Future changes in the trading of virtual water

The study addresses how global virtual water trade (VWT)—the green (soil moisture) and blue (renewable and nonrenewable) water consumed in producing traded agricultural goods—may evolve through the 21st century under changing socioeconomic and climatic conditions. Past work has focused on historical reconstruction of VWT and has shown that VWT doubled between 1986 and 2007, with growing reliance on nonrenewable groundwater for crop production in some regions. Yet, the spatial and temporal characteristics of future VWT across all water sources remain largely unknown. The purpose of this study is to provide the first comprehensive projections of future VWT, quantify how much green, blue, and nonrenewable groundwater will be embedded in agricultural trade, and identify key exporting and importing regions and basins. This is important for understanding how international trade may alleviate or exacerbate water stress, inform sustainable groundwater management, and guide policy on food-water security under population and climate change.

The paper synthesizes prior research showing: (1) VWT alleviates water stress in importing regions; (2) historical VWT intensified, roughly doubling from 1986 to 2007; (3) most traded water is green or renewable blue, but extraction and international transfer of nonrenewable groundwater have increased, with documented regional impacts such as subsidence, sea-level rise contributions, and water quality degradation; (4) previous global estimates vary due to different trade datasets and inclusion of processed commodities and by-products; and (5) while the historical VWT network has been reconstructed and drivers explored, comprehensive forward-looking projections under socioeconomic and climate scenarios have been lacking. This study fills that gap by linking SSP-RCP futures to trade-driven VWT evolution, including nonrenewable groundwater components.

The analysis uses the Global Change Analysis Model (GCAM), a market equilibrium, multi-sector model linking energy, water, land, economy, and climate. Key elements: (1) Scenario: SSP2 (reference, medium challenges) paired with RCP6.0; five GCMs (ISI-MIP datasets) provide climate impacts on renewable water supply (via Xanthos hydrologic model at 235-basin resolution), crop yields, hydropower availability, and building energy demands. (2) Water representation: Basin-level cost resource curves for renewable water, nonrenewable groundwater, and desalination determine source shares via a logit formulation; increasing depletion raises extraction costs, feeding back into sectoral prices and production. (3) Agriculture and trade: Crop production is endogenous with rainfed/irrigated shares, using MIRCA 2000-based historical patterns and exogenous water consumption coefficients (green and blue) by country/crop aggregated to 32 GCAM regions for 12 crop types plus two biomass categories. Trade follows a Heckscher-Ohlin single global market with uniform commodity prices; regions choose supply/demand at that price without bilateral preferences. (4) Virtual water accounting: Exports are traceable to basins using production shares; imports are not traceable bilaterally and are allocated proportionally. Basin-level surplus/deficit is computed by allocating regional demand to basins in proportion to basin production of each crop and growth type (rainfed or irrigated). Virtual green exports (VGE) are computed using green water consumption per crop and rainfed exports; virtual blue exports (VBE) use blue water consumption and irrigated exports. Virtual nonrenewable groundwater exports (VGWE) equal VBE multiplied by the ratio of groundwater depletion to total blue water withdrawals in the basin. Regional imports of green/blue/groundwater virtual water are derived by scaling global virtual export pools by each region’s import shares. Calculations are performed for 2010 through 2100 in 5-year steps across five GCMs; interregional trade is between GCAM’s 32 regions; intra-regional trade is not included in totals that explicitly state this exclusion. Basin-level analyses identify hotspots of export for green, blue, and nonrenewable groundwater.

• Global virtual water exports increase substantially under SSP2-RCP6.0. Virtual green water exports and virtual blue water exports more than triple from 905 and 56 billion m³ in 2010 to over 3200 and 170 billion m³ by 2100, respectively; uncertainty spans across GCMs.
• Virtual nonrenewable groundwater exports increase markedly: approximately fivefold by mid-century relative to 2010, and at least double by 2100.
• Crop drivers: Oil crops, corn, and wheat dominate virtual green water trade growth; wheat and rice are major drivers of China’s future blue water exports.
• Regional roles shift: China transitions from a net importer historically to a significant exporter by 2100 due to slowing and then declining population after 2030, freeing production for export. Many African regions become larger importers as rapid population growth outpaces domestic production.
• Export origins over time: Early-century exports intensify from water-stressed regions (Middle East, Pakistan, India). Later-century exports increasingly originate from water-rich regions with lower water requirements per unit production.
• United States becomes a major future exporter of virtual blue water via corn, fibers, and oil crops, while importing mainly miscellaneous crops (fruits, vegetables, nuts) as some Southwestern production shifts away from these.
• Nonrenewable groundwater export hotspots: United States, Mexico, western South America, and northern Africa. Saudi Arabia and the Indus Basin export nonrenewable groundwater by 2050 but largely cease by 2100 as depletion raises costs.
• Basin hotspots: Intensification of green/blue exports in many Chinese basins; strong exports from the Missouri River, La Plata, and Murray-Darling basins. Nonrenewable groundwater exports persist from U.S. aquifers such as the High Plains (excluding areas with net renewable recharge like the Missouri River basin), Central Valley, and Mississippi Embayment. The Nile, La Plata, and Murray-Darling basins use nonrenewable groundwater to produce rice, fibers, and corn for export.
• Comparisons to historical estimates show differences due to dataset and product aggregation choices; this study focuses on primary agricultural products (pre-processing).

Linking GCAM’s global market and water constraints with climate impacts provides the first integrated projection of how VWT might evolve through 2100, including explicit accounting of nonrenewable groundwater embedded in trade. Results suggest a restructuring of agricultural trade under demographic and climate pressures, with shifting exporter/importer roles and increased reliance on specific basins and aquifers. The anticipated growth in virtual blue and nonrenewable groundwater trade has implications for sustainability, as continued depletion can exacerbate regional environmental impacts (e.g., subsidence, contribution to sea-level rise, water quality degradation). The study highlights the importance of international trade in balancing regional food-water deficits but also underscores risks of embedding unsustainable groundwater use into global supply chains. While nearly 90% of blue water consumption is agricultural, other traded sectors (energy, industry) also influence water use and trade patterns; extending projections to these sectors and exploring a broader set of socioeconomic and climate mitigation pathways would refine understanding of future VWT trajectories and associated risks.

This work provides the first comprehensive, scenario-based projections of future virtual water trade across green, blue, and nonrenewable groundwater components, revealing large increases through 2100 and identifying key exporting basins and shifting regional roles (e.g., China’s transition to exporter; growing imports in Africa, India, and the Middle East). It advances virtual water research by integrating socioeconomic and climate drivers within an equilibrium trade framework and quantifying nonrenewable groundwater embedded in trade. Future research should: (1) assess VWT under alternative SSP-RCP combinations and mitigation scenarios; (2) incorporate energy and industrial sectors’ virtual water trade; (3) improve representation of bilateral trade patterns and intraregional flows; (4) refine basin-level demand allocation and import tracing; and (5) analyze policy and management strategies to reduce reliance on nonrenewable groundwater in export-oriented production.

• Scope limited to agricultural crops; virtual water in energy and industrial goods is not modeled.
• Trade is modeled in a single global market across 32 regions without bilateral preferences; imports are not traceable to specific exporting basins/regions once pooled.
• Basin-level demands are approximated as proportional to basin production within a region; this downscaling assumption may not capture true subregional consumption patterns.
• Intra-regional trade is not included in certain totals; results reflect interregional flows.
• Water consumption coefficients are exogenous and aggregated; product aggregation differs from some literature (excludes processed products/by-products), affecting comparability.
• Uncertainty arises from climate model differences and scenario assumptions (only SSP2-RCP6.0 emphasized; other pathways yield broader ranges).
• Groundwater depletion-cost relationships and resource limits follow modeled cost curves and may differ from local hydrogeologic and policy conditions.