Assimilation of Temperature and Humidity Profiles from a Raman Lidar into Convective-Scale NWP Models and the Impact on the Forecast

The current observing systems fail to provide a satisfactory amount of vertically resolved observations of temperature and humidity in the lower troposphere despite their potential positive impact on numerical weather prediction. This is particularly critical for humidity which exhibits a very high variability in space and time. It has been shown that active and passive ground based remote sensing technologies like lidar and microwave radiometers, respectively, or drones, can fill this observational gap but operational maturity was so far lacking. Only recently, commercial lidar systems for temperature and humidity profiling in the lower troposphere have become available.

MeteoSwiss operates a custom-built Raman lidar that measures tropospheric humidity and temperature profiles with high temporal and vertical resolution. The MeteoSwiss lidar is fully automatic, operates unattended and yields a data availability of more than 50% on average. It demonstrates the potential of the Raman lidar technology for operational meteorology. Temperature and humidity profiles are available every 30 minutes reaching as high as 5 km on average during day and 10 km asl during nighttime in cloud free conditions. In order to assess the impact of the lidar profile observations on Numerical Weather Prediction (NWP), three two-day assimilation experiments have been performed using two different convective-scale NWP models. We used the the operational COSMO-LETKF-based ensemble data assimilation system of MeteoSwiss and the deterministic WRF-DA system with 3DVAR and 4DVAR assimilation schemes. A case of fog formation above the Swiss Plateau and two cases of summer convection have been analysed. Here, we compare analyses and forecasts using the standard set of observations from radiosondes, wind profiler, surface stations, aircrafts and radar-derived surface precipitation profile with those, in which the additional 30min-frequency observations from the Raman Lidar have been taken into account. We find a clear positive impact of the additional temperature and humidity profiles for the convective cases, where forecasted precipitation probabilities match much better with the observations. The impact of lidar measurements on the forecast for the fog case are neutral and show no clear impact on the forecast. A neutral impact was expected given the very local nature of fog events. We would however expect that a network of temperature and humidity lidars has a clear positive impact, but this is left for future investigations.

We present in detail the characteristics of the lidar measurements and the data assimilation system and discuss the results obtained for the three case studies. We confirm that additional tropospheric temperature and humidity profiles are beneficial for numerical weather prediction, in particular for convective situations, and conclude that lidar is a suited technology to provide these observations.