Description
Date depot: 1 janvier 1900
Titre: Estimation of dynamics from multiscale image assimilation. Application to: Rain nowcasting from ground radar images
Directrice de thèse:
Isabelle HERLIN (Inria-Paris (ED-130))
Directeur de thèse:
Dominique BÉREZIAT (LIP6)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
{{Context and Objective}}
Rainy events are often not accurately forecasted by meteorological models. This is due to a weak description of rain and clouds and to the spatial resolution of the models, which does not allow accurately assessing the small scale processes. A strong collaboration for research and software development has been initiated between Inria and the SME Weather Measures for the definition of local meteorology forecast tools. The end-users are farmers, which require a forecast at a temporal horizon of 4 hours and at the sub-parcel scale. These 4 hours allow to better plan and optimize the irrigation and use of fertilizers for instance.
A number of research subjects were jointly defined by Inria and Weather Measures, which are the core subject of the PhD, on the issue of rain nowcast from a local network of ground- radars.
{{Scientific Content}}
The research work on the nowcasting system includes two main parts. First, the atmospheric dynamics is estimated from data acquired by the network of ground-radars monitoring the studied site. Experiments are currently conducted in the region of Clermont- Ferrand and one in the Beauce. The estimation method is based on data assimilation of the images with a numerical model describing the local advection, convection and diffusion processes. The work will concern the design of multiscale data assimilation techniques, which permit the fusion with the low resolution images of the ARAMIS network, operated by Météo-France, and pointwise pluviometers measures.
Second, the short term forecast will be investigated. It needs advanced numerical schemes for synthetizing the future rain quantities and HPC tools for producing near real time results.
This research in applied mathematics requires inputs on the physical processes, on software development and on HPC technics. A support on these points will be available from radar specialists and computer scientists of Weather Measures and from researchers of Maison de la Simulation in Saclay.
Doctorant.e: Zebiri Aniss