Assessment of horizontal shifting errors in open-source DEMs and their impact on 1D hydraulic modeling of meandering river channel

Study Region

The Ganges River, Bangladesh

Study Focus

Horizontal shifting error in open-access GDEMs has received very little attention for detecting and assessing its impact on hydraulic modeling. This study uses three integrated techniques to detect horizontal shifting errors in ASTER and SRTM DEMs by comparing DEM river centerlines and real-time river centerlines, measuring river channel dynamics, and analyzing water level elevation profiles. Four strategies were applied to assess the impacts: evaluating DEM-generated river channels, creating a 1D hydraulic model for inundation mapping, assessing inundation accuracy, and visualizing the horizontal shifts.

New Hydrological Insights for the Region

The results reveal several significant issues. Firstly, both DEMs exhibited notable horizontal shifting errors, with observed maximum, average, and minimum errors of 3200 m, 1800 m, and 680 m, respectively, though the intensity of these errors varied spatiotemporally. Secondly, the maximum riverbed effectiveness (≥73.38 %) was observed in 2000, but subsequently decreased to ≤ 41.25 %, ≤ 47.94 %, ≤ 46.75 %, and ≤ 54.08 % in 2006, 2010, 2015, and 2020, respectively. Thirdly, the horizontal shifting error was found to have severe implications for 1D inundation mapping accuracy. Lastly, strong negative correlation was observed between river channel dynamics and DEM river channel effectiveness (r ≥ -0.963). The results indicate that both DEMs have horizontal shifting errors, which are expected to produce severe inaccurate results in hydraulic and flood modeling, especially in areas with dynamic topography.