Home | Wildfires | Contact Info

Wildfire Surrogate Modeling

Overview

This work evolved from an MSc research project at Imperial College London with the goal of developing a foundational deep learning surrogate model for wildfire spread, based on the SCALED model developed by Yueyan Li. Following the methodology of Burge et al. (2023), data was generated using ELMFIRE.

This visualization compares the model's prediction of the wildfire spread and the ground truth over 12 hours of autoregressive rollout, at each step using the model's prediction as input for the next prediction. The first row shows the ground truth from ELMFIRE, while the second row shows the model's prediction for time of arrival of the fire front, spread rate, flame length, and burn scar.

You can find the repository for this project at https://github.com/matheuboucher/WildfireModeling.

Wildfire Simulation Result

Region of Interest

10,000 ELMFIRE simulations were generated with real-world fuel model data for the purposes of training and evaluating the model's ability to generalize to unseen fuels and topography. The training set was generated from a region spanning parts of eastern California and western Nevada, USA, while the test set was generated from a region of western and northern California.

Work is ongoing to publish the datasets used for this project to a public repository.

Ignition Areas for Real-World Fuel Model Dataset

Future Work

Future work includes generating more diverse datasets and refining the model.