A new treatment of depression for drainage network extraction based on DEM

Abstract

Depressions in landscapes function as buffers for water and sediment. A landscape with depressions has less runoff, less erosion and more sedimentation than that without depressions. Sinks in digital elevation models (DEMs) can be considered the real features that represent depressions in actual landscapes or spurious features that result from errors in DEM creation. In many hydrological and erosion models, all sinks are considered as spurious features and, as a result, these models do not deal with the sinks that represent real depressions. Consequently, the surface runoff and erosion are overestimated due to removing the depressions. Aiming at this problem, this paper presents a new method, which deal with the sinks that represent real depressions. The drainage network is extracted without changing the original DEM. The method includes four steps: detecting pits, detecting depressions, merging depressions, and extracting drainage network. Because the elevations of grid cells are not changed, the method can also avoid producing new flat areas, which are always produced by the conventional filling methods. The proposed method was applied to the Xihanshui River basin, the upper reach of the Jialingjiang River basin, China, to automatically extract the drainage network based on DEM. The extracted drainage network agrees well with the reality and can be used for further hydrologic analysis and erosion estimation.