中获得的高质量现场数据,建立了由现场测量遥感反射率分别反演三要素浓度的神经网络模型。反演的平均相对误差分别叶绿素32.5%,黄色物质8.9%,总悬浮物24.2%。同时分析了神经网络模型在水色反演模式应用中的稳定性。A retrieval method for the chlorophyll , total suspended matter , and gelbstoff concentrations of case Ⅱ waters in Yellow Sea and East China Sea Based On Artificial Neural Network （NN） is presented. In this paper, several neural Network models are established to retrieve the three major components concentrations from remote sensing reflectance （Rrs）. The data set was obtained from the Joint Ocean Color Experiments in Yellow Sea and East China Sea in spring 2003. Three types of NN models are proposed. They derive each components concentrations with an individual model. The averaged relative errors are 32. 5%for chlorophyll , 24. 2%for total suspended matter （TSM） and 8. 9% for gelbstoff , respectively. The NN models presented here are the preliminary and usable results for the area. The Stability of the Neural Network models was analyzed in this paper.

关 键 词：二类水体 三要素浓度 神经网络模型 

分 类 号：TP722.4[自动化与计算机技术—检测技术与自动化装置]