Localized Forest Fire Risk Prediction: A Department-Aware Approach for Operational Decision Support
Nicolas Caron, Christophe Guyeux, Hassan Noura, Benjamin Aynes
TL;DR
This work introduces a department-aware, national-scale wildfire risk dataset for metropolitan France and a corresponding ordinal, multi-class benchmark for two operational targets: Fire Occurrence and Burned Area. By aligning data with French departmental fire services and enriching it with calendar, forest cover, land cover, and fire indices, the authors enable region-specific risk forecasting on a 2 km × 2 km × 1 daily grid, solved via a diverse set of models including GraphCastGRU which achieves the best overall performance. Key findings show that multi-class, ordinal predictions outperform binary formulations at this scale and that FO is more predictable than BA, though both remain challenging; SHAP analysis highlights historical and short-term signals as dominant features with potential generalization costs. The dataset and benchmark have practical impact for field operations by enabling department-tailored predictions, and future work focuses on ordinal losses, federated learning, and advanced spatial-temporal architectures to improve generalization and realism.
Abstract
Forest fire prediction involves estimating the likelihood of fire ignition or related risk levels in a specific area over a defined time period. With climate change intensifying fire behavior and frequency, accurate prediction has become one of the most pressing challenges in Artificial Intelligence (AI). Traditionally, fire ignition is approached as a binary classification task in the literature. However, this formulation oversimplifies the problem, especially from the perspective of end-users such as firefighters. In general, as is the case in France, firefighting units are organized by department, each with its terrain, climate conditions, and historical experience with fire events. Consequently, fire risk should be modeled in a way that is sensitive to local conditions and does not assume uniform risk across all regions. This paper proposes a new approach that tailors fire risk assessment to departmental contexts, offering more actionable and region-specific predictions for operational use. With this, we present the first national-scale AI benchmark for metropolitan France using state-of-the-art AI models on a relatively unexplored dataset. Finally, we offer a summary of important future works that should be taken into account. Supplementary materials are available on GitHub.