Horizontal wind profiling with Doppler lidars: long-term evaluation of the perpendicular vertical sweeps reconstruction method

Over the last 30 years, the demand for wind profile observations in the lower troposphere has rocketed, carried by weather agencies, airports and the wind energy industry. Doppler lidars are favoured for their compactness, easiness of operation, and versatility in the scanning strategy. Several methods have been developed to reconstruct the horizontal wind profile from the raw radial wind observations recorded in different directions. The most common is the Doppler Beam Swinging (DBS) technique, which is implemented in commercial lidars software. However, DBS leaves a blind zone near the ground that can damper the observation of very low-altitude phenomena like certain low-level jets. 
Another horizontal wind reconstruction method consists in combining observations from two vertical sweeps of the Range-Height Indicator (RHI) type recorded in perpendicular directions, by binning the data into horizontal layers. To our knowledge, this cross-RHI technique has only been used twice [1, 2] and applied to only a few tenth  of hours of lidar scans, so that this method still needs to be fully validated over a longer period and under more varied conditions.
In this study, the cross-RHI and DBS techniques were compared using observations recorded by two Doppler scanning lidars from the Leosphere/Vaisala company, installed at two contrasting sites in France: a flat coastal site (Dunkerque, North Sea coast) for four months, and an urban hilly site (Paris) for two months. Compared to the previous studies and to the DBS method, the cross-RHI technique was improved by adding filtering steps designed to remove range-folded echoes from middle-level clouds. In addition, the flow inclination on the hilly site was taken into account by tilting the wind binning layers and minimizing the total intra-layer variance. 
The horizontal wind speed values retrieved using both techniques were in very good agreement on both sites, with correlation coefficients ~0.92 in the first 200 m above the lidar. The regression slope was 0.93 and the intercept was below 0.4 m/s on both sites, drawn by a small share of points where the DBS grossly overestimated the wind speed due to range-folded echoes. This problem disappeared at higher altitudes, where the correlation coefficients exceeded 0.97, with slopes ~0.97 and intercepts lower than 0.1 m/s. In Dunkerque, where the DBS were averaged over 10 consecutive cycles, the horizontal wind direction difference was smaller than 5° (resp. 10°) for 61% (resp. 83%) of observations in the first 200 m above the lidar, and these numbers also improved with increasing altitude. Additionally, the cross-RHI technique proved to be more efficient to reconstruct the wind in pristine conditions yielding low lidar signal. 
This method’s ability to capture very low-altitude phenomena while providing turbulence information opens new perspectives for urban studies and wind farm site assessment. 

References
[1]    R. M. Banta et al., “Nocturnal Low-Level Jet Characteristics Over Kansas During Cases-99,” Bound.-Lay. Meteorol., 105(2), 221–252, 2002, doi: 10.1023/A:1019992330866.
[2]    T. A. Bonin et al., “Evaluation of turbulence measurement techniques from a single Doppler lidar,” Atmos. Meas. Tech., 10(8), 3021–3039, 2017, doi: 10.5194/amt-10-3021-2017.