Validation of the Aeolus L2B wind product with airborne wind lidar measurements in the polar North Atlantic region and in the tropics

Abstract

Abstract. During the first three years of European Space Agency’s Aeolus mission, the German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt, DLR) performed four airborne campaigns deploying two different Doppler wind lidars (DWL) on-board the DLR Falcon aircraft, aiming to validate the quality of the recent Aeolus Level 2B (L2B) wind data product (processor baseline 11 and 12). The first two campaigns, WindVal III (Nov/Dec 2018) and AVATAR-E (Aeolus Validation Through Airborne Lidars in Europe, May/Jun 2019) were conducted in Europe and provided first insights in the data quality at the beginning of the mission phase. The two later campaigns, AVATAR-I (Aeolus Validation Through Airborne Lidars in Iceland) and AVATAR-T (Aeolus Validation Through Airborne Lidars in the Tropics), were performed in regions of particular interest for the Aeolus validation: AVATAR-I was conducted from Keflavik, Iceland between 9 September and 1 October 2019 to sample the high wind speeds in the vicinity of the polar jet stream. AVATAR-T was carried out from Sal, Cape Verde between 6 September and 28 September 2021 to measure winds in the Saharan dust-laden African easterly jet. Altogether, 10 Aeolus underflights were performed during AVATAR-I and 11 underflights during AVATAR-T, covering about 8000 km and 11000 km along the Aeolus measurement track, respectively. Based on these collocated measurements, statistical comparisons of Aeolus data with the reference lidar (2-µm DWL) as well as with in-situ measurements by the Falcon were performed to determine the systematic and random error of Rayleigh-clear and Mie-cloudy winds that are contained in the Aeolus L2B product. It is demonstrated that the systematic error almost fulfills the mission requirement of being below 0.7 m s-1 for both Rayleigh-clear and Mie-cloudy winds. The random error is shown to vary between 5.5 m s-1 and 7.1 m s-1 for Rayleigh-clear winds and is thus larger than specified (2.5 m s-1), whereas it is close to the specifications for Mie-cloudy winds (2.7 to 2.9 m s-1). In addition, the dependency of the systematic and random errors on the actual wind speed, the geolocation, the scattering ratio and the time difference between 2-µm DWL observation and satellite overflight is investigated and discussed. Thus, this work contributes to the characterization of the Aeolus data quality in different meteorological situations and allows to investigate wind retrieval algorithm improvements for reprocessed Aeolus data sets.

Validation of Aeolus wind profiles using ground-based lidar and radiosonde observations at La Réunion Island and the Observatoire de Haute Provence

“…Therefore, the statistical parameters include the correlation coefficient, SD, MAD, bias, slope, and intercept. We can compare our results to other instruments that weren't mentioned during previous sections, such as ALADIN Airborne Demonstrators (A2D) (Witschas et al, 2020;Lux et al, 2020a;, airborne Doppler Wind Lidars (DWL) (Witschas et al, 2020;Witschas et al, 2022) and Radar Wind Profilers (WPR) (Zuo et al, 2022;Guo et al, 2021;Belova et al, 2021;Iwai et al, 2021). Close to Wu's (2021) observations, we observe the same consistency and similarities with the more recent studies.…”

The impact of Aeolus winds on surface wind forecast over tropical ocean and high latitude regions

“…With respect to the impact of different seasons, the results show that Aeolus has a more positive impact on wind forecast during the winter months of each hemisphere than during the summer months. This is partly attributed to the seasonal variation of solar background noise, which leads to larger random errors of Rayleigh-clear winds during summer months over polar regions and in the stratosphere (Reitebuch et al, 2022), thus resulting in larger forecast errors correspondingly.…”

“…After the successful launch, calibration and validation works have been widely carried out worldwide. Owing to the continually improved data processing chain, from Baseline 10 with M1-temperature-based bias correction and daily updates of global offset bias removal (Data Innovation and Science Cluster, 2020), the systematic errors of both Rayleigh-clear winds and Mie-cloudy winds are almost within 0.5 m s -1 despite some cases in the polar regions, and the random errors mainly vary between 4 m s -1 and 8 m s -1 for Rayleigh-clear winds and between 2.0 m s -1 and 5 m s -1 for Mie-cloudy winds (Belova et al, 2021;Iwai et al, 2021;Witschas et al, 2022;Zuo et al, 2022). However, what should be noted is that Aeolus has been suffering unexpected signal loss since the launch, probably due to the decreasing emitted laser energy for the FM-A period (August 2018 -June 2019) and/or laser-induced contamination for the FM-B period (July 2019 -September 2022) (Straume-Lindner et al, 2021).…”

Quality control and error assessment of the Aeolus L2B wind results from the Joint Aeolus Tropical Atlantic Campaign

“…The reason is most likely the fact that the Rayleigh-clear wind results with higher EE that were validated by the 2-µm DWL are largely located in dust-laden areas including the PBL where the shorter range bin thickness leads to lower SNR in addition to the attenuation by aerosols. This topic is discussed in more detail by Witschas et al (2022b).…”