Assistant Professor University of Texas at Arlington
Estimating surface water extents accurately is key to improved understanding of their role in water resource research and management, especially in a changing climate. However, overlooking the structural and spatial inconsistencies existing between available datasets capturing surface waters, can make future water security and sustainability uncertain. To assess the degree of such uncertainties in the continental United States (CONUS), we performed a novel CONUS-scale analysis of six surface water or potential surface water datasets, including the (1) HydroLAKES (Messager et al., 2016), (2) LAGOS (Cheruvelil et al., 2021), (3) 10-meter European Space Agency Sentinel satellite-based WorldCover (ESA., 2022), (4) 30-meter Joint Research Centre Landsat satellite-based Global Surface Water Occurrence (Pekel et al., 2016), (5) 30-meter National Elevation Dataset-based surface depressions (Rajib et al., 2020, Wu et al., 2019), and (6) 90-meter OpenStreetMap Surface Water Layers (Yamazaki et al., 2019). Our results reveal the source of these uncertainties to be differences in (1) data origins and methods used for water body delineation, (2) data formats (vector vs. raster), (3) spatial and temporal resolutions, and (4) data-specific definition and nomenclature of surface waters. These differences result in dissimilar total areal coverage, number, and size distribution of surface water bodies across these datasets. For example, 10-m ESA detects 132,000 km2 of total surface water extents across CONUS, whereas 90-m OSM detects only 97,000 km2 reflecting methodological differences in the original data sources; or Lake Mead, a hydropower reservoir formed by the Hoover Dam on the Colorado river, is 347 km2 in the gridded ESA dataset but is 40% larger with a considerably different shape in the HydroLAKES polygons. With these findings, our study creates the first baseline to quantify the uncertainties in surface water availability estimates of the U.S. and therefore guides the selection of a suitable dataset for management and policy decisions.
