Leveraging Genetic Algorithms and Mathematical Wildfire Propagation Models to Determine Optimal Firebreak Placement

Overview

Each year, wildfires destroy billions of dollars of infrastructure, cost billions of dollars to suppress, emit billions of tonnes of carbon, and claim many lives. These negative effects are increasingly exacerbated by climate change, making efficient wildfire containment-optimization systems critical. Current assistive wildfire treatment software, like WFDSS and FLAMMAP, lack decisive algorithmic capabilities that would be used to optimize wildfire containment—through optimizing firebreak construction—in the complex circumstances that characterize wildfires. While the feasibility of Genetic Algorithm (GA) wildfire containment-optimization models has been demonstrated, there have been few recent attempts at developing one. Bridging this gap, we developed an algorithmic fire containment-optimization system that utilizes a novel Hybridized RPGA (Real Parallel-Stream Genetic Algorithm) to optimize firebreak placement on a wildfire simulated by a novel adaptation of the Alexandridis Cellular Automata (CA)-based wildfire propagation model that supports standardized LANDFIRE and WindNinja data. The system was evaluated on the 2022 Oak Fire in California, where the wildfire simulation model was found to have an average 3.33% error in burnt area and the H-RPGA containment-optimization model was found to improve wildfire containment by decreasing the final burn area by 9.44% at best compared to actual containment. The effectiveness of the H-RPGA containment-optimization model was also demonstrated by the highly significant (p < 0.001, r = -0.922) relationship between generations and fitness score, where a lower fitness score indicates better performance. Our research underscores the serious potential of CA/H-RPGA-based wildfire containment-optimization systems in improving wildfire suppression.

Leveraging Genetic Algorithms and Mathematical Wildfire Propagation Models to Determine Optimal Firebreak Placement

Overview

Source coverage

Awards (2)

Competition history

Resources

Related projects

Mathematical Fire Fighting: Combating Fire with Delaunay Triangulation and Longitudinal-Reversible Cellular Automata

It's Flaming Out: Using Artificial Intelligence To Emulate Critical Aspects of Wildfire Growth

Cellular Automata-Based Mathematical Model for the Spread of Forest Fires

A Novel AI Pipeline for Predicting Next Day Wildfire Spread

Developing a Novel Wildfire Spread Prediction Model Using Spatio-Temporal Analysis and Time-series Forecasting

A Data Analysis Approach to Wildfire Prediction

Development of a Novel Method to Predict Wildfire Ignition Sites in San Diego County

BlazeAlert: A Novel Spaciotemporal Deep Convolutional Neural Network Model for Real Time Wildfire Risk Assessment

Browse more like this