Learn how WindGuard stumbled upon a surprising effect some remote sensing devices show when it comes to measuring wind speed in turbulent wind flow and how a different approach to evaluating LiDAR data can help to decrease uncertainty in your wind measurement campaign.

Wind measurements with LiDAR and SoDAR have become increasingly popular in the wind energy industry to perform wind measurements, especially in locations where traditional measurements with met masts and cup anemometers are difficult or even impossible. Since the official publication of standard IEC 61400-12-1, Ed.2 in March last year, remote sensing devices are increasingly accepted as an equal alternative to wind measurements with cup or ultrasonic anemometers.

The deviations occurred only under certain conditions. When high turbulence occurred, the LiDAR would measure a much higher wind speed than the cup anemometer. As the LiDAR – a Windcube V2 – had shown high accuracy at a prior calibration, we were puzzled as to why the sensitivity with regards to wind turbulence intensity suddenly increased by 10% in comparison to previous classification tests. In close cooperation with LEOSPHERE, the manufacturer of the LiDAR, we conducted a thorough investigation to find the cause of the deviations.

We soon became aware that the phenomenon was not specific to this model of LiDAR, but there rather seemed to be a more general problem at the root of this. The whole scenario reminded us of earlier discussions about nacelle LiDARs, where similar mistakes were observed. What if the same mistake that lead nacelle LiDARs to produce inaccurate results was causing a malfunction in our measurement with the ground-based LiDAR? A look into the way most LiDARs evaluate wind speed set things in motion. For historic reasons most ground-based wind LiDARs use scalar averaging as the algorithm to calculate wind speed.

As expected, the difference of the scalar and vector averaged wind speed calculated from Windcube measurement data proved unrealistically high under certain conditions. It was a multiple higher than the difference evaluated based on simultaneous measurements of a cup anemometer and vane. The sensitivity of the Windcube on turbulence intensity almost vanishes when comparing the vector averaged wind speed measured by the LiDAR to the scalar average of the cup anemometer. In addition, the analysed sensitivity of the wind speed measurements of the Windcube on the wind shear is reduced by a factor of about 2 by using vector averaging, which is likely caused by the correlation of wind shear and turbulence intensity. Also, the absolute agreement of cup anemometer measurements to the LiDAR measurements improves when using vector averaging.

The results indicate a true measurement error of the Windcube, caused by its monostatic character in combination with scalar averaging. This is the exact same phenomenon that had already been reported in the case of 2-beam monostatic nacelle LiDARs. As a conclusion, we identified scalar averaging as a key driver of the relatively high sensitivities of the wind speed measurement by monostatic RSDs on environmental variables. By the implementation of vector averaging, the measurement uncertainty due to the sensitivities is reduced significantly. The results strongly suggest that the classical scalar averaging approach used in most of today’s commercially available monostatic ground-based LiDARs and SoDARs should be replaced by vector averaging.