首页 | 本学科首页   官方微博 | 高级检索  
     检索      

赤潮期东海水体不同遥感分类算法应用分析
引用本文:申力,许惠平,吴萍.赤潮期东海水体不同遥感分类算法应用分析[J].海洋环境科学,2012,31(1):102-106.
作者姓名:申力  许惠平  吴萍
作者单位:同济大学海洋地质国家重点实验室,上海,200092
基金项目:科技部海洋地质国家重点实验室重点项目资助,上海市科学技术委员会重点项目资助
摘    要:对赤潮期长江口及邻近东海海域水体进行了三种不同遥感分类方法的水体类别提取。首先进行了最大似然监督法和ISODATA非监督法两种代表性的水体遥感分类,之后提出了两种基于水体光谱曲线特征的波段比值阈值分割模型,即C1=Lrs(645)/Lrs(469)和C2=Lrs(555)/Lrs(469)。对比发现,三种方法分类结果均揭示出了长江口及邻近东海海域近岸到远洋的水体空间变化规律,依次为悬浮泥沙水体、富营养化水体、赤潮水体、赤潮边界水体和远洋洁净水体。其中最大似然监督法对像元亮度值较为均一的悬沙水体和富营养化水体提取精度较高。而ISODATA分类结果误判较严重,可以作为辅助判别依据。本文提出的阈值分割模型1对悬浮泥沙水体更为有效,总体精度为91.34%,Kappa系数为0.88;模型2对赤潮水体更有效,总体精度为94.14%,Kappa系数为0.92,精度均高于前两种方法。

关 键 词:赤潮  水体分类  最大似然法  ISODATA法  光谱特征

Classification methods of different remote sense in the East China Sea during an algal bloom
SHEN Li , XU Hui-ping , WU Ping.Classification methods of different remote sense in the East China Sea during an algal bloom[J].Marine Environmental Science,2012,31(1):102-106.
Authors:SHEN Li  XU Hui-ping  WU Ping
Institution:(Key Lab of Marine Geology,Tongji University,Shanghai 200092,China)
Abstract:This paper focused on extracting water classification of the Yangtze River Estuary and the adjacent East China Sea during an algal bloom using three remote sensing approaches.Firstly,both Supervised Maximum likelihood and Unsupervised ISODATA methods were applied into the classification extraction.Then,based on the spectral characteristics of different waters a new method named Band-ratio Threshold Method consisting of Model 1 and Model 2 was developed,C1= Lrs(645)/Lrs(469) and C2= Lrs(555)/Lrs(469).The results derived from three methods showed the spatial distribution of waters in our study area from the coastal area to open ocean.They were the suspended sediment water,eutrophication water,and algal bloom water,boundary water of algal bloom and open ocean water.The maximum likelihood supervised method was more specific to classes of comparatively homogenous pixel values such as suspended sediment water and eutrophication water,while ISODATA unsupervised method could serve as a reference proof due to its severe misjudgments.Compared with the aforementioned two methods,the method put forward in this paper had a higher accuracy.The results of Model 1 was superior in identifying suspended sediment water with an overall accuracy of 91.34% and the Kappa coefficient of 0.88;Model 2 could better detect the spatial range of the algal bloom water with an overall accuracy of 94.14% and Kappa coefficient of 0.92.
Keywords:algal bloom  water classification  maximum likelihood  ISOADATA  spectral characteristic
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号