中文摘要：

由于DEM数据不包括河流、湖泊（水库）及流域边界和堤坝等信息，因此常规仅依靠DEM提取水系的方法不能反映平坦区域及受人类影响强烈区域水系的真实特征。该文建立了融合DEM、河流、湖泊水库、流域边界和堤坝等多源数字信息的流域水系提取方法，对多源信息进行栅格化，采用所提出的高程一距离函数对DEM进行校正，使得提取的数字水系与实际水系精确拟合。将该方法应用于山区太湖西苕溪流域老石坎水库及其上游集水区和平原丘陵分布区的淮河史灌河蒋家集站、梅山、鲇鱼山水库区间流域，提取的河网结构与地图信息拟合较好。

英文摘要：

Distributed hydrological models need a detailed definition of watershed＇s internal drainage network structure. There are many watershed automatic extraction approaches which based on raster DEM and the eight flow direction method （D8）. However, the important difference of these approaches is the method of depression-filling and flow direction-assigning in plat ar- ea. In this paper, a new approach, using multiple map information such as river, lake,dam or watershed boundary network as in- put in addition to DEMs, has been developed. The map is digitized and stored by a raster matrix in this method. Two elevation- distance functions are used to adjust the DEM according to the rater map information. This new approach allows for an accurate fit between the map and the modeled drainage structure. The proposed approach has been tested on the Xitiaoxi River watershed in southern Tai Lake and Shiguanhe River watershed in the Huaihe River. The modeled watershed drainage structure showed a high level of coherence with the map. A comparison between the results obtained with the DEM-based approach and those obtained by the proposed approach clearly demonstrated an improvement over the modeled drainage structure. The proposed approach will benefit the distributed hydrological models which require data such as a flow direction matrix, river,lake and subwatersheds for drainage structure information.