Abstract: The extraction of hydrological information from Digital Elevation Models(DEM) is an initial step in the application of spatially distributed hydrological models.Although,many automated methods have been proposed to extract the drainage structure from DEMs,the realistic river network and watershed boundary may not always be derived from conventional DEM processing methods.As a result,enhanced (or corrective) approaches have been proposed to improve the performance of existing single flow direction (D8) algorithms.However,those corrective measures may not always work over flat regions where the great uncertainty exists in DEM.Ambiguities in flow direction and parallel flow paths may arise if the surrounding cells have the same elevation as the central cell.In this paper,to downscale the DEM and extract drainage structures and watershed boundaries with improved accuracy a new method is proposed based on the Rural Infrastructure Digital Elevation Model is able to incorporate ancillary data and take into consideration of anthropogenic land cover features(i.e.roads and streams,buildings,some ponds and pipe systems) that influence drainage patterns.The method is applied to a watershed in the Shanghai drainage basin.The watershed can be characterized as a flat area undergoing a rapid urbanization process.The delineated maps and extracted stream networks are compared to those of using traditional methods.The result shows that the proposed method is able to produce more realistic results.The method is effective and easy to implement.