Threshold Selection of River Network Extraction Based on Different DEM Scales Using ATRIC Algorithm

Abstract

The extraction precision of digital network in watershed depends on drainage area threshold. Due to the multi-scale effect of DEM, different scales of DEM would provide different terrain information. Taking Qingjiang watershed as the subject and the DEM under 30m∗30m and 90m∗90m resolution as the foundation, this research extracts the digital river network and studies the optimal drainage area threshold from DEM with different resolutions through ATRIC algorithm. It's concluded that under the same resolution, the river network becomes sparser with the increase of threshold; the optimal drainage area threshold under higher resolution is larger than that under lower one. Compared with DEM data for 90m resolution, the river network extracted from DEM for 30m resolution enjoys a better goodness of fit with the real river network, with an overlapping rate of 68.8%. ATRIC algorithm is more suitable for DEM data under high resolution.