Development of High-Resolution Regolith Thickness and Topographic Datasets for Regional/Global Land-Surface Modeling

Abstract:

Earth’s terrestrial subsurface environment can be divided into a relatively porous layer of unconsolidated material (herein termed regolith) above consolidated saprock/bedrock. Variations in the thickness of this layer control water-storage capacity and runoff potential. Currently, Earth System Models assume a constant thickness (e.g. 2-5 m) for this layer globally despite the fact that regolith is absent in some regions and tens of meters thick in other regions. We report on the development of a high-resolution global raster dataset of regolith thickness in uplands (eroding landscapes of relatively thin (<5 m) regolith over bedrock) and lowlands (depositional landscapes with regolith often more than 10 m thick). The data products are 1 km/pixel resolution but they utilize topographic data at 30 m/pixel resolution using the ASTER V2 GDEM as input. Uplands and lowlands are distinguished at the 30 m/pixel scale using a valley network extraction algorithm and geologic age criteria. Regolith thickness on uplands is predicted using geomorphic/pedogenic models dependent on topography, rock type, and climate. Regolith thickness on lowlands is a function of topography (with narrower valleys having thinner regolith) and geologic map unit age. The predictive models are calibrated using the CONUS-Soil database (for uplands) and available data on the thickness of unconsolidated sediments from groundwater wells in U.S. states where digital well data are available (for lowlands). Data from the European Soils Database (for uplands) and a subset of U.S. well data (for lowlands) are used for validation. As a byproduct of our work, we developed a global drainage network/wetland map and other useful hydrologic data at 30 m/pixel resolution. These datasets will be made available to the research community and should prove useful in improving the accuracy of ecohydrological processes in regional and global land-surface models.