The wind resource of Northeast Brazil as a climate variable: decadal variability, regional model bias and projected change at turbine hub height

Wind is more than just a weather event; it is a climate variable that powers entire electrical grids. In the Northeast of Brazil, the wind is the second-largest source of electricity, generating enough energy to light millions of homes and drive the local economy. This region relies heavily on steady breezes that blow across the coast and over high plateaus, but the strength of these winds is not constant. It shifts over time, influenced by large-scale patterns in the ocean and atmosphere, such as the El Niño–Southern Oscillation, which warms and cools the Pacific Ocean, and temperature shifts in the tropical Atlantic. For engineers and planners, understanding whether these winds are getting stronger, weaker, or simply changing their rhythm is a matter of adaptation. If the wind resource shrinks, the energy supply becomes risky; if it grows, the region could become even more powerful. The critical question is whether the changes seen in recent years are a permanent shift caused by global warming or just a natural, temporary fluctuation that will eventually reverse.

A team of researchers set out to answer this question by looking at the wind exactly where modern turbines spin, rather than relying on old measurements taken near the ground. They gathered over a million hourly records from twelve tall metal towers, known as meteorological masts, located in three different wind farms across the states of Rio Grande do Norte and Ceará. These towers stand between 78 and 95 meters high, matching the height of the turbine hubs where the blades capture the most energy. The data covers a period from 2011 or 2015 up to 2025. The scientists then compared these real-world measurements against computer simulations of the future climate, using a regional model called Eta that was nested inside four different global climate models. These simulations projected what the wind might look like under different scenarios of greenhouse gas emissions, stretching out to the middle of the century. The goal was to see if the models could accurately predict the wind at turbine height and to determine if the wind was truly changing or just cycling through its natural phases.

The story the data tells is one of dramatic swings rather than a slow, steady decline. The researchers found that the wind was exceptionally strong from 2015 to 2017, a period that coincided with a very strong El Niño event. Then, starting in 2018, the winds dropped significantly and remained calm through 2024, a period that aligned with a rare sequence of three consecutive La Niña events and a warming phase in the tropical Atlantic. By 2025, the winds began to recover. This pattern was not a gradual, linear drop that would suggest a permanent loss of wind power. Instead, it was a sudden step down, a shift from one level of activity to another, followed by a return to higher levels. In the Rio Grande do Norte sites, this drop was statistically significant, representing a decrease of 5 to 12 percent per decade if one were to draw a straight line through the data. However, the researchers emphasize that this is a natural, decadal variability synchronized with the ocean, not a forced signal of climate change. The windiest years and the calmest years were simply different chapters in the same book, driven by the ocean's mood.

When the scientists tested the computer models against these real measurements, they found that the models were good at predicting the seasons but poor at predicting the specific years. The simulations correctly showed that the wind is strongest between August and November and weakest during the rainy season, but they consistently underestimated the actual speed of the wind, sometimes by as much as 36 percent. This underestimation was most severe during the rainy months, when the models failed to capture how the wind accelerates over the high plateaus. More importantly, the models could not predict the specific sequence of windy and calm years because they were not running on the same calendar as the real world. This led to a surprising discovery about how scientists try to connect model predictions to real-world locations. The researchers tested whether machine-learning tools, which are often used to "translate" model data into local forecasts, could fix this problem. They found that these tools failed completely when applied to future climate scenarios. Because the models and the real world were out of sync, the machine learning algorithms could not learn the true relationship between the two. Instead, they simply averaged out the data, transmitting only a tiny fraction of the projected change and effectively hiding the signal of future climate shifts.

To get a clearer picture of the future, the researchers used a different method that focused on the statistical distribution of the wind rather than trying to match year-by-year patterns. They adjusted the model data to correct for the known biases, particularly the underestimation during the rainy season. When they looked at the corrected projections for the period between 2025 and 2054, the results were surprisingly modest. The models suggested a slight increase in wind speed of about 3 to 5 percent on the high plateaus of Ceará, while the coastal sites in Rio Grande do Norte showed almost no change. These projected changes are small, especially when compared to the natural swings the region has already experienced. The difference between a windy year and a calm year in the recent past was far larger than the entire difference predicted between today and the middle of the century. Furthermore, the different computer models disagreed with each other more than they agreed on the specific scenario of emissions, meaning that the uncertainty in the models is currently larger than the signal of climate change itself.

The implications of these findings are profound for how energy planners should think about the future. The study suggests that the wind resource in Northeast Brazil is not in danger of disappearing or collapsing due to climate change in the coming decades. The region remains one of the best places in the world for wind energy. However, the natural variability of the wind is a much bigger factor than the slow march of global warming. A wind farm built today could face a decade of weaker winds simply due to natural ocean cycles, which would reduce its power output significantly, even if the long-term climate trend is stable. This means that energy estimates for new projects must account for these natural ups and downs, not just the average climate of the future. The researchers conclude that for the next thirty years, the most important factor for planning is not the climate change signal, but the ability to adapt to the natural rhythm of the ocean. The wind will continue to blow, but its strength will ebb and flow in ways that are larger and more immediate than the slow changes predicted by climate models.