A fast fire rate of spread model leveraging machine learning

Principal Investigators: Adam Kochanski, Angel Farguell Caus, Jan Mandel

IAB Mentor: Josh Hacker (Jupiter Intelligence)

This project utilizes an innovative data-driven machine learning fire spread model, constructed using the fire progression data acquired from recent California fires, to develop a comprehensive fire spread simulator. The objective of this endeavor is to harness the power of machine learning (ML) to establish a functional correlation between weather conditions, fuel characteristics, and fire parameters. The resulting relationship derived from the ML model will serve as a valuable complement to the semi-empirical Rothermel fire spread model, whether employed independently or as part of coupled fire-atmosphere models. The project further intends to seamlessly integrate the newly developed ML model with gridded weather and fuel data, thereby producing an advanced and versatile fire spread simulator.