Suppression of precipitation bias in wind velocities from continuous-wave Doppler lidars

<p>In moderate to heavy precipitation, raindrops may deteriorate the accuracy of Doppler lidar measurements of the line-of-sight wind velocity because their projected velocity in the beam direction differs greatly from that of air. Therefore, we propose a method for effectively suppressing the adverse effects of rain on velocity estimation by sampling the Doppler spectra faster than the time taken for a raindrop to transit through the beam. By using a special averaging procedure, we can suppress the strong rain signal by sampling the spectrum at <span class="inline-formula">3</span> kHz. A proof-of-concept field measurement campaign was performed on a moderately rainy day with a maximum rain intensity of <span class="inline-formula">4</span> <span class="inline-formula">mm h⁻¹</span> using three ground-based continuous-wave Doppler lidars at the Risø campus of the Technical University of Denmark. We demonstrate that the rain bias can effectively be removed by normalizing the noise-flattened 3 kHz sampled Doppler spectra with their peak values before they are averaged down to <span class="inline-formula">50</span> Hz prior to the determination of the speed. In comparison to the sonic anemometer measurements acquired at the same location, the wind velocity bias at <span class="inline-formula">50</span> Hz (<span class="inline-formula">20</span> ms) temporal resolution is reduced from up to <span class="inline-formula">−1.58</span> <span class="inline-formula">m s⁻¹</span> for the original raw lidar data to <span class="inline-formula">−0.18</span> <span class="inline-formula">m s⁻¹</span> for the normalized lidar data after suppressing strong rain signals. This reduction in the bias occurs during the minute with the highest amount of rain when the focus distance of the lidar is <span class="inline-formula">103.9</span> m and the corresponding probe length is <span class="inline-formula">9.8</span> m. With the smallest probe length, <span class="inline-formula">1.2</span> m, the rain-induced bias is only present at the period with the highest rain intensity and is also effectively eliminated with the procedure. Thus, the proposed method for reducing the impact of rain on continuous-wave Doppler lidar measurements of air velocity is promising and does not require much computational effort.</p>