Comparison of multi-source DEM data in river network extraction in plain watershed

Abstract

[Objective] To investigate the suitability of different Digital Elevation Models(DEMs) for extracting river networks in plain areas, [Methods] ALOS DEM, ASTER GDEM, and SRTM DEM data sources were collected. Seven representative watersheds were selected for the study, and the accuracy of river network extraction was analyzed using the fitting difference calculation.[Results] The result showed that all DEM data exhibited significant errors in extracting river networks in flat terrain, lake areas, and urban areas. ALOS DEM demonstrated the highest accuracy in extracting digital river networks. The precision of the digital river network can be significantly improved by using the DEM correction function of Arc Hydro Tools. ASTER GDEM exhibited higher accuracy in hilly areas compared to SRTM DEM, while SRTM DEM showed higher accuracy in lake, farmland, and urban terrain conditions compared to ASTER GDEM. The optimal threshold for watershed area extraction using ALOS DEM was 50,000, for ASTER GDEM it was 20,000, and for SRTM DEM it was 5,000.[Conclusion] By selecting multiple DEM sources and analyzing the extraction of digital river networks from different perspectives, this study provides valuable insights for extracting digital river networks in plain areas.

Keywords

DEM data / multi-source / ArcGIS / river network extraction / DEM reconditioning / river network accuracy / influencing factors