Solar + Wind + Battery Atlas

How it works

The Goal: Can solar, wind and battery storage, on their own, deliver constant round-the-clock "baseload" power 24/7/365, and at what cost? This tool tests that question at thousands of locations worldwide. Set wind to zero and you get the original solar-and-battery atlas, unchanged.

The model

Everything is sized per 1 MW of constant, round-the-clock demand, so results scale to any load. You choose how much solar and onshore or offshore wind (MW per MW of demand) and battery (MWh per MW of demand) to build.
We start from ten years of hourly weather data (2015–2024) for each location: NASA POWER for solar and ERA5 for wind.
For each hour, sunlight and temperature go through a tilted-panel solar model: fixed tilt facing the equator, calibrated to the Global Solar Atlas's annual yield, with a DC-to-AC ratio of 1.3 that clips the peaks.
An hour-by-hour simulation then dispatches that energy: solar and wind meet demand directly as they generate; surplus charges the battery (accounting for round-trip losses); the battery discharges to cover shortfalls; whatever is still unmet is recorded as a gap.
Averaged over the ten years, this gives each location's Capacity Factor; tooltips add the worst year for the build on screen. It is pre-computed offline for 3,216 solar, wind and battery combinations per location and loaded into your browser a wind level at a time, so panning the map is instant. The map covers 4,465 regions; Antarctica and Greenland are left out.
Wind is sited, not sampled. Each region's wind level is the mean net capacity factor of the windiest 20% of its eligible land, scored against the Global Wind Atlas (at least 500 km², or all of it if less). Offshore follows the same rule at sea. Every region with eligible land or sea gets its wind, however weak, and the cost search decides whether it is worth building.
Calibrated to new wind farms. Levels are scaled to IRENA's capacity factors for new wind farms: ×0.879 onshore in Europe, ×0.777 offshore everywhere (fitted on four countries) and 1.0 onshore elsewhere.
Hourly shape from ERA5, level from the atlas. ERA5 100 m winds at the chosen site give the hour-by-hour pattern for all ten years, scaled by one factor so that their ten-year mean equals that level. Some regions, mostly mountains, cannot reach it; see Wind siting below.
LCOE is calculated live in your browser from those capacity factors and your own cost assumptions: change the costs, financing or year and results update immediately.

Key Metrics

Capacity Factor (CF): the share of a year's constant load that the build supplies: energy served divided by energy demanded. A CF of 90% means solar, wind and batteries supply 90% of the energy a flat load would use over the year, averaged over 2015–2024. It is not the share of hours in which the load is fully met.
LCOE (Levelised Cost of Energy): the average lifetime cost per MWh, accounting for up-front build cost (CAPEX), operating costs (OPEX), financing (WACC), solar and wind output degradation (weighted by each one's share of the energy), and equipment lifetimes (30 years for solar and wind, 20 for batteries). You can switch between global cost assumptions and local, region-specific costs, roll costs forward to a future year, and optionally add a diesel or gas backup generator to firm up the last few per cent of supply.

Wind siting

Eligible land excludes built-up areas, IUCN protected areas, water plus a 2 km buffer and severe terrain (World Bank/ESMAP masks), every designated or inscribed protected area in the WDPA (October 2026; offshore, every marine protected area) and every GHSL settlement cell at least 10% built, with no further setback. Land below sea level counts as land where Natural Earth and the ESMAP water layer agree (polders, the Caspian Depression, Qattara); lakes carry no offshore wind, and the Caspian's depth is measured from its own surface. Forest and cropland are not excluded. 4,383 of the 4,465 regions have eligible land and 1,199 have eligible sea; 647 onshore and 140 offshore of them are weak, their windiest fifth below 15% before calibration. Only regions with no eligible land or sea are drawn hatched.
Calibration: the European factor applies by location, to regions in Europe and Russia west of 60°E. Against IRENA's new-build capacity factors, the onshore level falls within ±10% in 70% of the 27 countries with data; Russia, left out of the fit, comes to 1.00. The offshore factor matches its four countries by construction, so it is a fit, not a test; the current check leaves China out, because IRENA's fleet figure for it is implausibly low. Known limitation: onshore capacity factors in Latin America are probably understated. Brazil's level, for example, reaches 0.73 of IRENA's new-build figure.
The scaling cap: the factor on ERA5 wind speed is capped at 1.8, so 700 of the 4,383 onshore regions – mostly mountains – simulate below their level, 304 of them below 15% and 387 below 20% (Bogotá: 18% level, 10% simulated). Offshore, 9 of 1,199 fall short, one below 15%. This is the standard correction: ERA5 alone understates the level by about 17%.
Against the centre point: the simulated level runs a median 4 capacity-factor points above the region's centre, and in more than a quarter of regions the centre would run below 15%.
Offshore is offered where a region has eligible sea at least 10 km from shore, screened for fixed-bottom depths of up to 60 m, and costed at the mean depth and distance from shore of the windiest fifth it is simulated on. Floating area further out is recorded but not simulated.

Notes on the controls

Wind sliders: onshore and offshore are separate builds. Combined builds are simulated for onshore 1–4 + offshore 1–3 MW (5 at most in total); otherwise raising one sets the other to 0. Wind off reproduces the original solar and battery atlas.
Include offshore wind: each region uses onshore or offshore wind, whichever wins the search (the cheaper build when the target CF is met). Untick to use onshore only.
Technologies: switch solar, batteries, onshore or offshore wind off and the cost search uses only builds without them; uptimes come from the same simulations. Offshore is the same switch as Include offshore wind. Wind share and wind saving need wind. The saving is measured against the cheapest build with the same technologies minus wind; batteries alone deliver nothing, so with solar off the target is met only with wind.
Builds beyond the sliders (solar above 20 MW, battery above 36 MWh per MW of load) are searched only where the sliders cannot reach 99.9%. Their battery is interpolated between simulated sizes and always rounded up, and their ten-year weather record repeats – the battery starts the record with what it held at its end – while every slider build starts with an empty battery. Seasonal battery storage at this scale is a thought experiment, not a proposal. In 'cost within a budget' mode they are used only at simulated sizes, so the two modes can differ slightly for the same region.
Firm backup: covers any unmet load after solar, wind and batteries with a 1 MW dispatchable generator. Each region uses whichever is cheaper: diesel, or OCGT gas where a regional wholesale price exists (IGU 2024). The target CF still applies to solar, wind and batteries; the generator fills the remaining load to 100%.
Cost decline timeline: applies multipliers to the solar, battery and wind capex values (global mode); local mode interpolates each region's capex, and wind O&M, by year. The cost of capital does not change with the year.
Sample Days: daily totals for 2024 only (NASA POWER solar, ERA5 wind). A day can look fully met and still run short at night or in still hours, so these shares are not the uptime the other views report. The battery starts empty on 1 January. Each region is coloured by the source that supplied most of the day's load.
Potential, solar: ground-mount (40 MW/km² on suitable land) + rooftop (100 MW/km² on building footprints, Joshi et al. 2024 SSP2 2020). Technical excludes rugged terrain, urban/industrial areas, forests and very remote zones; Policy adds regulatory limits (cropland protection, conservation areas). Siting constraints apply to solar only.
Potential, wind: technical potential, the capacity that could physically fit, not what is likely to be built; very large build-outs would also slow the wind itself, so real limits are lower. Generation = capacity × the cell's capacity factor × 8,760 hours.
Potential, onshore: 5 MW/km² (NREL, Lopez et al. 2012; low case 3 MW/km², the observed average across whole US wind projects, Denholm et al. 2009) on eligible land that reaches at least 15% net capacity factor before calibration; weaker land still gets its wind in the simulation but does not count as potential. Eligible land is as in Wind siting; forest and cropland are not excluded, so it is broader than NREL's global estimate. The capacity factor is the cell's calibrated level (the windiest 20% of its eligible land), scaled down where the hourly wind profile can't reach it – mostly mountains. So in most cells generation leans high: globally about 18% above what the median site would give, calibrated the same way (1,157 against 982 PWh/yr).
Potential, offshore: the World Bank/ESMAP convention (World Bank Group / ESMAP, Global Offshore Wind Technical Potential, 2021) – 3 MW/km² at 7–8 m/s and 4 MW/km² above 8 m/s, within 200 km of shore; fixed-bottom up to 50 m deep, floating 50–1,000 m (the atlas's simulation and costing split at 60 m instead). Floating is not costed in the atlas. US floating is shown at 1,298 GW against ESMAP's 2,787 GW (the rest is missing from ESMAP's published map data). Nothing is excluded for shipping, fishing or military use, nor for sea ice, which is why Arctic shelves look large. Generation uses the atlas's offshore capacity factor for the cell; cells without eligible sea show no estimate. The atlas's 68 PWh a year fixed-bottom and 149 PWh floating (floating in the data files only) are about half the IEA's technical potential of about 420 PWh a year (Offshore Wind Outlook 2019), with the same split: about 70% in deep water. Sea is assigned to the nearest cell, not by country, so don't total cells by country. At low zoom about 2% of offshore capacity sits in a neighbouring drawn cell, and about 41% of coastal cells differ by more than 5% from the drawn cell, some by tens of per cent.
Potential, all: solar + onshore wind + fixed-bottom offshore wind; floating is left out. Gaps count as zero: offshore cells with no estimate, and solar outside its data latitudes. Solar and onshore wind can compete for the same land, so the sum overstates what both could use.

Data Sources

Solar & Weather Data: NASA POWER (Prediction of Worldwide Energy Resources), hourly 2015–2024, run through a tilted-panel model and calibrated to the Global Solar Atlas (World Bank Group / Solargis). Limits: the calibration factor is capped at 1.10, which holds 22 regions below it; in the Atacama they sit about 10% below the Global Solar Atlas. Above about 60°N, where it has no data, a region takes the median factor of its 8 nearest calibrated regions (0.3% of people, 3.9% of demand, mostly in Russia and the Nordics). Snow losses are not modelled.

Wind resource: Global Wind Atlas 4 (DTU Wind Energy and the World Bank, 2025, CC BY 4.0) for siting and long-run capacity factor; ERA5 100 m winds (Copernicus Climate Change Service, via the analysis-ready copy on Google Cloud) for the hourly shape, 2015–2024. Land eligibility uses the World Bank / ESMAP Global Photovoltaic Power Potential masking layers (built-up areas, protected areas, water and severe terrain), the World Database on Protected Areas (UNEP-WCMC and IUCN, October 2026) and the Global Human Settlement Layer (JRC, GHS-BUILT-S 2020); bathymetry is ETOPO 2022 (NOAA). Levels are calibrated to new-build capacity factors from IRENA, Renewable Energy Statistics 2026 and Renewable Power Generation Costs in 2025. Method after Peña-Sánchez et al., Nature Communications, 2026.

Wind costs: IRENA, Renewable Power Generation Costs in 2025. Global mode starts from IRENA's global averages: onshore $1,085/kW (the 2023–2025 mean), falling to IRENA's 2035 projection and on to 2050 at the same rate; offshore $2,931/kW, falling along the Danish Energy Agency's fixed-bottom path (DEA Technology Catalogue). Local mode gives each cell its country's installed cost. Onshore, that is IRENA's 2023–2025 mean where IRENA lists the country (figure 2.4), otherwise its 2025 regional figure (table 2.1). Saudi Arabia, the UAE, Qatar, Oman, Kuwait, Bahrain, Turkmenistan, Kyrgyzstan and Tajikistan have no IRENA value; they take the mean of IRENA's 2025 values for Egypt, Uzbekistan and Kazakhstan ($1,377/kW), markets supplied largely by Chinese manufacturers, instead of the regional $1,713. Onshore capex follows IRENA's projected decline to 2035 and on to 2050 at the same rate; offshore follows DEA's paths. Offshore capex is adjusted to the depth and distance from shore of the sea the cell's capacity factor comes from (the area-weighted mean over its windiest fifth), using DEA's cost relationships, with HVDC beyond 100 km; countries without an offshore market take IRENA's global average adjusted the same way, and other Asian countries, Taiwan among them, IRENA's Asian average excluding China (figure 4.4). Offshore capex is never below 1.5 times the cell's onshore capex, a conservative floor: IRENA's global offshore cost is about 2.7 times onshore. It binds in 254 cells, mostly in Asia, and sets all of Japan ($4,217/kW), which has no IRENA offshore value. All of that sea is at most 60 m deep, so offshore is costed as fixed-bottom. Wind O&M comes from IRENA's LCOE assumptions – by country where IRENA lists one (onshore table A3, offshore table A6), otherwise its OECD or non-OECD group, which covers about 65% of cells onshore and 87% offshore – and follows DEA's O&M trajectory to 2035 and 2050 (2050 at about 0.90 of 2024 onshore and 0.64 for fixed-bottom offshore). O&M therefore falls more slowly than capex. Global mode's onshore capex is IRENA's global weighted average, which Chinese installations weight heavily: a China-weighted world average, not the cost of a typical European or North American project. Offshore costs include the export grid, so they sit above tender prices where the grid operator pays for it (the Netherlands, Germany, Denmark, Belgium and France). UK and Irish auction prices sit above the atlas partly because they carry network charges (high in Scotland), and the UK's Contracts for Difference also carry balancing costs and a 20-year term. Both modes treat their cost data (IRENA's 2025 figures) as 2026 costs, the atlas's fixed cost year, and decline from there, so the two modes start from the same year; the legend states the cost year. Local mode's cost of capital is held at 2025 rates across the timeline. Wind is assumed to last 30 years (NREL ATB; Danish Energy Agency); IRENA's own LCOE convention is 25 years. At 25 years, adding wind lowers the cost of 90% uptime for places that use 79% of the world's electricity and for 49% of its people, against 81% and 52% at 30 years.

Population Data: NASA Socioeconomic Data and Applications Center (SEDAC), Gridded Population of the World (GPW), v4

Electricity Consumption Data: Chen et al. (2022), Scientific Data 9, 202. https://doi.org/10.6084/m9.figshare.17004523

Power Plant Data: Global Energy Monitor, Global Coal Plant Tracker, Global Oil & Gas Plant Tracker, Global Bioenergy Plant Tracker, Global Nuclear Plant Tracker, December 2024 release. Data available under Creative Commons Attribution 4.0 International License.

Solar Potential (PVOUT): Global Solar Atlas (World Bank Group / Solargis). Used in the Potential Map to estimate realistic annual generation per area.

Wind Potential: onshore from the atlas's own wind siting, at 5 MW/km² (low case 3 MW/km²) on eligible land that reaches at least 15% net capacity factor before calibration; offshore from World Bank Group / ESMAP, Global Offshore Wind Technical Potential (2021), CC BY 4.0, reallocated to the atlas's cells. Method in Notes on the controls.

Grid Reliability & Electricity Access: High Resolution Electricity Access (HREA), University of Michigan; Min et al. (2024), Joule 8, 1–17. https://doi.org/10.1016/j.joule.2024.05.001. Data: hrea.isr.umich.edu/data.html. Used in the Supply-Demand Matching demand layers.

Cost of Capital (WACC): Country-level WACC from Hatton et al. (2025), Sci Data 12, 1912. https://doi.org/10.1038/s41597-025-06177-0. Local mode uses the v3 dataset (Zenodo, 2026; https://doi.org/10.5281/zenodo.18662192) for solar and wind: each Voronoi cell takes its country's 2025 rate, or a neighbouring country's where the dataset has none, applied as published, not deflated, and held flat over the timeline. Wind has separate onshore, offshore and floating rates per cell. Global mode prices wind at the WACC input plus a premium: by default +0.0 points onshore, since onshore wind equals solar PV in the dataset's 2025 values, and +0.92 points offshore, the median offshore-over-onshore gap across 174 countries (v3, 2026).

Local CAPEX Assumptions: Regional solar and battery CAPEX assumptions from IRENA/IEA/BNEF; Oceania and Saudi Arabia from IRENA's 2024 installed costs (Renewable Power Generation Costs in 2025), Saudi Arabia's also for the other Gulf states, and Korea's for Taiwan. Local mode maps each Voronoi cell to a region and interpolates 2026/2035/2050 values by year; wind capex and O&M are interpolated the same way (see Wind costs). The 66 cells missing from the local cost files use the global inputs, including the global WACC and wind premiums.

Wind Output Degradation: 0.2%/yr onshore and offshore by default, after Hamilton et al., Joule 4 (2020), who find 0.17%/yr for newer US wind plants (https://doi.org/10.1016/j.joule.2020.04.005); older fleets in the US, Sweden and Germany lose 0.5–0.6%/yr. We found no peer-reviewed offshore-specific estimate, so offshore takes the onshore figure.

Diesel Prices: Retail diesel pump prices (World Bank / GlobalPetrolPrices), clamped to a global unsubsidised floor. Used for the optional diesel-backup comparison.

Gas Prices: Regional wholesale gas prices (IGU Wholesale Gas Price Survey 2024); regions with no gas market default to diesel. Local mode uses these and per-country pump prices for diesel (World Bank / GlobalPetrolPrices 2025); Global uses the inputs in the panel.

Credits

Created by the Electrotech Revolution team, 2026.

Population Analysis

Demand & Supply by Latitude

Population by Latitude · Capacity Factor by Latitude

Demand (top axis) Supply (bottom axis)

Demand vs Supply Distribution

Daily energy balance

Select a location...

Subset Distribution

Value

--

System Config

LCOE Parameters

Technology

Siting Constraints

Display

Demand

Supply

CF Parameters