"Decision-Relevant" Multi-Model Hydrologic Evaluation in the Delaware River Basin for Assessments of Water System Vulnerability under Drought

The Delaware River Basin (DRB) provides drinking water to over 15 million people and is managed through a complex, multi-agency governance structure that strives to meet human and ecosystem needs while remaining adaptive to changing conditions. Internal basin water demands and out-of-basin transfers must be balanced with water temperature requirements for habitat and minimum flows to prevent upstream incursion of saltwater from the estuary that threatens infrastructure. The DRB weathered a major drought in the 1960s and water resource managers are concerned with meeting water demands during a future drought of comparable magnitude. To support these needs, we are building an integrated modeling system that links models of socio-economics, water infrastructure, atmospheric dynamics, inland hydrology, coastal dynamics, and water quality to "stress test" the DRB water system under drought conditions combined with future climate warming, land cover change, and sea level rise. As part of the first phase of this effort, we evaluated baseline performance of a selection of models with respect to major water budget components as well as the fluxes and states that are directly tied to legal operational mandates (e.g., minimum flow requirements). Specifically, we compared simulations from different meteorological drivers (Daymet, Livneh et al. 2015, U.S. National Oceanic and Atmospheric Administration's Analysis of Record for Calibration, Weather Research and Forecasting (WRF) Model 4-km "CONUS404") and hydrologic models (WRF-Hydro, Precipitation Runoff Modeling System (PRMS), Water Evaluation and Planning System (WEAP), MODFLOW). We compared water partitioning (evapotranspiration, recharge, runoff), streamflow, baseflow, and reservoir levels where possible. In addition to quantifying model skill, we also characterized the importance of process complexity (e.g., reservoir management, aquifer dynamics) when estimating decision-relevant metrics (e.g., number of days that the minimum flow requirement is violated). The model comparison results provide a foundation for improving models for future change scenarios as well as understanding impacts of model uncertainty on assessing water system vulnerabilities.