Divergent effects of climate change on future groundwater availability in key mid-latitude aquifers

Groundwater, a vital resource for both humans and ecosystems, is increasingly critical in mid-latitude arid and semi-arid regions where surface water is scarce. The growing demand for groundwater, driven by population growth, is exacerbated by climate change, increasing the risk of severe droughts. Understanding the interplay between climate-driven and anthropogenic impacts on groundwater storage (GWS) is therefore paramount. Climate change affects groundwater through alterations in the hydrological cycle, impacting infiltration, percolation, and recharge. Rising temperatures increase evapotranspiration, reducing groundwater replenishment. Anthropogenic impacts primarily stem from groundwater pumping and the indirect effects of irrigation and land-use changes. Most large-scale GWS change estimates rely on numerical modeling, often using offline hydrological models driven by General Circulation Models (GCMs). These offline approaches suffer from uncertainties in GCM projections, downscaling methods, and a failure to capture crucial land-atmosphere feedbacks. This study overcomes these limitations by employing a fully coupled Community Earth System Model – Large Ensemble Project (CESM-LE), which incorporates land, atmosphere, ice, and ocean components, providing a holistic Earth system perspective and accounting for internal climate variability. The model includes a physically based groundwater parameterization within the Community Land Model (CLM4.0), simulating water table depth, recharge, discharge, and interactions with overlying soils. While acknowledging limitations at local scales, the coupled nature of CESM-LE is well-suited for analyzing large-scale interactions and feedbacks within the hydrological cycle. The study focuses on projecting climate-driven GWS changes using CESM-LE simulations (2006–2100) under the RCP8.5 high-emission scenario, excluding anthropogenic water use effects to isolate climate impacts.

Previous studies estimating large-scale GWS changes often relied on offline simulations of hydrological models driven by GCMs. These studies faced challenges related to uncertainties in climate projections stemming from internal climate variability, inter-model differences, and scenario uncertainties. Offline simulations also introduced uncertainties associated with downscaling techniques and the inability to capture land-atmosphere feedbacks. Previous research highlighted the feedback mechanisms between groundwater and the atmosphere, emphasizing the need for coupled climate modeling to accurately assess GWS changes. This study contrasts with previous work by using a fully coupled climate model, specifically the CESM-LE, allowing for a more comprehensive understanding of the complex interactions within the Earth system.

This research utilizes the Community Earth System Model Large Ensemble Project (CESM-LE) to simulate climate-driven GWS changes. CESM-LE, a fully coupled climate model encompassing land, atmosphere, ocean, and ice components, was employed to account for internal climate variability using a large ensemble approach (30 members). The model's land surface component, CLM4.0, incorporates a physically-based groundwater parameterization that simulates water table depth, recharge, discharge, and interactions with overlying soils. Simulations were conducted for the period 2006-2100 under the RCP8.5 high-emission scenario, excluding anthropogenic effects (groundwater pumping, dam impoundment, and water transfer) to isolate the climate change signal. Seven key mid-latitude aquifers, previously identified as severely depleted, were selected for analysis. These aquifers are: Central Valley (California), Southern Plains (Central U.S.), Middle East (Tigris-Euphrates basin), Northwestern India (Indus-Ganges basin), North China Plain, Guarani Aquifer (South America), and Canning Basin (Northwestern Australia). The CESM-LE simulations provided monthly data on various hydrological variables, including precipitation, rainfall, snowfall, evapotranspiration (ET), snowmelt, surface runoff, infiltration, groundwater recharge, and groundwater storage. Annual mean GWS trends and seasonal variations were analyzed using the ensemble mean of 30 members. To assess the contribution of rainfall, snowmelt, and ET to groundwater recharge changes, a multiple linear regression analysis was employed. The study also used satellite-based estimates of GWS changes (2003-2014) derived from GRACE (Gravity Recovery and Climate Experiment) data and GLDAS (Global Land Data Assimilation System) to compare with the CESM-LE projections. Finally, a comparison was made between climate-driven and combined (climate-driven + anthropogenic pumping) GWS changes using twentieth-century simulations that included the effect of groundwater pumping from a previous study by the authors.

The study's key findings reveal a complex and spatially heterogeneous response of GWS to climate change. Changes in GWS are not simply a reflection of precipitation trends; instead, they are significantly influenced by changes in evapotranspiration and snowmelt. In some regions, increased evapotranspiration and reduced snowmelt offset the positive effects of increased rainfall on GWS. Specifically: * **Central Valley (California):** No significant long-term trend in GWS is projected under climate change alone. Competing effects of increased rainfall (but less snowmelt) and increased evapotranspiration result in a near-neutral impact. * **Southern Plains (Central U.S.):** A significant decline in GWS (-23.3 ± 11.4 mm dec⁻¹) is projected due to reduced infiltration and snowmelt. * **Middle East:** A substantial decline in GWS (-15.2 ± 3.4 mm dec⁻¹) is predicted, primarily due to reduced snowmelt and increased evapotranspiration. * **Northwestern India:** In contrast to other regions, an increase in GWS is projected due to increased rainfall outweighing increased ET and decreased snowmelt. * **North China Plain:** Similar to Northwestern India, increasing precipitation results in a projected increase in GWS. * **Guarani and Canning Basins:** These regions show projected increases in GWS, driven primarily by increased precipitation exceeding increases in evapotranspiration. The regression analysis revealed that rainfall dominates groundwater recharge changes in monsoon and humid regions. In snow-dominated regions, snowmelt is a key factor. In drier regions, evapotranspiration emerges as the dominant factor influencing groundwater recharge. Comparing climate-driven GWS changes with estimates that incorporate anthropogenic pumping, the study highlighted that the negative impacts of pumping significantly outweigh natural replenishment in heavily pumped regions. The GRACE-based estimates of GWS trends (2003-2014) indicate that recent depletion is primarily due to anthropogenic pumping in the Central Valley and Northwestern India, while climate variability plays a larger role in the Guarani and Canning Basins. The projected future climate-driven groundwater depletion in the Southern Plains and Middle East greatly exceeds that observed in the 20th century.

This study provides valuable insights into the divergent impacts of climate change on groundwater resources, demonstrating that simple correlations between precipitation and GWS are insufficient to fully understand future trends. The findings highlight the importance of considering other hydrological processes, such as evapotranspiration and snowmelt, which are affected by climate change and interact in complex ways. While the model excludes the effects of future pumping, the comparison between climate-driven changes and the combined effects of climate change and pumping suggests that anthropogenic impacts are likely to dominate in many regions. The large spatial heterogeneity in GWS changes observed underscores the need for regionally specific assessments. The use of a fully coupled climate model addresses limitations of previous studies that employed offline modeling approaches, allowing for a more accurate representation of land-atmosphere feedbacks and the complex interactions within the Earth system. The study's findings have important implications for water resource management, especially in regions already facing significant groundwater stress, such as the Southern Plains and Middle East. The model's relatively coarse resolution might underestimate the contribution of snowmelt in mountainous areas, suggesting a need for further refinement of hydrological parameterizations in complex terrains. Although the study provides valuable projections, the model-dependence and scenario-dependence of the results should be acknowledged, and further research using different fully coupled Earth System Models and various emission scenarios is warranted.

This study employs a fully coupled Earth System Model to project climate-driven changes in groundwater storage in seven globally important aquifers. The results demonstrate the complex and spatially heterogeneous impacts of climate change, highlighting the crucial role of evapotranspiration and snowmelt alongside precipitation. The findings underscore the need to integrate both climate-change projections and anthropogenic impacts when assessing future groundwater availability. Future research should focus on refining model parameters, incorporating improved representations of local-scale processes, and explicitly considering projected future socioeconomic factors and groundwater management practices to produce more precise and reliable predictions.

This study acknowledges several limitations. Firstly, the relatively coarse resolution of the GCM may underestimate the influence of snowmelt in mountainous regions. Secondly, the model simplifies local-scale processes, potentially leading to some inaccuracies. Thirdly, the study excludes future socioeconomic factors and groundwater management practices, which could significantly influence groundwater availability. The analysis focuses on a specific high-emission scenario (RCP8.5), limiting the generalizability of findings to other emission scenarios. Finally, the results are model dependent, suggesting that further research using alternative coupled ESMs is needed to build confidence in the projections.