Regional-scale groundwater quality degradation from nonpoint source pollution threatens the long-term sustainability of major alluvial aquifer-aquitard systems worldwide. Upscaled models can efficient represent nonpoint source transport, but fail to accurately characterize non-Fickian (anomalous) transport caused by mean flow direction transience. In this study, we demonstrate that hydrogeologic factors explain this failure. Specifically, vertical anisotropy in K and seasonal pumping and recharge in typical alluvial aquifer systems can fundamentally change hydraulic gradients and shift the mean flow direction between mostly horizontal and mostly vertical flow. Detailed 3D flow and transport simulations in a heterogeneous alluvial aquifer under varying mean flow directions indicate that alterations to hydraulic gradients which control the mean flow direction can lead to increasingly non-Fickian transport. Under mostly horizontal flow, diffusion and slow advection dominant low-K facies slow mass transfer rates from low-K material, and preferential flow along connected high-K networks causes increased spatial spreading along the mean flow direction. Conversely, predominantly vertical flow caused by spatially distributed pumping and recharge shifts mass transfer processes in low-K material from diffusion and slow advection dominant to advection dominant, which results in vertically oriented particle trajectories that compactly migrate through high- and low-K facies alike, leading to increasingly Fickian transport. Thus, mean flow direction transience driven by vertical anisotropy in K and seasonal pumping and recharge can create oscillating transport patterns, ranging from persistently non-Fickian to more Fickian. Results illustrate the hydrogeologic factors that explain why upscaled transport models fail to capture non-Fickian effects resulting from mean flow direction transience.

Global food systems rely on irrigated agriculture, and most of these systems in turn depend on fresh sources of groundwater. In this study, we demonstrate that groundwater development, even without overdraft, can transform a fresh, open basin into an evaporation dominated, closed-basin system, such that most of the groundwater, rather than exiting via stream baseflow and lateral subsurface flow, exits predominantly by evapotranspiration from irrigated lands. In these newly closed hydrologic basins, just as in other closed basins, groundwater salinization is inevitable because dissolved solids cannot escape, and the basin is effectively converted into a salt sink. We first provide a conceptual model of this process, called “nthropogenic asin losure and groundwater inization” (ABCSAL). We then examine the temporal dynamics of ABCSAL using the Tulare Lake Basin, California, as a case study for a large irrigated agricultural region with Mediterranean climate, overlying an unconsolidated sedimentary aquifer system. Even with modern water management practices that arrest historic overdraft, results indicate that shallow aquifers (36 m deep) exceed maximum contaminant levels for total dissolved solids on decadal timescales. Intermediate (132 m) and deep aquifers (187 m), essential for drinking water and irrigated crops, are impacted within two to three centuries. Hence, ABCSAL resulting from groundwater development in agricultural regions worldwide constitutes a largely unrecognized constraint on groundwater sustainable yield on similar timescales to aquifer depletion, and poses a serious challenge to global groundwater quality sustainability, even where water levels are stable.

Water security hinges on water storage. Although the public and water resources planners habitually look to surface reservoirs for storage solutions, by far the largest ‘space’ to store water is underground. The very nature of freshwater distribution on Earth foreshadows future water storage solutions, as 97% of all circulating freshwater globally is in groundwater. Similarly, although 140 surface reservoirs in California can store 52 km3(42 MAF), in the Central Valley aquifer system there is room for another ~170 km3(~140 MAF) owing to past depletion. Despite the state’s Mediterranean climate in which nearly all of the precipitation occurs between November and March when demand is lowest, historically massive snow storage and spring-summer snow melt synchronized well with surface reservoir replenishment during April-July. This system built around snow storage as a means of mitigating winter flood threats and delaying runoff until the beginning of the peak demand season is clearly demonstrating significant vulnerabilities to climate change and drought. Climate warming has already produced decades of declining snowmelt runoff, making surface reservoir storage more difficult. Moreover, as demonstrated during the 2012-16 drought, in the face of droughts longer than a few years, the surface storage offers inadequate long-term water security. This fact, the fact that California during pre-development times of the last millennium experienced far longer droughts, and ongoing climate change clearly indicate the need for a different strategy that more fully leverages both surface and subsurface storage. Kocis and Dahlke (2017) show that increasing winter runoff during wet and normal years provide enough high-magnitude flows to support a strategy of diverting flood flows for groundwater storage. This “flood-MAR” (managed aquifer recharge) approach will require a massive change in winter water and land management that exploits recharge opportunities on irrigated farm lands and in areas with suitable soils and subsurface geology. A case study in the American-Cosumnes Rivers portion of the Central Valley shows how total system water storage can be increased dramatically through diversion of high-magnitude flows and reoperation of both the surface and subsurface reservoirs including economic incentives.

Airborne electromagnetic (AEM) data can be inverted to recover models of the electrical resistivity of the subsurface; these, in turn, can be transformed to obtain models of sediment type. AEM data were acquired in Butte and Glenn Counties, California, U.S.A. to improve the understanding of the aquifer system. Around 800 line-kilometers of high-quality data were acquired, imaging to a depth of ~300 m. We developed a workflow designed to obtain, from the AEM data, information about the large-scale structure and heterogeneity of the aquifer system to better understand the vertical connectivity. Using six different forms of inversion and posterior sampling of the recovered resistivity models, we produced 6006 resistivity models. These models were transformed to models of sediment type and estimates of percentage of sand/gravel. Exploring the model space, containing the resistivity models and the derived models, allowed us to delineate the large-scale structure of the aquifer system in a way that captures and communicates the uncertainty in the identified sediment type. The uncertainty increased, as expected, with depth, but also served to indicate, as areas of high uncertainty in sediment type, the location of both large-scale and small-scale interfaces between sediment type. A plan view map of the integrated percentage of sand/gravel, when compared to existing hydrographs, revealed the extent of lateral changes in vertical connectivity within the aquifer system throughout the study area.

Upscaled transport models are needed to address regional-scale groundwater quality degradation in major aquifers worldwide, and these models must accurately characterize anomalous (non-Fickian) transport (e.g., early solute arrival due to preferential flow, and late time tailing). Although it is well known that seasonal pumping and recharge can cause vertical hydraulic gradient values to exceed horizontal gradients by 10-100 times, the relationship between large shifts in vertical to horizontal gradient ratio (VHGR) and plume migration is less understood. We simulate flow and transport of a conservative solute in a heterogeneous alluvial aquifer under varying VHGR representative of typical seasonal hydraulic gradient fluctuations and find that VHGR strongly impacts anomalous transport, plume migration, and mass transfer rates between hydrofacies. At low VHGR (e.g., ambient hydraulic gradient conditions corresponding to a more horizontal flow direction), low-K facies are diffusion-dominated, resulting in low mass transfer rates out of low-K material, increased spatial spreading along the mean flow direction, and tortuous particle trajectories. In contrast, at high VHGR (e.g., representative of pumping and recharge, and corresponding to an increasingly vertical mean flow direction), results in increasingly Fickian transport, explained by a shift from diffusion- to advection-dominated mass transfer in low-K facies, and illustrated by relatively non-tortuous, vertically oriented particle trajectories. We conclude that seasonal fluctuations in VHGR in a typical alluvial aquifer system may create oscillating transport behavior, with implications for contaminant transport and cleanup. The strong dependence of anomalous transport on VHGR implies that approaches to upscale anomalous transport may benefit from incorporating information about the VHGR, and highlights specific hydrogeologic forcings that cause upscaled transport methods to fail under transient boundary conditions.