We worked on developing new tools for wildfire behavior forecasting during the 2016 CEMRACS Summer Program with colleagues from INRIA, LIMSI and the University of Maryland. A new method based on image segmentation was adapted for any wildfire spread model (level-set or lagrangian-type). The key idea is to correct the fire front trajectory as infrared images become available. This method was successfully applied to the Rim fire in Yosemite National park in 2013.
This work is supported by the French national programs, Labex AMIES and LEFE/INSU, as well as by the NSF-WIFIRE project funding in the USA.