中文摘要：

水系自动提取是保证基础地理数据现势性的关键,而获得最佳集水面积阈值是提高水系提取精度的前提条件。集水面积阈值一般依靠定性分析的方法获取,采用均值变点分析法定量确定水系的集水面积阈值,可以满足基础地理数据精确性的要求。以云南省龙川江流域为研究区域,基于资源三号测绘卫星ZY-3 DSM和SRTM DEM两种数据,利用GIS工具分别提取该流域水系,再由均值变点分析法确定最佳集水面积阈值,并比较分析这两种水系的提取精度。研究结果表明：1随着集水面积阈值的增加,提取的水系密度减小且趋缓。ZY-3 DSM和SRTM DEM的最佳集水面积阈值分别是6.0 km~2和2.75 km~2;2在最佳集水面积阈值下,ZY-3 DSM提取的河流数比SRTM DEM多20.6倍,河流等级多2级,各级支流数为SRTM DEM的4.7~9.5倍不等,水系密度达到SRTM DEM的4.5倍;3与1：250 000水系数据相比,ZY-3 DSM和SRTM提取的河流点位平均相差395.23 m,RMSE为391.83 m。由此可见,利用均值变点分析法基于ZY-3 DSM提取的水系精度更高,完全满足测绘地理数据更新的现势性、准确性要求。

英文摘要：

Automatically extracting river system is the key to ensure the data currency of the basic geographic data. The drainage area optimal threshold is a prerequisite for improving the extraction accuracy of river system. The drainage area threshold is obtained by qualitative analysis method usually. In order to meet the requirements of the accuracy of the basic geographic data, the change point method is used, and compared the accuracy of automatically extracted river system based on ZY-3 DSM and SRTM DEM in this paper. The Long Chuan River of Yunnan is selected as a study area, using GIS tool to extract the river system and obtain the drainage area optimal threshold by the change point method. The results show：（1） with the increase of the drainage area threshold, the extracted water density is decreased and the decreasing trend becomes slowly. And the drainage area optimal threshold of ZY-3 DSM is 6.0 km2 and SRTM DEM is and 2.75 km2;（2） By using the drainage area optimal threshold, the extraction river number from ZY-3 DSM is 21 times than the SRTM DEM, channel classification levels are more than 2, drainage density reached 4.5 times over the SRTM DEM;（3）Taking the 1：250 000 river network as the reference, the difference between extracted river points from ZY-3 DSM and SRTM DEM is 395.23 m, and RMSE is 391.83 m.Thus, the extracted river system from ZY-3 DSM has higher accuracy than from SRTM DEM, and it completely satisfied the requirements of surveying and mapping geographic data currency and accuracy.