Doppler lidar wind profiling: long-term assessment of the perpendicular vertical sweeps reconstruction method

Abstract. Doppler lidars have become a relatively common tool to measure wind profiles in the lower troposphere, as commercial instruments improved and became more accessible to answer the demand, notably from the wind energy industry. Most often, the Doppler Beam Swinging (DBS) technique is used to reconstruct the horizontal wind profile from the radial wind raw observations, because it is implemented in the software of commercial Doppler lidars. However, the DBS leaves a blind zone near the ground that makes it difficult to observe very low-altitude phenomena such as certain low-level jets, plus the wind profiles can be noisy if most of the lidar observation time is dedicated to other types of sweeps. However, the horizontal wind can also be reconstructed by combining observations from two vertical sweeps of the Range-Height Indicator (RHI) type, recorded in perpendicular directions. Radial wind observations are binned into horizontal layers, which allows to retrieve both the average wind and its variance, providing an estimation of the horizontal part of the turbulent kinetic energy (TKE). This work introduces a method to take into account the flow inclination over sloping terrains, as well as filter the range-folded echoes from high-altitude clouds. The horizontal wind speed and direction retrieved with this cross-RHI method were compared with standard DBS observations over six months of observations in total, recorded at two urban sites in France, one coastal flat city (Dunkerque) and one continental hilly megacity (Paris). The wind speed intercomparison below 200 m altitude yielded very good correlation coefficients around 0.85 on both sites, with a cross-RHI-vs-DBS fitting slope of 0.93. At higher altitudes, up to 3 km, results were even better as range-folded echoes from clouds disappeared from the DBS, with correlation coefficients and slopes both around 0.97 for altitudes above 1 km. The wind direction showed higher dispersion, though in Dunkerque, 83 % of points below 200 m had an error lower than 10°, and 62 % in Paris, where there was no averaging over several DBS cycles. Again, results improved with increasing altitude, demonstrating that the cross-RHI wind reconstruction method can be used to retrieve full-height wind profiles without limiting the sweeps to low-elevation beams. In addition, case studies highlighted the ability of the cross-RHI technique to operate with low aerosol loads, range-folded echoes, and convective conditions. The two-dimensional flow inclination angle could be optimized successfully for about 88 % of the RHI pairs on both sites. The most frequent tilt angle was around 0.6° in Dunkerque and 1.0° in Paris, with 95 % of the values below 8°, though the direction of the steepest slope could only be partially related with the orography. In Paris, the decrease in horizontal TKE permitted by the flow inclination correction was −5.2 % at the lidar level, but could reach −12 % around 1 km altitude, though these values were very sensitive to the altitude band used to optimize the tilt angle. The comparison with a near-ground ultrasonic anemometer in Dunkerque showed that the cross-RHI retrieval overestimated the horizontal TKE by about 13 %, even when accounting for the flow inclination, though this may result from the slightly higher altitude at which lidar data were recorded.