MITRE Fireline is a set of tools that blend fire projection modeling and reinforcement learning (RL) to create robust wildfire mitigation strategies. The cornerstone of our research is a flexible fire simulator suite for generating realistic wildfire scenarios, combined with an RL engine that simulates and evaluates firefighter interventions, considering various factors like available resources and land value. This engine generates strategic plans that maximize value retention and optimize resource allocation.

Our aim is not only to control and minimize damage during active wildfires but also to devise preventative strategies to preempt future fires. By integrating AI and practical firefighting tactics, MITRE Fireline promotes a data-driven approach to wildfire management, taking into account the inherent complexities and unpredictable nature of wildfire, as well as an uncertain climate future.

Burn Mitigation Dataset

To conduct our research on Fireline, we needed a dataset to train the RL agents and to confirm our simulator accurately portrayed how real fires spread.

We found fire data across multiple agencies to be incompatible due to inconsistent tracking of fire spread and mitigations and, consequently, observed that fire spread models suffer from incomplete data. To overcome this problem, we merged several datasets and cleaned the data to create the first wildland fire dataset that includes community mitigation efforts.

We’ve provided the dataset here, free to use by others for modeling and predicting wildfire spread and mitigation techniques.

We hope this dataset will enable the AI community to come together with the greater wildfire community to build and validate future modeling techniques in this domain.

View and download the Burn Mitigation Dataset here: fireline.mitre.org/burnmd

Wildfire Mitigation Tool Suite

To better prepare for and react to the increasing threat of wildfires, more accurate fire modelers and mitigation responses are necessary. We introduce SimFire, a versatile wildland fire projection simulator designed to generate realistic wildfire scenarios, and SimHarness, a modular agent-based machine learning wrapper capable of automatically generating land management strategies within SimFire to reduce the overall damage to the area.

We hope this publicly available system allows researchers and practitioners the ability to emulate and assess the effectiveness of firefighter interventions and formulate strategic plans that prioritize value preservation and resource allocation optimization.

View and contribute to SimHarness and SimFire here: https://github.com/mitrefireline