For the period July 14- August 2, three different assimilation cycles have been reproduced, based on the 3D-VAR AROME system (current high-resolution forecast system of Météo-France). The control cycle was based on the standard observations and did not take into account any additional COPS observations (to remain close to the routine operations). The second cycle additionally considered the LEANDRE water vapor observations whereas in the third cycle both LEANDRE and WALES observations were included. From these three sets of AROME analyses, three 30h-AROME forecasts were issued each day starting at 00 UTC.
The overall impact of the lidar data assimilation on the daily precipitation forecast was found to be neutral. For most cases, the impact (slightly positive or negative) remained insignificant except for the case of August 1 for which the precipitation forecast was significantly improved. For this specific date, night time observations (ie observations closer to the forecast initial time) were available and could explain the improvement.
Further experiments were carried out starting the forecasts at 12:UTC to make better use of the observations collected in the late morning. Improvement of the forecast is particularly clear for July 15. The isolated storm which developed over the Black Forest in the early afternoon was not forecast in the control experiment whereas it was well captured in the two other experiments.