AI Surrogate Modelling Framework Accelerates Wildfire Hazard Mapping Across Canada

Subject: Computer Science > Artificial Intelligence

arXiv: 2610.30170 [cs.AI]

Submitted: 16 October 2026

Status: Preprint — under review

Authors: Hager Radi Abdelwahed, Toluwaleke Olutayo, Chirag Goel, Jana Osea, Aldo Zaimi, Tom Swystun, Sandy Erni, Francisco Emiliano Lopez Saavedra, Yan Boulanger, Gaétan Marceau Caron

Abstract

Canada's increasing wildfire activity requires frequent national hazard mapping to guide proactive mitigation, including fuel management, land-use planning, infrastructure protection, and evacuation routing. However, process-based wildfire simulation frameworks such as BurnP3 and BurnP3+ rely on computationally intensive Monte Carlo simulations to model burn probability, fire intensity, and rate of spread. These workflows can take days to weeks to process a single regional modelling unit.

To overcome these computational bottlenecks, we present three contributions developed through collaboration between AI researchers and wildfire modelling experts:

1. A standardized framework for training, fine-tuning, and evaluating AI surrogate models against process-based wildfire simulators.
2. A multi-task deep learning surrogate trained on national-scale data to emulate burn probability, mean fire intensity, and mean rate of spread at a fraction of the computational cost—on the order of seconds.
3. Domain-specific metrics and counterfactual evaluation protocols to probe out-of-distribution generalization.



Integrating geospatial data (fuel, terrain, ignition likelihood) and regional weather, our model approximates BurnP3+ simulation outputs, achieving strong performance across agreement, rank-correlation, error, and spatial-overlap metrics. It preserves downstream hazard classifications and produces physically plausible responses to expert-informed counterfactual interventions. On a representative region, the surrogate achieves 806-fold end-to-end CPU acceleration and uses 13% of BurnP3+'s host memory, facilitating rapid scenario analysis.




Key Highlights



• Problem: Process-based frameworks (BurnP3, BurnP3+) require days to weeks of Monte Carlo simulation per regional modelling unit.
• Solution: A multi-task deep learning surrogate trained on national-scale geospatial and weather data.
• Performance: Strong agreement, rank-correlation, error, and spatial-overlap metrics versus BurnP3+ outputs.
• Efficiency: 806× end-to-end CPU acceleration with only 13% of BurnP3+'s host memory on a representative region.
• Robustness: Preserves downstream hazard classifications and yields physically plausible responses to expert-informed counterfactual interventions.



Metadata



• Comments: Preprint — under review
• Subjects: Artificial Intelligence (cs.AI)
• Cite as: arXiv:2610.30170 [cs.AI]

via ArXiv AI