Revealing dominant controls in a continental-scale model of submarine groundwater discharge and seawater intrusion by utilizing intrinsic model variability

Groundwater is vital to sustain coastal freshwater consumption and agricultural activities around the globe. In the US, groundwater withdrawal has more than doubled from 1950 to 2015. In 2015, almost half of the coastal counties in the US relied on groundwater as their primary water source. Large volumes of groundwater withdrawal at the coast have caused groundwater level declines: almost half (44%) of the well water level observations made within 1 km of the coast are below sea level. We know that such reduction of groundwater reduces the amount of fresh submarine groundwater discharge (SGD) - fresh groundwater flowing into the ocean - and may trigger seawater intrusion (SWI), harming coastal ecosystems and deteriorating groundwater quality for domestic and agricultural use. Previous continental-scale and global models of SGD and SWI have simulated steady state conditions. To understand which factors drive these two fluxes and how coastal aquifers are impacted by sea level rise and changes in groundwater recharge, we have developed a MODFLOW-like modeling framework that can simulate transient density-driven groundwater fluxes on large scales (G³M-D). For our investigation, we focus on a simulation of North America. The model simulates SWI as an interface between potable and non-potable (i.e., too saline) groundwater. Established sensitivity-analysis methods that would allow pinpointing dominant controls inside a model often require hundreds to thousands of model simulations. Here, we utilize the intrinsic variability of the model to analyze drivers of coastal groundwater exchange. We show which factors drive the exchange fluxes between groundwater and ocean for different model domains. We also discuss whether the large-scale representation fits our perceptual model of coastal processes.