Watershed Merging: A Simple and Effective Algorithm for Channel Network Identification and Extraction

Abstract

Channel network identification is an important practice in not only hydrologic analysis but also hydraulic computation. In this paper, a new algorithm, watershed merging, is proposed to automatically identify and extract channel networks. In the water-merging algorithm, based on the fact that the sink cell of a dendritic watershed is either a depression cell or a flat cell, a macroscale approach is proposed to treat the depression and flat areas (DAFA) and determine the flow direction within those areas, where the conventional D8 slope calculation fails. The separated neighboring watersheds are merged together using information of neighboring watersheds instead of the D8 cells. This progressive merging process starts from small neighboring watershed to larger ones. The example and applications demonstrated that the proposed watershed-merging algorithm is effective in resolving the DAFA problems and identifying channel networks.