Probabilistic Forecasting of Localized Wildfire Spread Based on Conditional Flow Matching

2026-03-27

AI Summary

This study addresses the challenges of insufficient uncertainty quantification and high computational cost in wildfire spread prediction by proposing the first probabilistic surrogate model based on conditional flow matching. The approach models fire arrival time as a conditional stochastic process given the current fire state and environmental conditions—including wind speed, temperature, humidity, terrain, and fuel type—enabling high-resolution local modeling with substantially reduced computational overhead while explicitly capturing uncertainties inherent in fire–atmosphere interactions. Trained on coupled WRF-SFIRE and North American Mesoscale simulation data, the model accurately captures the variability of fire evolution in both single-step (3-hour) and recursive multi-step (24-hour) forecasts, generating high-quality probabilistic ensembles that demonstrate its effectiveness and scalability.

Abstract

This study presents a probabilistic surrogate model for localized wildfire spread based on a conditional flow matching algorithm. The approach models fire progression as a stochastic process by learning the conditional distribution of fire arrival times given the current fire state along with environmental and atmospheric inputs. Model inputs include current burned area, near-surface wind components, temperature, relative humidity, terrain height, and fuel category information, all defined on a high-resolution spatial grid. The outputs are samples of arrival time within a three-hour time window, conditioned on the input variables. Training data are generated from coupled atmosphere-wildfire spread simulations using WRF-SFIRE, paired with weather fields from the North American Mesoscale model. The proposed framework enables efficient generation of ensembles of arrival times and explicitly represents uncertainty arising from incomplete knowledge of the fire-atmosphere system and unresolved variables. The model supports localized prediction over subdomains, reducing computational cost relative to physics-based simulators while retaining sensitivity to key drivers of fire spread. Model performance is evaluated against WRF-SFIRE simulations for both single-step (3-hour) and recursive multi-step (24-hour) forecasts. Results demonstrate that the method captures variability in fire evolution and produces accurate ensemble predictions. The framework provides a scalable approach for probabilistic wildfire forecasting and offers a pathway for integrating machine learning models with operational fire prediction systems and data assimilation.

Problem

• wildfire spread
• probabilistic forecasting
• uncertainty quantification
• conditional flow matching
• fire-atmosphere interaction

Innovation

• conditional flow matching
• probabilistic forecasting
• wildfire spread
• surrogate modeling
• uncertainty quantification