Simulation et assimilation de données radar pour la prévision de la convection profonde à fine échelle

Abstract : The thesis work dealt with the use of radar data in the field of verification and initialisation of kilometric-scale atmospheric models. First, a modular advanced radar simulator is developed within the high-resolution atmospheric Meso-NH model. Sensitivity tests on the different formulations used to describe each physical process involved in radar measurements (reflectivity and radial Doppler winds) are carried out. As a result, a radar simulator for verification purposes is specified as well as an observation operator for the assimilation of radar data. Second, an original method is developed to assimilate reflectivities. This two-step method, called 1D+3DVar, first retrieves humidity vertical profiles from observed reflectivity profiles through a Bayesian technique that makes use of neighbouring model consistent humidity and reflectivity columns. The retrieved pseudo-observations are in turn assimilated by a 3DVar assimilation system at 2.5-km resolution. The method is evaluated and tuned with both simulated and real data. The method is found to be beneficial for short-range forecasts of the exceptional flash-flooding event that occurred in September 2002 in Southern France.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00222771
Contributor : Olivier Caumont <>
Submitted on : Tuesday, January 29, 2008 - 2:13:43 PM
Last modification on : Friday, April 5, 2019 - 8:14:07 PM
Long-term archiving on : Monday, June 27, 2011 - 5:42:12 PM

Identifiers

  • HAL Id : tel-00222771, version 1

Collections

Citation

Olivier Caumont. Simulation et assimilation de données radar pour la prévision de la convection profonde à fine échelle. Océan, Atmosphère. Université Paul Sabatier - Toulouse III, 2007. Français. ⟨tel-00222771⟩

Share

Metrics

Record views

489

Files downloads

696