DEM数据作为基本地形信息的载体，被广泛应用于分布式水文模型中。但是由于各种人为的和自然的原因，DEM中普遍存在着许多大小不等的低洼地（包括洼地和平地），为水系的生成和汇流计算造成障碍。因此，DEM中低洼地的预处理是我们使用DEM进行水文模拟的前提，而如何快速有效地处理低洼地则成为数字水文学研究的热点。本文采用新的洼地填充算法（溢水法）填平洼地，并提出独创的“出口汇聚流向模式”为平地区域设置坡度，将平地中的水导出。该方法与传统填充法相比，简单明了，不需要计算水流方向，也不需要考虑复合洼地的嵌套关系。整套处理方法经编程实现，并由实例验证，表明该方法正确可靠。使用这套算法处理后的DEM中，任何一点的流向都是确定的，而整体地貌对比于处理前的改变接近最小。使用该方法处理得到的DEM能够满足分布式水文模型中洪水汇流计算的需要。

DEM is widely used in distributed hydrological models as the basic topographic information carrier. Because of various man-made and natural causes, many depressions of every size (including depressions and flat areas) are widespread in DEM. That causes an obstacle for river network generation and concentration calculation. So, preprocessing depressions in DEM is the premise of all the other work in flood simulation. And it has become a hot issue to deal with depressions problem quickly and efficiently in digital hydrology research. In this paper, a innovative depression filling algorithm（Overflow method）was proposed to process depressions, and an improved outlet-collect flow direction mode was put forward to assign tiny slopes for flat area. In this way, water was exported from flat areas fluently. Comparing with traditional algorithms, this method is simple and clear. There is no need to calculate the flow direction and it is easy to deal with depressions nesting structure. This method is proved correct and reliable after programming and example verification. The difference between the original DEM and the processed one is tiniest. DEM processed by this method could be used for flood concentration calculation in distributed hydrological models in a watershed.