The Science

Scientists and engineers use hydrology models to simulate water flow across and beneath the Earth’s surface. Hydrology models have traditionally used simplified representations of the landscape and must be calibrated to match observations made under current conditions, which creates uncertainty when the models are used in new conditions. This study found that a new model version that represents hydrologic processes in greater detail can match streamflow data without significant calibration of the model. Avoiding calibration improves overall confidence in a hydrology model as a tool for understanding how climate and land use change will affect water supply.

The Impact

Advancing understanding of how watersheds function is becoming increasingly important as warming climate conditions affect water resources. This study evaluated the performance of a high-resolution, process-based hydrology model in reproducing streamflow and evapotranspiration data from seven diverse catchments. Model performance was good in five catchments using only community data products to define model inputs. In the other two catchments, good model performance was realized after correcting the data products to be consistent with known geology. This study shows that high-resolution process-based hydrology models supported by community data products can improve understanding of water supply threats.

Summary

A team of researchers from Oak Ridge National Laboratory evaluated the performance of a high-resolution surface/subsurface hydrology model, the Advanced Terrestrial Simulator (ATS), using streamflow and evapotranspiration data from seven diverse catchments. Community data products were used to define model inputs without calibration. ATS performance for evapotranspiration was good in all seven catchments using default data products. ATS with default data products performed reasonably well on streamflow for five catchments. Model performance was significantly improved in the other two catchments by adding local information on subsurface properties below the soil layer. ATS performance was also compared to a semi-distributed model called the Sacramento soil moisture accounting (SAC-SMA) model, which was calibrated for each catchment. Uncalibrated ATS performance was comparable to the calibrated SAC-SMA model in terms of streamflow, but overall was found to be better than the SAC-SMA model at reproducing evapotranspiration. Good performance of ATS without catchment-specific calibration provides new confidence in spatially resolved, process-based models as tools for advancing understanding of the function of watersheds in a changing environment. The community data products needed to support these types of models are widely available, but subsurface properties need to be independently verified.

Principal Investigator

Scott Painter
Oak Ridge National Laboratory
paintersl@ornl.gov

Program Manager

Jay Hnilo
U.S. Department of Energy, Biological and Environmental Research (SC-33)
Data Management
Justin.Hnilo@science.doe.gov

Funding

The work was supported by the ExaSheds project within the U.S. Department of Energy’s (DOE) Biological and Environmental Research (BER) Program. An award for computing time was provided by the Advanced Scientific Computing Research (ASCR) Leadership Computing Challenge allocation program. This research used resources from the National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science user facility.

References