- 1Météo-France/CNRS CNRM-UMR3589, GMAP, Toulouse, France (eric.bazile@meteo.fr)
- 2Météo-France/CNRS CNRM-UMR3589, GMME, Toulouse, France
- 3Direction de la Climatologie et des Services Climatiques, Météo-France, Toulouse, France
- 4Météo-France/CNRS, CNRM-UMR3589, CEN, Saint Martin d’Hères, France
The ARRA re-analysis is based on the AROME numerical weather prediction (NWP) system at 1.3 km over France. ARRA will cover the period 1961-2020 and will catch up the real time in 2026. ARRA, because of its high horizontal resolution, uses as the lateral boundary conditions the UERRA re-analysis produced at 11km for the period 1961-1984 and the European Copernicus CERRA reanalysis, produced at 5.5 km resolution after 1984.
With this new re-analysis at the kilometer scale with a non-hydrostic model, small-scale phenomena such as convection, orographic wind and extreme precipitation should be significantly improved compared to the ERA5 global re-analysis or the European Copernicus re-analysis (CERRA).
The ARRA reanalysis has a 3-hour cycle for the surface analysis (air temperature and humidity at 2 meters, snow and soil moisture) associated with the Incremental Analysis Update (IAU) technique for the upper atmosphere using UERRA or CERRA analysis. An other component, called ARRA-Land and produced at the same resolution, will focus on surface variables, snow and soil moisture. ARRA-Land is an offline simulation of the SURFEX modeling platform with more advanced surface physics than in ARRA and is driven by the ARRA atmospheric forcing fields and by an analysis of 24-hour cumulative precipitation (MESCAN (Soci et al. 2016)).
After a brief description of the system configuration and production status, a preliminary comparison with ERA5 and CERRA will be presented, focusing on several cases of extreme precipitation over France such as Nimes (1988), Vaison-La-Romaine (1992), … and in the Alps such as 22-25 September 1993 in Switzerland. The added value of the 24hour-precipitation analysis for the selected cases will be discussed in the perspective of the ARRA-Land product.
ARRA re-analysis data will probably be used for training purposes in the perspective of an AI based version of AROME.
How to cite: Bazile, E., Le Moigne, P., Van Hyfte, S., Selly, Y., Verrelle, A., Willemet, J.-M., Blanc, P., Monteiro, D., and Rousset, F.: Preliminary evaluation of the ARRA re-analysis over France : focus on extreme precipitation, EMS Annual Meeting 2025, Ljubljana, Slovenia, 7–12 Sep 2025, EMS2025-648, https://doi.org/10.5194/ems2025-648, 2025.