Applying Optimization Methods to Wildfire: A Simulation Study
Author(s)
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Wildfire response is a wide field that spans many disciplines, from first response and fire thermodynamics to aircraft operations and system-of-systems analysis. While many fires can be addressed with only ground crews, fires in high-risk conditions (high temperature, low humidity, high wind, etc.) can be difficult to tame even for high-cost air-tankers. In such situations, aerial retardant drops are a critical tool, yet their effectiveness depends strongly on where and when they are deployed. Quantifying the quality of a retardant drop is therefore essential for guiding operational decisions and improving suppression outcomes. Wildfire growth, however, is inherently a multi-objective problem. It involves tradeoffs between minimizing burned area, reducing risk to human populations, limiting emissions, and slowing fire spread. This complexity makes it difficult to define a single measure of success and to apply optimization techniques directly. To address this challenge, this work compares multiple novel cost functions to compress one or more spatial variables related to wildfire growth into a single scalar variable representing the “level of undesirability” of future fire growth. The cost functions integrate metrics of area burned, fire progression, population exposure, or emissions, thereby capturing both local and global impacts of suppression strategies. Using these cost functions, an optimization of retardant drop locations for aircraft operations was conducted using brute force search, Bayesian optimization, and differential evolution. Results demonstrate that the proposed framework can systematically evaluate tradeoffs between competing objectives and provide actionable guidance for aerial firefighting operations. Beyond immediate tactical applications, this approach offers a foundation for integrating ecological, social, and environmental priorities into wildfire management, supporting more resilient and sustainable suppression strategies.
Sponsor
Date
2026-01
Extent
Resource Type
Text
Resource Subtype
Paper
Rights Statement
Unless otherwise noted, all materials are protected under U.S. Copyright Law and all rights are reserved