Ending groundwater overdraft without affecting food security

The study investigates how ending groundwater (GW) overdraft affects global food production, prices, trade and hunger under climate change, and which policy measures can offset adverse impacts. Contextually, irrigated agriculture supplies about 40% of global food from 20% of cultivated land and accounts for a major share of freshwater withdrawals, with roughly 42% of irrigation water sourced from GW. Rapid GW expansion has supported food and income growth but has caused severe depletion and degradation in many basins, particularly in South and East Asia, West Asia and the USA, threatening long-term food security. Climate change is intensifying pressures by reducing surface water (SW) reliability, altering GW recharge, and increasing irrigation demand. With global population projected to reach about 9.2–9.7 billion by 2050 and food demand rising substantially, managing GW sustainably poses a key trade-off with achieving zero hunger (SDG 2) and sustainable water management (SDG 6). This paper quantifies the impacts of eliminating unsustainable GW use on food systems by 2050 and evaluates targeted interventions—increased agricultural R&D for irrigated crops, more effective rainfall management (ERM), and reduced meat consumption in high-income countries (RMC)—to mitigate negative food security outcomes.

Extensive literature documents both the benefits and externalities of intensive GW development. Prior work highlights GW’s role in stabilizing irrigation and livelihoods but underscores depletion-driven risks: declining water tables, rising pumping costs, land subsidence, saltwater intrusion, reduced base flows and transboundary tensions. Studies link GW depletion to sea-level rise and even shifts in Earth’s axis of rotation, and note that solar pumping—while lowering emissions—can accelerate depletion. Empirical evidence from India and elsewhere connects severe depletion to reduced cropping intensity, migration and constrained adaptation capacity. Global assessments reveal hotspot regions of depletion and uncertain recharge trajectories under warming. Food demand projections for 2010–2050 suggest a 35–56% increase with millions more at risk of hunger, raising concern that reliance on depleting GW undermines food security. Previous modeling studies have shown detrimental effects on food production when curbing depletion, though some omitted aquifer contributions to environmental flows and did not explore complementary policy investments. The present study builds on and extends this literature by explicitly integrating a hydrology–water–economy modeling framework, incorporating aquifer environmental flow contributions, and systematically testing policy levers to reduce food security trade-offs of GW conservation.

The analysis employs the International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT), a linked suite of biophysical–economic models integrating hydrology, water management, crop stress and multi-market trade/demand. Key components include: (1) IMPACT Global Hydrology Model (IGHM) simulating monthly SW availability, soil moisture and runoff at 0.5° grid resolution; (2) IMPACT Water Simulation Model (IWSM) optimizing allocation of water among sectors (domestic, industrial, livestock, irrigation) at 320 food production units (FPUs), subject to water availability, infrastructure capacities, and priorities (irrigation lowest); and (3) the IMPACT multi-market model determining supply, demand, trade and prices annually across 62 commodities for 158 countries/regions. Climate and socioeconomic forcings: Three GCMs (IPSL, GFDL, HadGEM) under RCP 8.5 are used, coupled with SSP2 trajectories (global population ~9.2 billion by 2050; average income ~US$25,000/person). Climate impacts on crops are simulated using DSSAT. Sensitivity across GCMs shows small differences in 2050 outcomes; a broader climate analysis is acknowledged as beyond scope. New GW module and coupling: A new module links IWSM and IGHM to capture GW–SW interactions. Steps: (1) IWSM projects monthly GW withdrawals at FPU level; (2) downscale to 0.5° grid-cell demands; (3) IGHM simulates shallow/deep GW storage balance with imposed grid-cell GW demand; (4) aggregate grid-cell pumping back to FPUs; (5) rerun IWSM/IMPACT with updated GW pumping and altered baseflows, iterating annually over 2005–2050. GW storage is modeled with vertical flows only, using a two-layer representation: a replenishable shallow aquifer and a non-replenishable deep aquifer. Pumping first draws from shallow storage up to availability; unmet demand draws from deep storage (always depleting), with irrigation return flows percolating to shallow storage next month and domestic/industrial GW return flows routed to SW within the same month. Environmental flow contributions from aquifers are represented via Q90 (monthly runoff exceeded 90% of the time). Calibration and capacities: IWSM’s GW withdrawal capacity per FPU is calibrated to FAO AQUASTAT 2005 withdrawals; capacities grow exogenously over time to reflect infrastructure/technology changes. Total withdrawal capacity equals observed base-year withdrawals times a factor driven by the coefficient of variation of demand and a tuning parameter φ chosen to minimize deviation from observed GW withdrawals. Scenarios: Baseline “No GWC” reflects climate change without GW conservation. The GW conservation (GWC) scenario caps GW withdrawals at net recharge less GW’s environmental flow contribution. Excess withdrawals in depleted basins are cut by 33% annually from 2025 to 2027 to reach net recharge by 2027; depletion is halted going forward (no aquifer recovery assumed). Three policy interventions are layered on top of GWC: (1) +R&D: accelerated investments to raise irrigated crop yields via improved varieties and agronomy (water-use efficiency, stress tolerance); (2) +ERM: improved effective rainfall management on irrigated and rainfed lands (e.g., conservation agriculture, mulching, terracing); (3) +RMC: reduced meat consumption in high-income countries via lower demand elasticities. A combined +R&D+ERM scenario exploits synergies. Least-cost investment levels for R&D and ERM are derived via multi-search sensitivity simulations minimizing squared deviations from pre-GWC food security metrics; for RMC, elasticities are reduced to lower LMIC hunger by 25%, corresponding to ~9% meat consumption decline in HICs. Outcome metrics: Changes in production, world prices, trade flows and population at risk of hunger by 2050 are assessed, averaging across the three climate models under RCP 8.5.

Baseline demand growth: IMPACT simulations project global food crop demand rising 40% from 2020 to 2050, with large absolute increases for fruits/vegetables and oilseeds. Irrigation water demand rises 17% globally (largest absolute increase in LMICs by ~201 bcm), while domestic and industrial demands grow faster in percentage terms. Extent of current depletion: Defining sustainable withdrawals as not exceeding net recharge, 25% of river basins are overexploited yet account for 61% of global GW withdrawals. Top overdraft basins are in India, Pakistan, China, Saudi Arabia, Iran, the USA and Egypt, totaling ~279 bcm or ~83% of global depletion. GW withdrawals tripled from 227 bcm (1990) to 879 bcm (2020); SW withdrawals doubled (963 to 2,697 bcm). Effect of GWC on withdrawals: Under GWC, total withdrawals fall by ~342 bcm by 2027 versus baseline, mostly in LMICs (~297 bcm), with HICs ~45 bcm. Country reductions include India −164 bcm (to 39% of baseline), China −34 bcm (to 69%), USA −12 bcm (to 76%) by 2027. Production and price impacts (2050): Halting depletion lowers production and raises prices. Approximate global production changes: rice −1.8%, wheat −1.5%, maize −1.0%, sugarcane −0.6%; all agricultural commodities −0.73%, all foods −0.66%. Price increases: rice +7.4%, wheat +6.7%, cereals on average +5.2%; other crops generally +1.4% to +4.0%. Trade adjusts: net trade in rice +10.8% and sugar +3.5%; global net wheat trade −3.3% due to reduced South Asia exports. Food security impacts: Higher prices increase the population at risk of hunger by ~26 million (+5.0%) globally by 2050 relative to No GWC (baseline ~520.6 million). Increases are concentrated in LMICs (+23.9 million; +5.0%), with HICs +2.1 million (+4.6%). Country examples: China +5.2 million (+8.0%), India +2.6 million (+6.7%), USA +0.7 million (+14.2%). Policy scenarios with GWC:

• +R&D (least-cost ~4.5% irrigated yield increase by 2050): Strong mitigation. Wheat price increase cut to +3.0% (from +6.7%); cereals +1.2% vs baseline; sugar +0.8%. Population at risk of hunger rises only +0.9% vs +5.0% under GWC.
• +ERM (least-cost ~4.5% improvement in effective rainfall use): Moderate mitigation. Cereal prices +1.9% vs baseline; maize prices fall more than in +R&D due to rainfed dominance; rice prices remain higher than in +R&D due to irrigation dependence. Hunger +1.9% vs baseline.
• +R&D+ERM (2% irrigated yield + 3% ERM): Combines benefits; lowers maize more than +R&D and rice/wheat more than +ERM; still above +R&D on cereals overall; further reductions for sugar and oilseeds. Hunger +1.1% vs baseline.
• +RMC (HIC meat demand elasticities −75%, ~9% consumption decline): Small global price effects except meat (−1.1% vs baseline); maize still +2.5% vs baseline; limited mitigation of hunger relative to technology/rainfall interventions. Overall, targeted investments in irrigated productivity and better rainfall management can largely offset the adverse food security impacts of halting GW depletion, whereas dietary shifts in HICs offer only marginal global relief.

The study demonstrates a clear trade-off: enforcing sustainable GW withdrawals stabilizes aquifers and baseflows but, absent compensatory measures, depresses crop output and elevates prices, exacerbating hunger—especially in LMICs and Asian breadbasket regions. However, technology-led productivity gains for irrigated crops and improved use of precipitation can substantially mitigate these impacts by raising water productivity and buffering production against constrained GW. The results emphasize interlinkages between SDG 6 (sustainable water) and SDG 2 (zero hunger) and support a transdisciplinary, multi-pronged strategy: combine water management improvements (ERM), crop genetic and agronomic gains (R&D), and supportive institutions. While reduced meat consumption in HICs modestly eases pressure on feed demand and prices, it is less effective globally than R&D and ERM, though it can enhance diet affordability in LMICs. Governance and implementation are pivotal: many basins suffer from weak GW institutions, misaligned energy subsidies, limited monitoring, and enforcement gaps. Emerging policies (e.g., California’s SGMA), social learning and community stewardship pilots in India/Ethiopia, and innovative energy–water schemes (e.g., selling solar power instead of pumping) show promise. Importantly, only about one quarter of basin areas are currently overexploited—albeit many are population and food-production centers—while other basins retain scope for sustainable GW expansion under appropriate monitoring and safeguards. Policymakers should anticipate transitional price and production impacts of GWC and plan compensatory investments to prevent worsening hunger.

Ending GW overdraft is essential for long-term water and food system resilience but, if enacted in isolation, would raise food prices and increase the number of undernourished by mid-century. The paper’s main contribution is quantifying these trade-offs under climate change and identifying least-cost, system-level interventions—targeted irrigated yield improvements and effective rainfall management—that can largely neutralize adverse food security impacts. A combined regulatory, technological, financial and behavioral approach, supported by robust institutions and monitoring, is recommended to align SDG 6 with SDG 2. Future research should: (1) improve aquifer-specific resolution to assess localized price and production impacts and capture river capture processes; (2) broaden climate and socioeconomic scenario coverage and uncertainty analyses; (3) evaluate costs, adoption dynamics and equity impacts of R&D and ERM at regional scales; (4) integrate managed aquifer recharge and conjunctive use strategies; and (5) assess policy instruments (pricing, permits, energy reforms) and governance innovations for practicality and effectiveness.

Key limitations include: (1) inability to draw aquifer-specific conclusions or price impacts from individual aquifer failures due to data uncertainty and model granularity; (2) climate forcing limited to three GCMs under RCP 8.5 and SSP2, with broader uncertainty exploration outside scope; (3) the IGHM does not simulate surface water withdrawal processes or increased river capture due to GW depletion; (4) the GWC scenario halts further depletion but does not model aquifer recovery; (5) base-year calibration relies on AQUASTAT withdrawals and inferred capacities; (6) implementation feasibility of GWC faces real-world constraints (subsidies, enforcement, illicit pumping) and the adoption of R&D/ERM technologies entails time, costs and capacity; and (7) results reflect average global effects; regional heterogeneity can be substantial, especially in hotspot basins.