China Southern Power Grid's decarbonization likely to impact cropland and transboundary rivers

China Southern Power Grid (CSPG)—covering Guangdong, Guangxi, Yunnan, Guizhou, and Hainan—has historically relied on coal generation. In support of China’s pledge to reach carbon neutrality by 2060, CSPG plans radical changes to generation and transmission, including large-scale expansion of solar, wind, and hydropower (>32 GW new hydro). A substantial fraction (≈78%) of the new hydropower capacity is planned on transboundary rivers (Mekong/Lancang and Nujiang/Salween), raising ecological and geopolitical concerns. The study asks: (1) What land is required to achieve carbon neutrality? (2) How much additional hydropower is needed to decarbonize CSPG? (3) How does protecting transboundary basins (i.e., excluding new dams on them) affect decarbonization pathways and land requirements? To answer these questions, the authors develop a spatio-temporal capacity expansion and operations model of the CSPG to co-optimize generation, storage, and transmission under multiple decarbonization and hydropower development scenarios.

The work is situated within a growing body of research on China’s carbon-neutral pathways and grid transformation, which highlights the accelerating cost decline of renewables and storage and the potential role of firm low-carbon resources. Prior studies emphasize the feasibility and investment planning for China’s net-zero transition, the role of solar-plus-storage, and the need for CCS in deep decarbonization. Parallel literature documents ecological and social impacts of large hydropower (e.g., river fragmentation, sediment/nutrient disruption, biodiversity impacts) in the Yangtze and Mekong, as well as land-use conflicts from utility-scale wind/solar. The study contributes by explicitly integrating geophysical constraints and cross-border hydropower siting issues into power system planning for CSPG, quantifying unintended consequences on land and water resources.

Modeling framework: A CSPG-specific implementation of the open-source GridPath platform co-optimizes capacity expansion and hourly operations over 2020–2060 in 5-year investment periods (with 2065 included to handle end effects). Spatially, nine load zones are modeled (five CSPG provinces plus four external trading regions). Operations are represented by 24 hours for 12 representative days (one per month) per planning year, balancing load in each zone with a 15% planning reserve margin.

Supply portfolio and techno-economics: Eleven technologies are modeled—coal, gas, biomass, nuclear, hydro (conventional), wind, solar, coal-CCS, gas-CCS, pumped storage hydropower (PSH), and battery storage. Existing assets (circa 2020) and candidate expansions are included. About 32 GW of new conventional hydro and 53 GW of new PSH are considered as candidates. Capital and O&M costs, fuel prices, heat rates, ramping, lifetimes, emissions factors, and transmission losses are compiled from Chinese sources and NREL ATB; time-varying capital cost trajectories are applied to wind, solar, batteries, coal-CCS, and gas-CCS. Fuel prices for coal/gas follow NREL ATB projections; others are static at 2020 levels.

Demand and trade: Province-level load is built from high-resolution 2020 data and extrapolated using CSPG/SGCC growth rates (6.5% to 2025, 3.5% to 2030, then 1%/yr to 2060). Cross-border trade on links to neighboring provinces/countries is held at 2020 levels owing to limited forward data.

Transmission: 2020 transmission corridors (with capacities, losses, lengths) are included; intra-CSPG lines can be cost-optimally expanded from 2025 onward, while CSPG-to-external links remain fixed. A generic loss factor of 1% per 100 miles is used. Line capital costs are from national planning sources.

Hydropower availability: Monthly hydropower availability for existing and planned dams is simulated using Xanthos (forced by WFDEI and NorESM-based climate), at 0.5° resolution, 1970–2010 for calibration and with projections under RCP4.5/8.5. Average monthly capacity factors are computed for existing plants; planned plants inherit factors from nearest similar existing dams within the same basin. Sensitivity runs vary monthly hydro capacity factors for 2025–2060 under climate scenarios.

Wind/solar siting and availability: Candidate wind and solar sites are identified via REZoning (MapRE-based), screening at 500 m resolution and aggregated to 50 km, applying geophysical and land-use constraints and estimating LCOE; multi-criteria prioritization yields 94 wind and 340 solar candidate sites. Existing capacities are spatially matched and deducted to obtain remaining potentials. Hourly availability time series for 2020 are generated using pyGRETA forced by MERRA-2 (GHI, TOA, 2 m temperature for solar; 50 m winds for wind), assuming single-axis tracking PV and GE 2.5 MW turbines at 100 m hub height. Land-use requirements are computed via installed capacity area factors (MW/km²).

Optimization: Mixed-integer linear programming in Python/Pyomo solved with Gurobi minimizes net present total system cost (investment + operations) with a 7% discount rate. Binary decisions govern selection of discrete wind/solar sites and hydro/PSH projects; continuous decisions for conventional additions and batteries. Emissions constraints linearly decline from 2020 levels to target year in decarbonization scenarios. Outputs include capacities by period, hourly dispatch, curtailment, interprovincial trade, emissions, and costs.

Scenarios: Core scenarios include (i) Reference (no carbon neutrality target), and decarbonization to neutrality by 2060, 2050, or 2040. Each is run with full hydropower plan and with Without Transboundary Hydropower (WTH) variants (excluding new dams on Mekong/Nujiang). Climate-induced hydro availability sensitivities are also tested. Available new hydro capacity is 32.34 GW; candidate solar/wind potentials correspond to 340 and 94 sites, respectively.

• Under Reference (no neutrality): By 2060, coal remains dominant—≈33% of new capacity and ~50% of generation; renewables expand late (post-2040). New additions by 2060 include ~283 GW solar and ~62 GW wind, together ~21% of generation; hydro provides ~22% with >32 GW new hydro. Utilization of available new capacity by 2060: solar 83%, wind 66%, hydro 100%.
• With neutrality by 2060 (Decarb. 2060): Rapid wind/solar buildout—new capacity added in 2030 totals ~163 GW (vs 74 GW in Reference). By 2060 new capacity includes ~291 GW solar and ~62 GW wind. Hydro totals ~172 GW (32.34 GW new), yielding ~25% of generation. Coal-CCS is deployed at scale to replace conventional coal and support VRE; by 2030 coal-CCS is 64–86 GW across decarb timelines (more for earlier targets). Near-total exploitation of wind/solar/hydro potentials is required to meet rising demand and neutrality.
• Accelerated neutrality (2050/2040): Front-load investments in wind/solar and larger/faster coal-CCS additions relative to 2060 neutrality.
• Transmission: System-wide transfer capacity expands by ~96–134% over 2020 levels. Expansion patterns differ under neutrality (e.g., Yunnan→Guizhou corridor to access hydro; more eastern interconnections to access wind/solar), boosting interprovincial trade.
• Hydropower siting/timing: The full 32.34 GW candidate hydro is selected in all scenarios by 2060; ≈78% is on transboundary Mekong and Nujiang. Decarbonization scenarios require much earlier dam development—~23 GW by 2025—with basins committing substantial shares by 2025: Mekong 48%, Nujiang 68%, Yangtze 100% of candidate capacity.
• Land requirements: Wind+solar require ~40,000 km² by 2060 (~one-quarter of Guangdong’s area), similar total across scenarios but much faster land conversion under decarbonization. By 2030, wind and solar land needs are ~1.3× and ~3× those in Reference, implying ~10,000 km² earlier conversion. Land burdens are uneven—Guangxi bears ~43% of total land needs; ~90% of sites overlap cropland/grassland/shrubland, implying socio-ecological trade-offs.
• Excluding transboundary dams (WTH): Carbon neutrality remains feasible by substituting additional coal-CCS (e.g., +~20 GW in Decarb. 2060 WTH vs Decarb. 2060) and more aggressive wind/solar build, which further accelerates early land conversion (especially pre-2030).
• Emissions and costs (2020–2060 cumulative): Decarbonization reduces CO₂ by 74–86% (≈6,254–7,246 Mt CO₂ reduction) at a total system cost increase of 6.3–10.6% ($76–127 billion) vs Reference. Excluding transboundary hydropower increases corresponding decarbonization scenario costs by ≤1% (≤$12 billion).

The results demonstrate that CSPG can feasibly achieve carbon neutrality by 2060—and even by 2050 or 2040—with a portfolio centered on wind, solar, hydropower, coal-CCS, and enhanced interprovincial trade. However, renewable capacities alone are insufficient to meet rising demand while eliminating unabated fossil generation; firm low-carbon coal-CCS plays a pivotal bridging role across all neutrality timelines. The analysis highlights major resource implications: rapid hydropower expansion—particularly on ecologically sensitive, transboundary Mekong and Nujiang—and large-scale land conversion for wind/solar concentrated in cropland/grassland. Excluding new transboundary dams is technically feasible at modest cost increases, but requires greater coal-CCS deployment and earlier, faster land conversion, intensifying siting trade-offs.

These findings address the core questions by quantifying land needs (~40,000 km² by 2060), the scale/timing of hydropower additions (32.34 GW with front-loaded deployment under decarbonization), and the system-level consequences of protecting transboundary basins (small cost penalty, larger CCS and land-use burdens). The study underscores ecological and geopolitical risks of river fragmentation and downstream impacts, and the need for proactive land-use planning (e.g., agrivoltaics, siting design, stricter environmental regulation) and multi-objective planning to manage trade-offs. It also emphasizes that near-optimal alternatives likely exist with similar costs, suggesting policymakers have flexibility to steer toward solutions that reduce socio-ecological conflicts while meeting climate targets.

CSPG can meet carbon neutrality by 2060—and potentially earlier—via rapid expansion of wind/solar, full utilization of planned hydropower, substantial coal-CCS deployment, and significant transmission upgrades and trade. Decarbonization entails substantial land (~40,000 km² by 2060) and water impacts, notably early dam development concentrated on transboundary basins. Nevertheless, neutrality without new transboundary dams is achievable at ≤1% additional system cost, with compensating increases in coal-CCS and accelerated wind/solar build.

Main contributions include: (i) a high-resolution, multi-decade co-optimization of CSPG expansion and operations integrating geophysical resource constraints; (ii) explicit quantification of land requirements and basin-level hydro development timing; and (iii) evaluation of neutrality pathways that avoid transboundary dam impacts.

Future research should: (1) explore multi-objective planning that balances cost, emissions, land, water, and biodiversity; (2) map and evaluate near-optimal solution spaces to identify robust, low-conflict portfolios; (3) assess CCS deployment feasibility, infrastructure, and environmental risks; (4) refine land-siting strategies (e.g., agrivoltaics, co-location, higher land-use efficiency) and socio-economic impacts; and (5) extend to broader electrification scenarios (transport, industry) and dynamic trade/transmission expansion planning.

• Objective function and scope: Single-objective cost minimization with emissions caps; does not internalize environmental/social externalities or explicitly optimize for land/water/biodiversity.
• Technology and cost assumptions: CCS at large scale is assumed available with mid-range future costs; significant uncertainty in costs, performance, storage availability, and policy. Many technology capital costs are held static except for select technologies; fuel price and cost projections are uncertain.
• Temporal/spatial resolution: Operations approximated by 12 representative days per year; nine-zone spatial resolution may mask intra-provincial constraints and siting conflicts. Transmission losses and distances are simplified.
• Hydropower availability: Monthly capacity factors derived from Xanthos with nearest-neighbor estimates for planned dams; limited climate scenarios (two GCM/RCP combinations) and monthly (not sub-monthly) variability.
• Demand and trade: Demand growth is exogenously projected; cross-border trades held constant at 2020 levels due to data limitations.
• Policy implementation: Emissions decline linearly to target year; real-world policy and institutional constraints on investment timing, permitting, and siting are not modeled.