Mesoscale data assimilation for high-resolution forecasting of heavy rain events
It has been shown by the past that convective scale atmospheric models can simulate with great realism heavy precipitation events. These simulations are however very sensitive to their initial
conditions, and in particular to the integration of mesoscale features in them such as cold pool or moist low-level jet.
Our work in the data assimilation field explored two directions: the adaptation of a 3D-VAR assimilation system to fine scale( 2-10km) and the improvement of the initialization of the mesoscale moisture field, with the development of assimilation of reflectivity radar data and
GPS zenithal total delay. Observation operators for the assimilation of precipitation radar observations and ground GPS data have been developped and an original 1D retrievial of moisture from
radar data have been designed. The mesoscale data assimilation system is based on the 3D-Var ALADIN scheme and the MESO-NH model is used to compute inovation vectors.
First results for heavy precipitation events over Southeastern France will be shown. For these cases, the mesoscale assimilation of conventional data, and in particular of mesonet near-surface observations, improves the simulation in low layers and gives a more realistic positioning of the precipitating areas. By modifying the moisture field, the assimilation of the radar and GPS data has an impact on the localization and the activity of the mesoscale convective systems.
HyMeX – Hydrological cycle in the Mediterranean Experiment 2010-2020