Towards predictive simulation of wildfire spread using a reduced-cost Ensemble Kalman Filter based on Polynomial Chaos approximation

Abstract

The sequential correction of a fire spread model parameters is performed via the assimilation of airborne-like fire front observations in order to improve the simulation and forecast of the fire propagation. An Ensemble Kalman Filter (EnKF) is applied to reduce uncertainties in the atmospheric and vegetation parameters for the Rate Of Spread (ROS) model. The non-linear relation between the parameters and the fire front position induced by the non-linearities of the fire spread is described stochastically over the EnKF members. In order to reduce the computational cost of the data assimilation algorithm, a surrogate model based on a Polynomial Chaos (PC) approximation is used in place of the forward propagation model. The merits of using the EnKF algorithm based on the PC approximation are highlighted in experiments using synthetical and real measurements.