Cerfacs Enter the world of high performance ...

PhD thesis (in collaboration with CNRM/Météo-France): Fire/atmosphere two-way coupling for on-request simulation in case of wildfire hazard


Required Education : Master 2 ou Diplôme d'Ingénieurs
Start date : 2 October 2017
Mission duration : 36 mois
Deadline for applications : 30 April 2017

Description: Fire/atmosphere coupled modeling is an active research topic with operational application perspectives for wildfire behavior forecasting. In this context, the coupling between the meso-scale atmospheric research model Meso-NH (http://mesonh.aero.obs-mip.fr) and the fire front-tracking simulator ForeFire (http://forefire.univ-corse.fr) has recently been designed. See the example on the Aullène fire (Corsica, 2009).


The Meso-NH model is base on anelastic approximation, meaning that equations for the atmosphere are discretized around a stationary hydrostatic reference state, for which horizontal and temporal variations of the air density are neglected. This assumption has some important limitations for the simulation of wildfires, which feature strong heat releases involving horizontal and temporal variations of the density. An extension to compressible equations has already been implemented in Meso-NH. In this framework, the first part of the PhD aims at developing the numerical components required to solve these compressible equations at a reasonable computational cost (time splitting method) to make it compatible with operational framework.

In the second part of the PhD, the objective will be to adapt the data assimilation algorithm developed at CECI to the ForeFire/Meso-NH coupled system to simulate realistic fire front evolution. ForeFire simulates the evolution of the fire front position through a semi-empirical rate-of-spread model subject to uncertainties and through the integration of near-surface winds provided by Meso-NH. The idea behind data assimilation is to correct the trajectory of the simulated fire front as new observations become available via an ensemble Kalman filter. This, in order to improve the quality and performance in wildfire behavior forecasting. Adapting the data assimilation algorithm to ForeFire/Meso-NH requires prior to the forecast, the spin-up of the atmosphere to make it compatible with the corrected position of the active flaming areas. The cost of the methodology will be an important criterion to analyze. Testing the added value of the coupled simulations (with respect to the stand-alone fire front-tracking simulator) on the forecast performance will be also an important contribution of the PhD for large-scale wildfire hazards.

The PhD student will also work on setting the ForeFire/Meso-NH coupled system for application on wildfire hazards in the South part of France.


Summary: The PhD thesis will last 36 months starting in October or November 2017. It will be carried out at CNRM/Météo-France (Centre National de Recherches Météorologiques) and CNRS/Cerfacs (research unit CECI – Climat, Environnement, Couplage et Incertitudes), two research institutes located on the Meteopole campus, 42 Av. Gaspard Coriolis, Toulouse. A mobility, mainly in France, will also be encouraged in the frame of the ANR-FireCaster project to enhance collaboration with Laboratoire d’Aérologie (Toulouse) and the CNRS/University of Corsica.


Qualifications: Strong skills in numerical simulation, numerical methods and programmation (Fortran, C/C++, Python) are required for this work. Analysis, autonomy and team works are highly valued qualities for this work. Good English skills are also an important aspect to communicate, both written and oral, on the scientific results through scientific publications and international conferences.


Contact: If you are interested in this job, do not hesitate to contact us (join CV and a motivation letter to apply).