Identifying directions for improving water prediction by evaluating multiple national-scale hydrologic models

Skillful hydrologic models are needed to address a broad range of water management issues, such as doughts and water availability. To address water issues at the national scale, research teams at the National Center for Atmospheric Research (NCAR) and the U.S. Geological Survey (USGS) have developed hydrologic models covering the contiguous United States (CONUS). These physically-based models, the National Water Model (NWM version 2.1) application developed by NCAR and the U.S. National Oceanic and Atmospheric Administration and the National Hydrologic Model application of the Precipitation-Runoff Modeling System (NHM version 1.0) developed by USGS, address comparable breadth and scope of water resource issues, but are based on different modeling software and approaches. To identify directions for developing the next generation of hydrologic models an assessment of current model skill to represent daily streamflow and select water budget components in the CONUS is being completed. Model skill is assessed by computing error statistics in a standardized evaluation and benchmarking approach. The models are evaluated at 5,520 streamgage locations which are nearly all part of the GAGESII subset of the USGS streamgage network, a quality-assured database designed to support such evaluations. Selected streamflow metrics at streamgage locations are associated with attributes of the upstream drainage areas to identify climatic, physiographic, biologic, and anthropogenic factors that contribute to model skill. Simulated water budget components (e.g., evapotranspiration, soil moisture, snowpack) are compared for a range of watersheds that represent various hydroclimatic regimes. Results indicate regional variations in model skill across the CONUS, with neither of the model applications performing consistently better for all regions. The preliminary results are an initial step of an evidence-based process for guiding model-enhancement activities and improving water budget prediction at the national scale. This effort is part of a larger USGS-NCAR collaboration entitled: A Community Testbed Project for High-Resolution Hydroclimate Science, Simulation, and Application.