基于SPOT5的红树林遥感分类

Remote sensing classification of mangrove based on SPOT5 image

以海南岛东寨港国家级红树林自然保护区为例,实地测量了红树林等地物光谱,获取了较为准确的红树林的光谱特征;在采用最大似然分类法提纯红树林信息基础上,结合不同红树种类的光谱特征及植被指数差异,选定可分性阈值,建立决策树分类规则,对SPOT5图像进行种间分类,并检验其分类精度。结果显示,各红树种类的使用精度都达到80%以上,且总体精度达到90%以上,为红树林生态系统业务化遥感监测奠定基础。

Taking the Dongzhaigang National Mangrove Reserve in Hainan Island as example,we executed field measurements to obtain mangrove spectral data.On the basis of maximum likelihood classification purification mangrove information,combining with different kinds of mangrove spectral features and vegetation index difference,selected separability threshold,we established decision-making tree classification rule of SPOT5 images based on images between classifications,and tested its classification accuracy.Classification result shows that accuracy for each mangrove type reached 80% and the total accuracy added up to 90%.This research offers a foundation for operational remote sensing monitoring of mangrove ecosystem.