Skip to main content
U.S. flag

An official website of the United States government

Deriving Distributed Parameters to Improve Community Land Model Version 5 Hydrological Predictions

Presentation Date
Thursday, January 25, 2024 at 8:00am - Thursday, January 25, 2024 at 9:30am



Identifying spatially distributed parameters and quantifying their uncertainties remains a significant challenge in land surface models and macro-scale hydrological models. Previous large-domain modeling approaches have predominantly relied on spatially uniform parameters derived from limited field studies or expert judgments. This study addresses this challenge by deriving cell-to-cell spatially distributed ensemble hydrological parameters for the conterminous United States (CONUS) using the Community Land Model Version 5 (CLM5) at a resolution of 1/8-degree grid cells, totaling 50,629 grid cells. In contrast to previous parameter regionalization studies that specify a single cell-level parameter value based on a selected error metric, our approach exploits different streamflow characteristics to constrain behavioral ranges for parameters. This strategy captures parametric spatial heterogeneity, uncertainty, and provides users with the flexibility to select parameter sets based on their specific application interests. To evaluate the derived ensemble parameters, we utilized data from 464 selected CAMELS (Catchment Attributes and Meteorology for Large-Sample Studies) basins across the CONUS. We evaluated the derived ensemble parameters using three distinct error metrics, which focused on different flow regimes including low flow, high flow, and water balance. The results obtained from the CAMELS basins demonstrate significant improvements in CLM5 flow predictions. However, the magnitude of improvements exhibits considerable regional and error metric specific variations. For each error metric, we then constrained behavioral parameters for each grid cell across the CONUS. The novel CLM5 distributed parameterization developed in this study enhances CLM's predictive capabilities and, for the first time, enables the CONUS-scale characterization of parametric uncertainty in key hydrologic processes and signatures.

Funding Program Area(s)