Evaluating time-lagged relationships between groundwater storage and river discharge using GRACE-based data: insights from the Potomac Basin

Abstract This study evaluates the utility of a recently available GRACE-based groundwater drought index (GDI) in supporting regional water supply management, with application to the Potomac River Basin, in the U.S. Middle Atlantic region. As the primary drinking water source for the Washington Metropolitan Area (WMA), effective management of the Potomac River’s resources is critical, especially in the context of climate change, with the expected increase in severity and frequency of extreme events. Our analysis integrates 22 years of data, including GRACE-based groundwater storage (GWS) index estimates, river discharge (Q) measurements, and meteorological records, to investigate trends and predictive relationships between past GWS, as determined by the GRACE-based drought index, and streamflow. Seasonal Mann-Kendall trend analyses consistently identified severe declining trends in groundwater storage (GWS), as well as moderate declines in minimum streamflow and well water levels over the past 22 years. Granger Causality (GC) tests revealed significant time lags of 49 weeks to 22 months at weekly and monthly scales, respectively depending on a region’s hydrogeomorphic characteristics. Vector Autoregressive (VAR) Models and Forecast Error Variance Decomposition (FEVD) highlighted the variable contributions of precipitation and temperature to the GWS-Q relationship, revealing a strong autoregressive component of Q, but also reveal that GWS plays an important role, and this role increases with time. These findings underscore the interconnectedness of groundwater and surface water systems and the importance of integrated predictive models to enhance water management strategies. Incorporating GRACE-based seasonal groundwater forecasts into drought preparedness tools could bolster efforts to mitigate regional climate change impacts and improve the resilience of water resources in the Potomac River Basin. While practical use of native GRACE data has been challenging for local, small-scale applications, this study demonstrates the utility of the GRACE-based GDI in forecasting low flows and informing regional water resource management decisions during droughts.