Dual-lidar profilers for measuring atmospheric turbulence

Abstract. A dual-lidar system, consisting of WindCube v2.1 profilers with one lidar oriented horizontally at 45° relative to the other, was deployed to estimate along- and cross-wind variances, velocity spectra, and turbulence intensity (TI) using the "variance method". This approach computes second-order statistics directly from the line-of-sight (LOS) velocity variances to infer the full three-dimensional velocity variance components. It is benchmarked against the "traditional method" commonly used in the wind energy sector, which reconstructs instantaneous velocity components from single-lidar LOS measurements before deriving second-order statistics. Both methods are evaluated at a single altitude using a 30-day collocated dataset, with reference measurements from a sonic anemometer classified by atmospheric stability conditions. Two key performance metrics are considered: the mean relative bias error (MRBE) and the relative root mean square error (RRMSE), as defined by DNV (Det Norske Veritas). Spectral analysis of the velocity components shows that, while the traditional method more closely matches the reference spectra at low frequencies, it tends to overestimate spectral energy at higher frequencies. In contrast, the variance method typically underestimates spectral energy in the along-wind component and overestimates it in the crosswind component. Furthermore, linear regression analysis reveals that the variance method captures 90–97 % of the reference variances across all stability regimes, while the traditional method tends to overestimate—especially for cross-wind variance under unstable and neutral conditions (up to 132 %). Overall, the variance method yields lower MRBE and RRMSE values for both along- and cross-wind TI. Specifically, for along-wind TI, MRBE decreases from 10.1 % to 7.5 % and RRMSE from 19.5 % to 11.7 %. For cross-wind TI, MRBE is reduced from up to 17.6 % to 12.3 %, and RRMSE from 23.4 % to 18.2 %.