The study, published in the journal Wind Energy Science, was led by Nikolas Angelou of the Technical University of Denmark and Camille Dubreuil-Boisclair of Equinor. Their instrument of choice was not a towering meteorological mast, which is prohibitively expensive to install in deep water, but a nacelle-mounted Doppler wind lidar bolted directly onto the turbine itself. The device, a Vaisala Wind Iris, fires four laser beams upwind at azimuth angles of plus and minus 15 degrees and tilt angles of plus and minus 5 degrees, sampling radial wind speeds at ten distances ranging from 50 to 400 meters ahead of the rotor. Because the optical head follows the yaw of the nacelle, the lidar keeps its beams pointed into the wind almost continuously, maximizing data availability in a way that fixed masts and floating buoys cannot easily match.

The turbine under scrutiny, designated HS4, is one of five Siemens Gamesa SWT-6.0-154 machines at Hywind Scotland, the first commercial floating wind farm, operating since 2017 off the east coast of Scotland. Each unit has a hub height of 98.6 meters and a rotor diameter of 154 meters, mounted on a ballasted spar buoy. Floating turbines move in six degrees of freedom, three rotational and three translational, and at Hywind Scotland the dominant motion is pitch: the nacelle tilts backward by up to 7 degrees at wind speeds between cut-in and rated power, settling to about 2 degrees above rated. That tilt matters enormously for the lidar, because it warps the measurement geometry. When the turbine pitches to 5 degrees, two of the four beams become nearly horizontal while the other two sweep across the upper part of the rotor, and a motion-correction procedure using a motion reference unit on the nacelle is required to untangle the true wind from the swinging platform.

To make sense of the raw laser returns, the researchers built a mathematical model that expresses the four line-of-sight velocities as a function of the horizontal wind components at hub height, their vertical gradients, and an induction factor describing how the turbine’s own presence slows the air approaching the rotor. The model assumes that shear and veer, the change of wind speed and direction with height, remain constant across the upper rotor, and that the inflow is horizontally homogeneous. For the vast majority of cases, 88.5 percent of the dataset, this simple parameterization worked well, reproducing the observed radial speeds with root mean square errors below 0.2 meters per second. The lidar data even captured the turbine’s induction zone clearly: at around one rotor diameter upwind, radial speeds dropped noticeably, corresponding to induction factors of 0.38 to 0.41, consistent with the velocity deficit a turbine generates when operating below rated power.

But in 11 percent of the 10-minute periods, the model failed spectacularly. The culprit was the low-level jet, a river of fast air that forms in the lowest few hundred meters of the atmosphere, often under stable stratification when the sea surface is cool relative to the air above it. Within these jets, wind speed can peak at a certain altitude and then fall off above, producing wind speed inversions that no linear shear assumption can capture. Some profiles even displayed two inflection points, a maximum and a minimum, between 90 and 180 meters, forming a distinct core in the wind field. The researchers found that a local maximum in the vertical profile could be detected in 10 percent of all cases, with inversion heights typically clustered around 130 meters, squarely inside the rotor-swept area, and wind speed differences across the inversion usually between 0.25 and 0.5 meters per second, though in a tenth of cases exceeding 2 meters per second.

The climatology that emerged from six months of usable data, drawn from January 2019, autumn 2019, and summer 2020, is striking. Negative shear accounted for 22 percent of the examined periods, concentrated in the months from June to September, a figure consistent with earlier rawinsonde studies over the North Sea. These events were not tied to a single wind direction sector; in June and August they appeared across the entire 90 to 270 degree range analyzed, and they occurred at all times of day. Most negative-shear and inversion episodes lasted about an hour, but eight periods persisted for 700 to 900 minutes, meaning the turbine operated for more than half a day under atmospheric conditions that standard power performance methods simply do not anticipate. A detailed 12-hour case study on 26 June 2020, with hub-height winds swinging from 6 to 14 meters per second, showed deviations from nominal power reaching up to 50 percent when referenced against the power curve of an equivalent fixed-bottom turbine.

To quantify the effect systematically, the team performed a formal power curve verification using three different estimates of the reference wind speed: a wind field reconstruction from the farthest range gates at 2.6 rotor diameters, a reconstruction using all available range gates, and the rotor-equivalent wind speed recommended in the IEC 61400-12-1 standard. The results were unambiguous. Below rated wind speed, power production differed by 5 to 10 percent between periods with negative shear and periods with positive shear, depending on which wind speed definition was used. Under negative shear, output fell by up to 18 percent relative to the reference power curve, and the scatter in the measured power, expressed as the standard deviation of mean power in each wind speed bin, was nearly twice as large as in positive-shear conditions. Part of the explanation lies in veer: negative shear was usually accompanied by negative veer, a clockwise-with-height rotation of the wind that further reduces the energy available to the rotating blades.

The implications ripple well beyond one Scottish turbine. Power curve verification is the contractual backbone of the wind industry, the procedure by which manufacturers prove their machines deliver the energy they promise. The current IEC standards were written for turbines on flat or complex terrain and do not account for platform motion, nor for wind profiles that deviate from the textbook logarithmic law. Yet this study shows such deviations occur in a third of offshore cases during summer and autumn. The authors also note a practical limitation: because the lidar only probed the upper half of the rotor, the exact altitude of inversions in negative-shear cases could not be pinned down, which prevented a full rotor-equivalent wind speed calculation for those conditions. Temperature gradient measurements were likewise unavailable, so atmospheric stability could not be classified directly, leaving open the question of whether the observed jets were driven by stable stratification as models suggest.

The remedy, the researchers argue, is a measurement strategy that captures wind across the entire rotor-swept area, not just at hub height. Nacelle-mounted lidars with more beams or more range gates, combined with induction-zone models to correct for the turbine’s own influence on the approaching flow, could provide the full three-dimensional picture that floating wind demands. As floating offshore wind expands into deeper waters worldwide, and as rotors grow ever taller into the layer where low-level jets live, the message from the North Sea is clear: the wind a floating turbine actually harvests is far stranger, and far more structured, than the smooth curve on the datasheet, and only by measuring it honestly can the industry certify, and ultimately improve, the performance of its machines.

Subject of Research: Vertical wind shear and low-level jet effects on floating offshore wind turbine power production

Article Title: Offshore wind profile characteristics and their impact on floating wind turbine power production

Article References: Angelou, N., & Dubreuil-Boisclair, C. (2026). Offshore wind profile characteristics and their impact on floating wind turbine power production. Wind Energy Science, 11(9), 3745-3762. https://doi.org/10.5194/wes-11-3745-2026

Image Credits: AI Generated

DOI: 10.5194/wes-11-3745-2026

Keywords: floating offshore wind, wind shear, low-level jet, nacelle lidar, power curve verification, Hywind Scotland, North Sea, wind speed inversion, rotor-equivalent wind speed, atmospheric boundary layer, wind energy, IEC standards