中文摘要：

针对地势平坦且空间范围较大的流域水系自动提取任务中,采用常规的洼地填充算法难以提取出完整水系,形成大量＂断头河＂的问题,提出了一种能够导出整个数字高程模型（DEM）所有像元被淹没的次序表的方法,由该次序表构成的矩阵代替原有DEM来实现流向的计算,进而提取出累积流向及分级河道。经用覆盖中心流域的高精度DEM以及覆盖黄河和长江等大型流域的DEM来测试,结果表明：用其他算法提取结果均发生了＂断头河＂错误,而该方法则能提取出完整的水系。这种方法能正确实施洼地填充和像元填平处理,可以适应于任意规模和精度的DEM填洼问题,具有较强的鲁棒性,克服了以往算法难以处理大河流域DEM洼地填充的不足。

英文摘要：

Currently there are a variety of commonly used algorithms for filling depressions and removing planar cells in extracting digital river networks from the digital elevation models（DEM）data.For the extraction for flat and large river basins,conventional depression-filling algorithm is still difficult to ensure the complete extraction of river networks.This paper proposes a method based on an existing algorithm,which takes the idea of gradual inundating the cells and using apriority queue.By modifying this algorithm,the inundating orders of all cells are exported to a matrix.Then the flow directions and accumulated flow directions are calculated from this matrix instead of the original DEM.For testing the algorithm,small scale and middle scale DEM with high resolution,the comparatively low-resolution DEM covering large river basins such as Yellow River basin and Yangtze River basin,the results show that the proposed method can extract complete river networks,but other algorithms have produced intermittent and erroneous river networks for large river basins.This indicates the proposed method is very robust,and it overcomes the shortcomings of the existing algorithms for depression filling of DEM data on large river basins.