Authors: Christina Hutchins, Adrienne Marshall – Colorado School of Mines
Title: Evaluating National Water Model Performance in Regulated River Systems
Presentation Type: Poster Presentation
Abstract: Reservoirs and water infrastructure fundamentally alter streamflow magnitude and timing in river systems by changing how, where, and when water flows through the system. Hydrologic models are valuable tools for understanding streamflow dynamics, but at continental scales they struggle to accurately represent human impacts on river systems. The National Water Model (NWM) is a continental-scale hydrologic model that provides short- and mid-range streamflow forecasting across most of the continental US, with a reanalysis run available over 1981-2023. The NWM currently includes 5,000 reservoirs in its framework but simulates them using only a simple fill-and-spill approach that fails to capture operational complexity.
While comprehensive testing of the NWM has been conducted, the effect of this extreme simplification of reservoir operations on model performance has not yet been evaluated. Here, we develop a statistical framework to systematically identify where streamflow regulation causes the largest errors in modeled streamflow relative to observations by analyzing hydrologic signatures that characterize streamflow timing and volume. Specifically, we anticipated that simulated flows downstream of reservoirs would occur earlier in the water year than observations, and would be too concentrated in a relatively narrow seasonal window. Aligned with these hypotheses, preliminary analysis reveals that the presence of reservoirs delays flow timing and attenuates peaks compared to NWM predictions. We plan to use this analysis to identify pilot locations based on these regulation-induced error patterns; these will be used to prioritize locations for developing enhanced reservoir operations modeling within the NextGen Framework.