Probabilistic Forecasting of Localized Wildfire Spread Based on Conditional Flow Matching
Authors: Bryan Shaddy, Haitong Qin, Brianna Binder, James Haley, Riya Duddalwar, Kyle Hilburn, Assad Oberai
Organizations: Department of Aerospace and Mechanical Engineering, University of Southern California · Department of Mathematics, University of Southern California · Cooperative Institute for Research in the Atmosphere, Colorado State University · Department of Computer Science, University of Southern California
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.
Figures & tables
Figure 1: Two training samples, with input variables and corresponding fire arrival time outputs displayed. The wind field shown is plotted using the magnitude and direction computed from the zonal and meridional wind components and fuel category data is additionally displayed as a single categorical image determined from the one-hot encoded data used for training.
Figure 2: Predicted fire area versus (a) u wind speed, (b) v wind speed, (c) relative humidity, (d) temperature, and (e) terrain slope. For each plot, the mean predicted area is shown along with variability indicated by the standard deviation.
Figure 3: Results for three cases from the testing dataset. The first row shows conditioning inputs, the second row shows six sampled predictions, and the third row presents the target, medoid prediction, pixel-wise standard deviation, and absolute error between the medoid and target.
Figure 4: Performance of the wildfire spread surrogate model on the testing dataset. Panels (a) and (b) show medoid predicted versus target burned area over the full and restricted ranges, respectively. Panels (c) and (d) show the Sørensen–Dice coefficient (SC) as a function of target burned area over the same ranges.
Figure 5: Results from recursive application of the wildfire spread surrogate model to predict growth from ignition to 24 hours for four cases. The first row shows the initial fire area, target growth, medoid prediction, standard deviation, and absolute error. The second row shows representative samples of predicted fire growth.
Figure 6: Results from recursive application of the wildfire spread surrogate model to a subsection of an active fire, initialized from the 12-hour perimeter and propagated to 36 hours. The first row shows the initial fire area, target growth, medoid prediction, standard deviation, and absolute error. The second row shows representative prediction samples.
Wildfire forecasting predicts where fires will start, how they will spread, and their likely intensity or impacts using environmental and fire observations.
In this work we present a probabilistic model for predicting the temporal evolution of wildfire size using conditional generative algorithms, namely flow matching. Rather than predicting spatially resolved fire progression, the proposed model forecasts a time series of fire acreage, substantially reducing computational cost and enabling efficient integration with downstream applications such as early warning systems and risk assessment tools. Given estimates of the fire size at the current and previous forecasting steps, alongside spatially averaged weather conditions, terrain characteristics, and fuel composition, the model predicts the change in fire area over a 3-hour interval. Training data are drawn from 140 coupled atmosphere-wildfire simulations generated using WRF-SFIRE, paired with weather data from the North American Mesoscale (NAM) model. Sensitivity analysis demonstrates that the model captures physically-consistent responses to variations in wind speed, temperature, and relative humidity. Model performance is evaluated against reference solutions from WRF-SFIRE. For single-step predictions, the model achieves a symmetric mean absolute percentage error (sMAPE) of 0.39 relative to the reference WRF-SFIRE solutions. Recursive multi-step predictions over 24 hours are performed for 12 held-out test cases, where the flow matching model achieves a final fire area sMAPE of 0.39 while further providing meaningful quantification of forecast uncertainty.
Isaac Lee, Bryan Shaddy, Assad Oberai
Department of Mathematics, University of Southern California · Department of Aerospace and Mechanical Engineering, University of Southern California
Reliable wildfire spread prediction is vital for risk-aware emergency planning, yet most deep learning models lack principled uncertainty quantification (UQ). Further, for boundary-sensitive cases like wildfire spread, evaluating models with global metrics alone is often insufficient. To shift the focus of UQ evaluation toward a more operationally relevant approach, the Fire-Centered Evaluation Region (FCER) framework is introduced as a spatially conditioned protocol to characterize UQ within critical fire zones. Using FCER, an Ensemble is compared against an distilled single-pass student model on the WildfireSpreadTS dataset. The student model demonstrates comparable calibration and complementary uncertainty ranking in boundary-relevant regimes. Code is available at https://github.com/jonasvilhofunk/WildfireUQ-FCER
Accurate large-scale wildfire spread modelling requires models that capture both the environmental conditions associated with fire occurrence and the local dynamics of fire propagation. We propose a three-stage framework that combines a Random Forest (RF) model with a cellular automaton (CA). First, an RF model trained on the 2021 Canadian fire season estimates daily pixel-level fire-occurrence probabilities. Second, quantile gradient boosting models provide optional spread-rate priors for sensitivity analysis. Third, an RF-informed CA combines the RF probability layer with neighbourhood-driven spread on a 5 km grid. The RF model achieved AUC values of 0.725--0.795 on the 2022--2024 datasets, while the RF-informed CA achieved substantially higher spatial overlap than the evaluated CA-only baselines in the 2023 simulation. A higher-resolution simulation provides an additional qualitative assessment of local spatial errors. These results suggest that combining RF-derived probabilities with local CA spread can improve large-scale wildfire simulations under the tested conditions.