Afforesting arid land with renewable electricity and desalination to mitigate climate change

The study addresses the challenge that, although afforestation is a widely practised carbon dioxide removal (CDR) method, its deployment is limited by competition for productive land and insufficient water resources, leading to potential conflicts with agriculture and higher food prices. CDR is expected to remove 150–1,180 GtCO₂ by 2100 depending on mitigation pathways, with global energy-sector emissions at 36.3 GtCO₂ in 2021 underscoring the urgency. Traditional afforestation/reforestation costs can be low, but land and especially water needs are high, and tree survival is water-limited in many arid projects. Seawater reverse osmosis (SWRO) desalination powered by increasingly low-cost renewable electricity offers a potential solution to overcome water scarcity. The purpose of this study is to evaluate the global CO₂ sequestration potential, costs, and energy-water system requirements of RE-based SWRO desalination for irrigating forests established on arid and semi-arid lands from 2030 to 2100, using a forest species mix suited to desert/tropical climates and a spatially resolved energy-water systems model.

The paper reviews the role of CDR in limiting warming to 1.5 °C, noting required removals of 150–1,180 GtCO₂ by 2100 and potentially higher cumulative removals depending on emissions reductions. Afforestation potential estimates include up to 1.5 GtCO₂ yr⁻¹ at US$50 tCO₂⁻¹ and up to 4.9 GtCO₂ yr⁻¹ at US$200 tCO₂⁻¹ by 2050. Despite relatively low afforestation/reforestation (AR) costs projected for 2100 (US$17–30 tCO₂⁻¹), AR has among the higher land and water requirements, creating competition with agriculture and raising food prices. Empirical evidence shows growth and survival of trees depend on reliable water supply, with mortality observed in arid-region projects. SWRO desalination has become more cost-effective and, when powered by RE, is a promising option for augmenting water supply under increasing water stress. Prior work assessed desalinated water for irrigating urban trees in Abu Dhabi; this study extends the idea to large-scale afforestation on arid/semi-arid lands over decades, using multi-species mixes and accounting for evapotranspiration, local precipitation recycling, and energy-water infrastructure. The paper also positions desalination-based afforestation relative to DACCS and BECCS, noting BECCS’ land and water constraints and DACCS’ declining costs and scalability, and highlights potential co-benefits of restoration (e.g., economic returns in Sahel).

Study scope and land selection: Potential afforestation areas were identified by intersecting restoration land (areas suitable for tree cover absent human influence) and bare land (neither vegetation, cultivation, urban, nor water bodies) with hydrological basins exhibiting: (a) high (40–80%) or extremely high (>80%) water stress; (b) low water stress (<10%) but low renewable water supply (<10 cm); or (c) arid/low water-use basins (demand <3 cm and supply <10 cm). Restoration nodes used maximum afforestable canopy area from literature; bare land canopy cover was capped at 20% (global average proxy), acknowledging environmental suitability is not fully considered. Areas climatically unsuitable for the selected tree mix were excluded using climate zone and USDA hardiness constraints; sites where temperature falls below 5 °C for more than 5 days were removed.

Forest composition and carbon accounting: A mix of eight tree species suited to desert/tropical climates was used: Coolibah, Paperbark (Melaleuca), Date palm, Turkish pine, Willow acacia, Ironwood (Casuarina), Kamani, and White mulberry. Tree allometry and CO₂ sequestration trajectories were developed from the Urban Tree Database and the CUFR tree carbon calculator (date palm allometry from dedicated studies). Shares of each species were tuned so that annual carbon storage matched mature tropical forest expectations, and carbon pools included above/belowground biomass, dead wood, litter, and soil. Mature trees were assumed at 40 years (range 20–70 in literature). Cumulative node- and country-level CO₂ storage were computed by summing across species, accounting for node afforestation area, species shares, tree density, and species-specific cumulative sequestration over time.

Water demand modeling: Tree water demand depends on local reference evapotranspiration (ET₀; FAO Map Catalog, 0.16° resolution, 1961–1990 mean), species-specific water-use coefficients, irrigation efficiency (sub-surface drip, 95%), and irrigated area share (up to 90% of canopy area). Water demand scales with canopy area as forests mature. Precipitation recycling due to increased tree cover was included with an upper bound of 26% across regions (data-limited). Desalinated seawater is sourced from the nearest coastline; horizontal distance and elevation determine pumping needs.

Energy and desalination systems modeling: A simplified LUT Energy System Transition Model (LUT-ESTM) was used to cost-optimize the electricity supply for SWRO desalination and water pumping at 0.5°×0.5° spatial and hourly temporal resolution for 2030–2100. Generation technologies include solar PV and wind with battery storage and power-to-gas supporting near-baseload SWRO operation. Component technical and financial parameters (CAPEX, OPEX, lifetimes, efficiencies) vary to 2050 per literature; post-2050 parameters are held at 2050 values due to data limitations. Component replacement is modeled by lifetime. Fixed O&M for SWRO includes membrane replacement (15% yr⁻¹). The model favors high full-load hours at SWRO, allowing curtailment over excessive storage; a 10% cap on excess electricity was applied and excess is credited at system LCOE.

Costs and metrics: The study computes levelized cost of electricity (LCOE), levelized cost of water (LCOW), and an annual historic CO₂ cost metric defined as the annualized sum of energy, desalination, irrigation, and land system costs divided by the average annual CO₂ sequestration rate for that decade. Land system costs include land rent, conversion, monitoring, O&M, and fertilizer. Sub-surface drip irrigation CAPEX/OPEX and decommissioning are included. A uniform WACC of 5% is assumed. Water transport infrastructure CAPEX/OPEX scales with vertical lift and horizontal distance. PV, wind, and battery capacities and land use are tracked. Outputs are aggregated to countries and nine major regions.

Geographic and temporal aggregation: The world is grouped into nine major regions within LUT-ESTM; country-level results (including those outside the nine regions) are reported. Key years highlighted include 2030, 2050, 2070 (approximate maturity of forests), and 2100.

• Cumulative CO₂ sequestration potential: 730 GtCO₂ (2030–2100) from afforestation on arid/semi-arid lands irrigated by RE-powered SWRO.
• Costs of CO₂ removal: Global average annual historic cost decreases from ~€457 per tCO₂ in 2030 to ~€99–100 per tCO₂ by 2100, driven by falling RE costs and rising forest sequestration rates. Nodes near coasts with abundant solar and cooler climates reach ~€50 per tCO₂ by ~2070; overall 2070 global cost range ~€20–300 per tCO₂.
• Regional potentials by 2070: MENA ~131 GtCO₂ (largest), Sub-Saharan Africa ~87 GtCO₂; Europe ~3.4 GtCO₂ and Eurasia ~1.2 GtCO₂ (limited suitable land/temperature).
• Annualized system costs and annual removals: Mid-century global annualized costs ~€1,499 billion for ~7 GtCO₂ yr⁻¹; by 2100 ~€1,462 billion for ~14 GtCO₂ yr⁻¹.
• Water and energy intensity: Global average water demand per tonne CO₂ ~198 m³ (Spain example ~129 m³ due to lower ET); energy demand ~1 MWh per tCO₂ by 2100 (ten largest-potential countries ~1–1.5 MWh tCO₂⁻¹ in 2070). Roughly 52% of energy for desalination; ~48% for pumping (country variation from <10% to >70% for pumping share).
• Energy and water costs: System LCOE declines from ~€46.9 MWh⁻¹ (2030) to ~€31.0 MWh⁻¹ (2100). LCOW decreases from ~€0.59 m⁻³ to ~€0.44 m⁻³ (2030–2100); water cost increases with distance/elevation from coast.
• Infrastructure drivers: Water transport infrastructure dominates annualized costs across regions; inland/raised sites (e.g., Iran with ~1,200 m weighted average lift vs global ~590 m) face higher costs.
• Technology mix: By 2050, >80% of electricity in countries with sequestration potential comes from solar PV; batteries meet up to 67% of electricity demand for steady SWRO operation. Modeled global excess electricity ~24% before capping to 10%.
• Spatial cost patterns: 2070 map shows lower CO₂ removal costs (€50–100 tCO₂⁻¹) near coastlines; detailed country results provided in supplementary data.
• Comparative CDR context (from Table 2): Afforestation with RE desalination averages ~€214 tCO₂⁻¹ in 2050 and ~€99 tCO₂⁻¹ in 2100; water demand ~416 m³ tCO₂⁻¹ (2050) falling to ~198 m³ tCO₂⁻¹ (2100); land efficiency improves from ~356 to ~192 km² per MtCO₂ between 2050 and 2100.

The findings demonstrate that combining low-cost renewable electricity with SWRO desalination can unlock substantial afforestation potential on arid lands by removing the primary constraint of water scarcity. With reliable irrigation, modeled forests reach high sequestration rates over decades, yielding cumulative removals of 730 GtCO₂ by 2100 and achieving long-run CO₂ removal costs competitive with key technological CDR options (DACCS and BECCS), particularly in coastal, high-irradiation, cooler-climate areas. The approach offers significant co-benefits beyond carbon, including combating desertification, reducing soil erosion and flood risk, and restoring ecosystems. However, costs are relatively high early on due to low initial sequestration rates, and benefits materialize after a multi-decade growth period. Spatial heterogeneity is pronounced: inland elevation and distance from coast drive higher pumping energy and infrastructure costs, while local climate (ET) affects water intensity per tCO₂. Energy system optimization with high PV shares and batteries enables near-baseload SWRO operation at declining LCOE and LCOW over time. Compared with DACCS, which may be cheaper by mid-century in some locations, desalination-based afforestation can still be more cost-effective in many countries and delivers ecosystem co-benefits. The results support including RE-desalination-supported afforestation as a major component of diversified, region-specific CDR portfolios.

This study introduces and quantifies a pathway to scale afforestation on arid and semi-arid lands by supplying irrigation through RE-powered SWRO desalination. Under modeled assumptions, the approach could remove ~730 GtCO₂ cumulatively by 2100, with global average costs falling toward ~€100 tCO₂⁻¹ and even ~€50 tCO₂⁻¹ in favorable coastal nodes by ~2070, while developing large-scale RE and water transport infrastructure. The work highlights the significance of water infrastructure costs, the dominance of solar PV with battery storage in the power mix, and the sensitivity of economics to distance/elevation from coasts and local evapotranspiration. The concept positions desalination-based afforestation as a viable long-term CDR option with substantial ecological co-benefits, complementary to DACCS and BECCS. Future research directions include: refining carbon sequestration trajectories for diverse, multispecies forests; improving estimates of precipitation recycling and local climate feedbacks; assessing long-term forest maintenance needs and carbon permanence risks (e.g., fire); optimizing brine management with mineral/chemical recovery; analyzing impacts on dust plume dynamics and teleconnections; improving cost models for long-distance pumping/piping and country-specific WACC; and integrating this option within broader sector-coupled energy-water-land planning frameworks.

Key limitations include: (1) Uncertainties in forest carbon sequestration projections—species mixes, growth in high-density forests, and use of urban tree allometry as proxies; (2) Limited species diversity (eight species) and simplified climatic suitability thresholds, potentially underrepresenting ecological complexity; (3) Water demand estimates rely on historical ET₀ datasets, species-specific coefficients, high assumed irrigation efficiency (95%), and a simplified, capped precipitation recycling effect (max 26%); (4) Land availability assumptions—bare land canopy cover set at 20% without full environmental suitability screening; restoration in cooler/boreal regions not considered; not all identified land will be restorable in practice due to competing uses and climate impacts (fires, floods); (5) Cost and technology projections fixed at 2050 values thereafter; uniform WACC (5%) across countries; linear scaling of pumping/piping costs with distance/elevation without detailed economies of scale; (6) Potential environmental externalities of desalination (e.g., brine disposal) are noted but not fully modeled; (7) Time-lag to significant sequestration (multi-decade forest maturation) complicates near-term planning; (8) Modeled excess electricity and sector coupling benefits are simplified. These factors affect the precision and generalizability of cost and potential estimates and warrant further detailed investigation.