基于多源遥感数据的水体信息提取研究

摘要

Satellite remote sensing technology has been widely used to extract water -body information .However , nowadays a single kind of remotely sensed data is usually used to extract water -body information , but the advantage of the combination of multi -source datum information is ignored .Consequently , the accuracy of water information extraction is often limited to the climate or spatial reso-lution.In this paper, a new method based on multi -scale and multi -platform remotely sensed data is explored to extract water -body information .The main research includes several steps as follows:first, decision tree classification algorithms based on relation-ships between spectrum are used to extract the accurate size of water -body on Landsat ETM +image due to its high spatial resolu-tion;second, the method of threshold value is used to extract water -body information on Radarsat SAR image crudely , then more ac-curate water-body information is obtained by eliminating the shadow influence combined with DEM ; finally, the flood extent is ob-tained by using differential technique of the water -body information extracted from the Landsat ETM +image before flood disaster and Radarsat SAR image during flood disaster .The new method explored in this paper can provide information support for flood disaster monitoring and evaluation .%卫星遥感技术已被广泛应用到水体信息提取中，但目前基于遥感技术的水体信息提取多采用单一的遥感数据源，而没有充分利用多源数据的信息复合优势，因此，提取结果经常受天气气候或空间分辨率限制。本文研究了不同尺度、不同平台的多种遥感数据源的水体信息提取方法。首先，基于波谱间关系决策树分类算法对Landsat ETM＋图像进行水体提取，利用其分辨率优势较准确地提取出水体范围；其次，在Radarsat SAR图像上利用阈值法粗提取水体信息后，结合DEM剔除阴影得到水体信息；最后，利用灾前Landsat ETM＋图像水体信息提取结果和灾中Radarsat SAR图像水体信息提取结果，进行差值处理，得到洪水淹没范围。研究结果可以为洪水灾害监测与评估提供信息依据。