Forest Fire Spread Prediction and Assimilation Using the Deterministic Ensemble Kalman Filter

Abstract

Computer simulation is an important method of forest fire spread prediction. However, inaccuracies stemming from input parameters and model errors can compromise predictions. To address this, we proposed a dynamic correction algorithm for forest fire spread prediction based on the deterministic ensemble Kalman filter (DEnKF). In comparison to the widely used ensemble Kalman filter (EnKF), this approach avoids “perturbation observations” to enhance robustness. We used Observing System Simulation Experiments (OSSEs) to validate the effectiveness of the proposed method in enhancing confidence in forest fire spread predictions and investigated the influence of wind conditions and the DEnKF algorithm parameters on the correction effect. This was the first attempt to apply DEnKF to forest fire spread simulation. The results confirm DEnKF superiority over EnKF in correcting forest fire spread, especially at fire line inflection points. Building upon this, we integrated the “Forest Fire Spread Prediction and Assimilation” system to provide guidance for emergency management of forest fires.