Plateforme de prévision incendie et de réponse d’urgence.

FIRECASTER a pour objectif principal de développer une plate-forme de prévision d’incendie et de réponse d’urgence, basée sur la simulation.

|Coordinateur | Jean-Baptiste Filippi SPE Corte|
|Correspondant CNRM |Christine Lac|
|Financement| ANR (ANR-16-CE04-0006) |
|Début| janvier 2017 |
|Durée | 54 mois |


The goal of the FireCaster project is to develop a prototype of Wildfire Decision Support System based on state-of-the-art methods and computing resources, which will provide new ways to forecast fire danger and estimate fire impact of an ongoing crisis for operational agencies. The system is based on the Meso-NH/SURFEX atmospheric model coupled with the Forefire fire propagating model.
Developing such a decision support system strongly requires further research to be generalized and scaled at a national level. In particular, the current developments of data assimilation, high resolution fuel models, combustion products and ensemble simulations are only available on complex case-by-case analysis. Such platfoms requires two parts, risk prediction and disaster management : Risk prediction will be performed by developing innovative next-day risk assessment, using high resolution coupled surface-fuel model, Monte Carlo mass ensemble simulation of fire size and associated economic impact. Risk assessment based on fire size is a highly innovative approach compared to the current burn probability which is not a direct indicator of fire consequence. The contribution of economists partners, in an interdisciplinary framework, is essential to translate the simulations into understandable products for risk assessment. This kind of approach, based on Monte Carlo simulations, is only available since the recent availability of supercomputers. The approach will be evaluated by the operational services of Météo-France (south-east and south-west fire weather forecasters).
The crisis, disaster management integrated platform will enable, in case of a declared fire, to forecast H+1 to H+12 fire-front position (with observation assimilation), burn probabilities (ensemble forecast), high resolution plume composition, fire meteorological effects and to calculate the efficiency of fighting scenarios with the cost of the wildfire.
These “crisis models” will perform “on-demand simulation” in case of verified alert (ignition reported) with initial data available for anytime, anywhere on the national/European territory.
The main challenge is here to deliver these new diagnostics immediately on a given territory and at a given forecast date. It requires to overcome a major issue : access to high-resolution (50 m) fuel models and data. In order to characterize these fuels and the potential pollution products, better knowledge on hazardous air pollutants from open vegetation fires is mandatory and requires field observations of gas and aerosols in the vicinity of the fire. The models will then have to be generalized to the whole French territory, by linking them to the SURFEX surface model, which simulates energy exchanges and water cycle in meteorological models.

Contribution of CNRM

The contribution of CNRM is twofold. The first objective is to compare the classical version of Meso-NH in its anelastic form with a full compressible version of the model in order to evaluate the impact of strong local variations of the air density induced by the wildland fire on the dynamics. The second one is to improve fuel parametrization between Surfex and Forefire.

Publications du CNRM

  • Filippi, J.-B., F. Bosseur, C. Mari, and C. Lac, Simulation of a large wildfire in a coupled fire-atmosphere model, Atmosphere, 9(6), 218, 2018.