Global potential for wind-generated electricity

Forty times. That's the number that grabs your attention when you first encounter this study. Forty times the world's entire current electricity consumption is available in the wind. Your first instinct is to reach for the asterisk — surely that disappears once you add real-world constraints, exclude forests and cities, and the frozen ground, and only count turbines that actually run at a meaningful fraction of their rated power. Lu, McElroy, and Kiviluoma conducted that very audit. The number held. Before discussing the results, it's important to understand why this study was needed at all. Previous estimates of global wind potential varied dramatically — not by ten or twenty percent, but by orders of magnitude — because researchers started from different data, different turbines, and different assumptions about usable land. Archer and Jacobson, working from around eight thousand surface meteorological stations, concluded that a fifth of global wind potential could supply roughly fourteen terawatts, about seven times the then-current global electricity consumption. That's a huge number, but their station network was heavily concentrated in the United States, Europe, and Southeast Asia, leaving vast regions effectively uncharacterized. What the field needed was a globally consistent baseline, built from a data source that didn't have gaps. That's where the GEOS-5 Data Assimilation System comes in. Think of it not as a weather forecast but as a retrospective best estimate of the atmosphere — a physics-based model that continuously ingests observations from surface stations, weather balloons, aircraft, ships, buoys, dropsondes, and satellites, then adjusts its internal state to match all of them simultaneously. The result is a three-dimensional wind field covering the entire globe, updated every six hours, on a grid roughly sixty-seven kilometers by fifty kilometers, with seventy-two vertical layers. The three lowest layer centers sit at about seventy-one, two hundred and one, and three hundred and thirty-two meters above the surface. Turbine hub height is one hundred meters, so the authors fit a cubic spline through those three lowest layer midpoints to interpolate wind speed at the hub. To check how much that interpolation matters, they also computed power by averaging across the layers themselves — and found the results differed by less than fifteen percent onshore and less than seven percent offshore. That's a meaningful uncertainty to consider, but it's quantified, not hidden. Once you have wind speed at hub height, you need to translate it into electricity. The key concept is the power curve — the turbine manufacturer's specification of how much power is generated at each wind speed. At low speeds, output rises roughly with the cube of wind speed. It flattens toward a constant rated power at moderate speeds, then drops to zero above a cut-out speed when the blades feather to protect the machine. The study uses General Electric power curves for a two-point-five megawatt land-based turbine and a three-point-six megawatt offshore machine, adjusted for actual air density at hub height. There's also the capacity factor — the ratio of actual power delivered to the maximum rated power. Lu and colleagues only count locations where capacity factors exceed twenty percent. For reference, turbines installed in the United States in 2004 and 2005 averaged close to thirty-six percent. So twenty percent is a conservative floor, not an optimistic ceiling. One more layer of realism: turbines in a wind farm steal energy from each other. Downstream turbines are affected by the wake of upstream ones, experiencing slower, more turbulent air. To keep wake loss below twenty percent, the study requires downstream spacing greater than seven rotor diameters and cross-wind spacing greater than four. For the two-point-five megawatt land turbine, with a one-hundred-meter rotor, that calculates to roughly 0.28 square kilometers per turbine. Now, consider where turbines can actually go. The study used Moderate Resolution Imaging Spectroradiometer land-cover data at one-kilometer resolution to exclude forests — where surface roughness hinders wind speeds at practical hub heights — as well as urban areas, permanent ice and snow, and inland water. What remains is open, ice-free, nonurban land. Offshore, the cutoffs are waters shallower than two hundred meters and within fifty nautical miles, or ninety-two point six kilometers, of the nearest coastline. That offshore boundary is divided into three depth bands — zero to twenty meters, twenty to fifty, and fifty to two hundred — because shallow water is cheaper to develop than deep water. The study also acknowledges that shipping lanes, fishing grounds, wildlife reserves, and aesthetic objections all limit where offshore turbines can realistically go, although it assumes one hundred percent of the available near-shore area within those depth and distance limits is usable. After all those exclusions, here are the numbers that emerged. A globally distributed network of land-based two-point-five megawatt turbines, operating at just twenty percent capacity factor, could supply more than forty times current worldwide electricity consumption — and more than five times total global energy use in all forms, not just electricity. If you raise the capacity factor threshold, the picture sharpens further: at fifty-three percent capacity factor, wind alone could meet total current global electricity demand. At thirty-six percent — the actual average for recent US installations — land wind would match total current global energy consumption across all sectors. Zoom into the United States, and the concentration of resource becomes vivid. The central plain states — the open, flat, wind-swept corridor running from Texas through the Dakotas — could accommodate as much as sixteen times total current US electricity demand on their own. Taken nationally, US onshore potential is about twenty-three times current US electricity consumption, with roughly eighty-four percent of that available on land. The obvious concern is variability. Wind is intermittent. When it drops in one region, does the whole system fail? Lu and colleagues addressed this by computing six-hour averaged potential supplies for three US regions — Montana, Minnesota, and Texas — and testing how correlated they are across seasons. In winter, the answer is: barely at all. Correlation coefficients between those regions fall below 0.07 from October through March. In summer, correlations rise — ranging from 0.28 to 0.37 — reducing the smoothing benefit. So a geographically dispersed network acts best as a reliability hedge in winter, which happens to be when US wind potential peaks. Onshore potential in January exceeds August by a factor of two point five; offshore, the ratio is two point nine. The seasonal mismatch between wind supply and electricity demand isn't perfect, but it's not catastrophic either — and the paper notes that surplus winter wind could be absorbed by plug-in vehicles or converted to hydrogen. What the study doesn't do is assume these numbers automatically translate into a working grid. Significant transmission expansion is required. Lu and colleagues cite estimates from the Electric Reliability Council of Texas suggesting that adding up to four point six gigawatts of Texas wind capacity would necessitate transmission investment of about one hundred eighty dollars per kilowatt, roughly ten percent of the wind installation capital cost itself. That's a solvable problem, not a showstopper, but it's a real cost. There are also honest limits on the analysis itself. The GEOS-5's horizontal resolution of roughly sixty-seven by fifty kilometers means sub-grid variations from topography and land cover are invisible to the model. A comparison with a high-resolution wind atlas for Minnesota found that atlas wind speeds were higher than GEOS-5 estimates across every land cover class — which indicates the study's land-based numbers likely underestimate true potential in complex terrain. The offshore estimates are probably closer, since the ocean surface is smoother and better resolved. The paper also highlights a physical limit at the extreme end of deployment: supplying all current US electricity from wind would introduce an atmospheric friction sink equivalent to about six percent of the total surface-friction sink over the contiguous United States. Large-scale extraction could modestly impact atmospheric circulation and meridional heat transport. Global mean surface temperature wouldn't change significantly, but it's not negligible. What Lu, McElroy, and Kiviluoma really delivered with this paper is a methodological baseline — globally consistent, built on a data assimilation product rather than sparse station networks, specific enough to support regional planning. The resource question, as they put it, is settled: there is enough wind. Enough to power the world many times over, even after you eliminate the forests, the cities, the ice, and the shallow ocean beyond fifty nautical miles. The remaining questions are about transmission, storage, economics, and political will. Those are hard questions. But they're not the same as asking whether the wind blows. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.