Large-scale photovoltaic solar farms in the Sahara affect solar power generation potential globally

The study investigates whether and how large-scale deployment of photovoltaic (PV) solar farms in the Sahara Desert could alter global atmospheric conditions and, consequently, solar power generation potential elsewhere. Solar energy is expanding rapidly and is highly sensitive to cloud variability and surface shortwave radiation (RSDS). Prior projections of climate change generally show modest regional RSDS changes (typically within ±10%), but massive PV deployments can modify land surface properties (e.g., albedo, roughness), potentially triggering regional-to-global climate responses. The authors pose a pilot case study using an Earth system model to quantify global impacts on PV power generation potential from hypothetical Sahara solar farms covering 5%, 20%, and 50% of North Africa, and to diagnose the underlying atmospheric forcing mechanisms. The work is motivated by the need to understand system-wide risks for energy security as solar becomes a dominant energy source.

The paper situates its contribution within several strands of research: (1) growing reliance on weather-sensitive solar power and the operational challenges for grids; (2) climate change impacts on solar generation, where most studies report modest regional RSDS and cloud cover changes (often within ±10%); (3) land surface changes from large-scale PV installations that can alter surface albedo and roughness, leading to regional climate responses; and (4) prior modeling over arid regions (e.g., North Africa) showing amplified regional climate responses via land–atmosphere–vegetation feedbacks and potential global teleconnections. The authors build on earlier Sahara-focused simulations that assessed climate and vegetation responses, extending explicitly to quantify impacts on global solar generation potential and diagnose large-scale atmospheric mechanisms.

Model: EC-Earth version 3.3.1 (EC-Earth3-Veg-LR configuration) couples atmosphere (IFS), ocean (NEMO), sea ice (LIM) and terrestrial biosphere (LPJ-GUESS). Atmospheric grid is T159 (~1°) with 62 vertical levels; ocean grid ~1° with 75 levels. The model reproduces key dynamics (general circulation, monsoon systems, teleconnections, and cloud structures) within CMIP6 performance ranges. Solar farm scenarios: Three experiments prescribe large PV coverage over North Africa (15–30°N, 20°W–45°E): S05 (5% area), S20 (20%), S50 (50%). PV sites are represented by modifying bare soil to an effective albedo of 0.235, reflecting both panel reflectivity and conversion of absorbed solar energy to exported electricity. Effective albedo calculation assumes panel reflectivity 0.147 and laboratory conversion efficiency 0.1548, yielding 0.1 + 0.15*(1−0.1) ≈ 0.235. This removes 0.135 of incoming shortwave from local heat budget. Simulation design: Simulations initialized from a default 1990 CE climate and integrated for 210 years to quasi-equilibrium; greenhouse gases, aerosols, and land-use fixed at 1990 CE values to isolate PV forcing. Last 60 years used for analysis. Sensitivity tests with 2010 CE boundary conditions (CTRL and S50) showed marginal differences, suggesting robustness to initial/boundary state. Additional exploratory solar farm simulations over other drylands (Central Asia, Central Australia, Southwestern US, Northwestern China) were conducted; global RSDS responses are provided in supplementary material. PV power potential metric: PVpot(t) = [1 + γ (T_cell(t) − T_STC)] × RSDS(t)/RSDS_STC, with RSDS_STC = 1000 W m−2, T_STC = 25 °C, γ = −0.005 °C−1, and T_cell approximated by surface temperature from EC-Earth. PVpot equals 1 under STC and decreases with higher temperature and/or lower irradiance. Atmospheric diagnostics: Vertical moist stability (VMS) defined as MSE_upper(200–500 hPa) − MSE_lower(1000–700 hPa) to assess convective stability in tropical/subtropical regions. Lower-troposphere dry static stability (potential temperature difference 700–925 hPa) and moisture flux/divergence diagnostics used. Seasonal (DJF, JJA), annual means, and interannual variability (standard deviation) of RSDS and clouds analyzed; statistical significance assessed via t-tests (95%).

- Cloud and RSDS response scales with PV coverage: Minimal significant change in S05; pronounced redistribution of global cloud fraction and RSDS in S20 and S50. Largest local cloud increase appears over North Africa with a Sahel-centered anomaly (up to >5% in S20), extending into Southern Europe and the southern Arabian Peninsula. Increased cloud also over India, North Asia, and Eastern Australia; decreased cloud over Central/South America, South Africa, Central & Eastern US, Central Asia, and NW China. - Strong seasonal signal: Responses are generally stronger in Northern Hemisphere summer (JJA) due to intensified West African Monsoon and associated land–atmosphere–vegetation feedbacks. Eastern Australia shows largest anomalies in DJF linked to Walker circulation changes. - RSDS changes largely follow cloud changes, with some regional deviations due to cloud properties (albedo, lifetime). Seasonal anomalies are robust and partially cancel in annual means. - Interannual RSDS variability decreases prominently over the tropical Pacific (linked to suppressed ENSO) but shows regionally contrasting responses elsewhere (e.g., Northern US ANN, Eastern Australia DJF, India JJA), implying regional processes dominate variability changes. - PV power generation potential (PVpot): In S05, impacts are small (e.g., seasonal <−3% in western Sahel; <+5% in South Africa). In S20 and S50, robust PVpot reductions occur in North Africa, the Middle East, most of Europe excluding Scandinavia, India, Eastern China, Japan, Eastern Australia, and the Southwestern US, especially in local summer; increases occur in Central and South America, the Caribbean, Central & Eastern US, Scandinavia, and South Africa. • Magnitudes: Annual PVpot change around −4% in sensitive regions (North Africa, Central Europe, India) for S20; JJA reductions exceed −8% over substantial parts of North Africa, and reach ±5% in India and the Northern US; ±3% in Europe, Southwestern US, and South Africa; remote regions seasonally reach about ±5%. • Aggregation: When averaged over continents or countries, mean PVpot changes are muted (typically within ±2%). - The magnitude of PVpot changes is comparable to those projected under strong climate change scenarios for solar energy (on the order of up to ±10% regionally).

The study demonstrates that large-scale PV deployment in the Sahara can modify atmospheric circulation sufficiently to alter cloudiness and RSDS globally, thus affecting solar generation potential. Mechanistically, reduced surface albedo over PV areas leads to localized warming and convection, surface convergence, and upper-tropospheric divergence over North Africa. This alters Atlantic SST gradients (Atlantic Niño-like), intensifies and shifts the Walker circulation, and strengthens the South Asian monsoon. In NH summer, baroclinic anomalies over North Africa and west-central Asia excite a circumglobal wavenumber-5 Rossby wave train, producing anticyclonic anomalies over East Asia, the North Pacific, and North America. These circulation changes suppress cloud formation in regions like Japan, Eastern China, and the Eastern US (mainly via mid- and low-cloud reductions due to moisture divergence and stability changes), while a Gill-type response induces descent over the North Atlantic and Amazon, reducing high cloud over Brazil and contributing to increased PV potential over Northern Europe via stationary Rossby wave links. Southwestern US shows increased high cloud linked to enhanced atmospheric rivers and coupling with Pacific fronts/eddies. Overall, these teleconnections explain the spatially heterogeneous PVpot impacts. The findings address the research question by quantifying global-scale consequences from a regional PV deployment and highlight potential inequalities in solar resource distribution, underscoring the need for coordinated planning across regions transitioning to solar power.

Large-scale PV solar farms covering 20% or more of the Sahara can significantly reshape atmospheric circulation, cloud cover, and RSDS across seasons and regions, leading to notable, though spatially heterogeneous, changes in PV power potential. While global/continental mean effects are modest, important regional and seasonal impacts (often up to several percent and exceeding −8% locally in summer) may disadvantage regions heavily promoting solar (e.g., Middle East, much of Europe, India, Eastern China, Japan, Eastern Australia, Southwestern US) and benefit others (Central/South America, the Caribbean, Central & Eastern US, Scandinavia, South Africa). As a pilot study using a single Earth system model and limited scenarios, the work highlights system-wide risks of deploying massive solar projects in drylands and the importance of international cooperation. Future research should employ multi-model ensembles, varied PV configurations (including realistic, variable albedo and efficiency), explicit treatment of dust and surface roughness effects, and expanded sensitivity experiments and regions to refine impact assessments and support equitable, secure solar energy transitions.

- Representation of PV properties is simplified via a fixed effective albedo (0.235), whereas real-world panel reflectivity and conversion efficiency vary with type, operation, and conditions; this may underestimate local surface heating effects when generating electricity, especially during daytime/summer. - Other PV-related surface parameter changes (e.g., surface roughness) are not fully represented, potentially affecting sensible heat fluxes and boundary-layer processes. - Dust processes are omitted; given their importance in the Sahara, responses may be underestimated. - Single-model approach (EC-Earth3-Veg-LR) and limited scenario set constrain robustness; multi-model, multi-scenario analyses are needed. - Results focus on quasi-equilibrium with fixed 1990 (and a 2010 sensitivity) forcings; real-world evolving GHGs/aerosols/land use may modulate impacts. - Aggregated impacts are small, but subregional variability and interannual variability changes imply potential local sensitivities not fully explored.