Modeling the Effects of Aridification on Hydrologic Fluxes and Reservoir Dynamics in the U.S. Southwest

Abstract Persistent droughts and anthropogenic warming have brought the southwestern United States—one of the world's most highly managed hydrologic regions—to the brink of a water supply crisis. Here, we examine the changes in hydrologic fluxes due to a shift toward aridity and the extent to which water storage in major reservoirs has offset these changes using a novel, high‐resolution (4 km) configuration of a state‐of‐the‐science land model, the Community Land Model. We leverage an emulation‐based parameter optimization approach to improve model hydrologic performance and reservoir parametrizations (generic and data‐driven) to simulate water storage and release. The hydrologic performance is notably improved across 121 basins in the Southwest, which translates into improved simulation of inflow, storage, and release across 133 reservoirs. The data‐driven approach outperforms the generic one, in simulating reservoir storage but not releases due to biases in inflows. Model simulations show a widespread increase in evapotranspiration (especially its soil evaporation component), driven primarily by increased absorption of solar radiation due to snowpack loss, which suppresses runoff. Increases in snowmelt mitigate runoff deficits during spring, which highlights the sensitivity of runoff response to the seasonal distribution of hydroclimatic supply and demand in the Southwest. Up to 40% of inflow deficits are met by surface water storage in large reservoirs (i.e., Mead and Powell). Simulated estimates of this supply‐demand imbalance are sensitive to biases in simulated inflows and releases. Caution is therefore warranted when using land models with reservoir parametrization for future drought risk and water scarcity assessments.