SHIFT: a spatial-heterogeneity improvement in DEM-based mapping of global geomorphic floodplains

Abstract

Abstract. Floodplains are a vital part of the global riverine system. Among all the global floodplain delineation strategies empowered by remote sensing, digital elevation model (DEM)-based delineation is considered to be computationally efficient with relatively low uncertainties, but the parsimonious model struggles with incorporating the basin-level spatial heterogeneity of the hydrological and geomorphic influences into the map. In this study, we propose a globally applicable thresholding scheme for DEM-based floodplain delineation to improve the representation of spatial heterogeneity. Specifically, we develop a stepwise approach to estimate the floodplain hydraulic geometry (FHG) scaling parameters for river basins worldwide at the scale of the level-3 HydroBASINS to best respect the scaling law while approximating the spatial extent of two publicly available global flood maps derived from hydrodynamic modeling. The estimated FHG exponent exhibits a significant positive relationship with the basins' hydroclimatic conditions, particularly in 33 of the world's major river basins, indicating the ability of the approach to capture fingerprints from heterogeneous hydrological and geomorphic influences. Based on the spatially varying FHG parameters, a ∼ 90 m resolution global floodplain map named the Spatial Heterogeneity Improved Floodplain by Terrain analysis (SHIFT) is delineated, which takes the hydrologically corrected MERIT Hydro dataset as the DEM inputs and the height above nearest drainage (HAND) as the terrain attribute. Our results demonstrate that SHIFT validates better with reference maps than both hydrodynamic-modeling- and DEM-based approaches with universal parameters. The improved delineation mainly includes better differentiation between main streams and tributaries in major basins and a more comprehensive representation of stream networks in aggregated river basins. SHIFT estimates the global floodplain area to be 9.91×106 km2, representing 6.6 % of the world's total land area. SHIFT data layers are available at two spatial resolutions (90 m and 1 km), along with the updated parameters, at https://doi.org/10.5281/zenodo.11835133 (Zheng et al., 2024). We anticipate that SHIFT will be used to support applications requiring boundary delineations of the global geomorphic floodplains.