The influence of assimilating Doppler wind lidar observations from the Swabian MOSES 2023 campaign on mesoscale wind variability over southwestern Germany

A wide range of weather phenomena, including for example valley circulations and convective initiation, are connected to mesoscale wind fluctuations. Their representation in convective-scale numerical weather prediction models, particularly in complex terrain, remains uncertain but may significantly affect forecast quality.
To quantify the potential added value of denser wind observation networks, we assimilate 3 months of data from a network of 12 Doppler wind lidars obtained during the Swabian MOSES campaign around the Black Forest region in southwestern Germany during summer 2023. Vertical profiles of the horizontal wind components up to approximately 4 km altitude retrieved from the wind lidars were assimilated using the regional forecasting system of the German Weather Service based on the Kilometer-Scale Ensemble Data Assimilation (KENDA) system using a Local Ensemble Transform Kalman Filter (LETKF) and the ICOsahedral Non-hydrostatic (ICON) model.Overall, ICON represents the wind fields well and the assimilation reduces short-term forecast errors. As expected, the observation influence is largest within the campaign region but also spreads horizontally and vertically away from it. Differences between observations and model tend to be particularly large during convective conditions. Moreover, assimilating the dense wind information leads to small but systematic differences in wind speed and direction compared to an experiment without Doppler wind lidar assimilation. On average, the zonal wind speed is slightly overestimated in the model, while the meridional wind speed is underestimated, resulting in a rotation of the wind direction. The underlying causes of this bias are currently under investigation.