Is the wind really dying?

A weak wind season can quickly become a referendum on renewable energy. Turbines slow, wind generation falls below expectations, power prices rise, and headlines begin asking whether the wind itself is disappearing. Terms such as “wind drought” and “global stilling” are increasingly invoked as evidence that wind power is becoming less dependable just as grids are relying on it more heavily. 

But a difficult year for wind does not necessarily tell us much about the long-term wind resource. 

There is a real scientific question behind the debate. Many land-based weather stations recorded declining near-surface wind speeds over several decades during the late twentieth century, a phenomenon known as terrestrial stilling. But the popular version of the story often collapses several different questions into one: Was a particular month or year unusually calm? Is the long-term wind climate changing? And what does either mean for the reliability and economics of a power system? 

These questions operate on different time scales and require different evidence. Understanding the distinction is increasingly important as wind becomes a larger part of the global energy system. 

A quiet year is not a climate trend

The distinction between weather and climate sounds elementary, but it becomes surprisingly easy to lose when megawatt-hours and revenue are involved. A wind plant owner experiences the resource through monthly settlements, quarterly debt tests and annual budgets. A climate trend, by contrast, is estimated from decades of observations or model output. The commercial impact can be immediate even when the climatological meaning remains uncertain. 

Consider Europe in 2021. Copernicus found annual mean wind speeds as much as 10% below the 1991–2020 average across parts of northwestern and central Europe. In portions of the United Kingdom and Ireland, third-quarter wind speeds were about 20% below average, with some affected regions registering their lowest or second-lowest values in the ERA5 record beginning in 1979. 

It was an exceptional and economically important year. But it was not, by itself, proof that every subsequent year would be calmer. 

The following year illustrates the point. Copernicus reported that the 2022 annual average wind speed over European land was less than 0.1% below the 1991–2020 average—effectively normal for Europe as a whole—even though substantial regional differences remained. 

A continental average can hide local deficits, while a local deficit can be mistaken for a continental trend. Both errors arise when a dataset is asked to answer a question at the wrong scale. 

What does “terrestrial stilling” actually mean? 

The scientific literature uses terrestrial stilling to describe a broad decline in observed near-surface wind speed over land, particularly from the 1960s or 1980s through the first decade of the twenty-first century. A major review by McVicar and colleagues synthesized widespread declines in station records. Proposed causes included changes in atmospheric circulation, increasing surface roughness from vegetation growth and urbanization, and observational issues such as changing instruments, station exposure and land use around measurement sites. 

Then the signal changed. 

A 2019 study led by Zhenzhong Zeng found that the global terrestrial decline reversed around 2010 in its station dataset. The authors linked much of the decadal behavior to ocean–atmosphere variability and estimated a sizable rebound in potential wind energy during 2010–2017. More recent work continues to find recovery in some regions while also emphasizing discrepancies among station observations, reanalyses and climate models. 

This does not mean the stilling literature was wrong, nor that climate change is irrelevant to future wind resources. It means that “the wind is slowing everywhere” is too blunt a conclusion. 

The answer depends on the period selected, the height above ground, the region, whether the site is onshore or offshore, which observational network or reanalysis is used, and which part of the wind-speed distribution matters for a specific turbine fleet. 

Measuring the wind resource is harder than it looks 

There is another complication: the wind fleet itself is changing. 

Installed wind capacity has grown rapidly, turbines have become taller and more efficient, and projects have moved into new regions. This makes it difficult to distinguish changes in the underlying wind resource from changes in the generation fleet. Near-surface weather observations also do not necessarily represent conditions at turbine hub height. 

Resource assessments therefore need hub-height, energy-relevant data—not a casual extrapolation from an airport anemometer. 

One way to address this is to ask a counterfactual question: how would today’s installed wind footprint have performed under each historical hour of weather? Vaisala Xweather’s Powerup product does this by modeling every wind project in a market as the fleet exists in 2026, then driving that fixed fleet with hourly ERA5 winds back to approximately 1980.

By holding turbine locations and technology fixed, the analysis helps separate changes in the underlying wind resource from the effects of continuing capacity additions.

The long-term climate signal is regional and uncertain 

Climate change can alter wind resources by shifting pressure gradients, storm tracks, atmospheric stability, land–sea temperature contrasts and the frequency of weather regimes that produce persistent calm or strong winds. It can also change electricity demand and the performance of other resources at the same time. 

But projections of future wind power are generally more spatially heterogeneous and less robust than projections of temperature. 

The IPCC’s energy-sector assessment emphasizes regional differences and uncertainty. Some studies project decreases in wind-energy potential in parts of the Mediterranean and other regions, while changes elsewhere are smaller, mixed or model-dependent. A project-level conclusion therefore needs to come from an ensemble of models, bias-corrected against credible historical data and translated through the plant’s technology—not from a single global percentage. 

It is also important to distinguish a change in average resource from a change in risk. 

A small shift in annual mean wind might matter less than a rise in the probability of multi-day low-wind events during periods of high demand. For lenders and grid planners, the tails of the distribution may be more consequential than the mean. 

The relevant questions become more specific: Are low-output events lasting longer? Are they becoming more spatially correlated? Do they coincide with low solar output, high demand, transmission constraints or weak hydro conditions? 

A better question than “Is the wind dying?” 

Wind stilling is a legitimate research topic, and long-term climate shifts deserve serious attention in renewable investment and grid planning. But the evidence does not point to a simple story of a steadily weakening global wind resource. The observational record includes multi-decadal decline, partial reversal, strong regional differences and continuing uncertainty about the causes and future trajectories. 

A calm year is evidence of exposure, not necessarily evidence of a trend. A trend is evidence for updated assumptions, not a verdict on a technology. 

The more useful question is not whether the wind is dying, but what the best available evidence tells us about where, when and how the wind resource is changing.