Examination of Physical Coupling Processes in Wildfires Through High-fidelity Ensemble Simulations

Wildfires pose serious threats to society, environment, and ecosystems as they can disrupt, damage, and destroy infrastructure, services, and properties. To examine the complex interaction of wildfires, arising from strong coupling precession between combustion, atmospheric flow, heat-transfer, topography, and fuel properties, we present a simulation framework that integrates a high-fidelity ML-enabled simulations framework for wildfire predictions with a sampling technique to perform high-resolution ensemble simulations of large-scale wildfire scenarios. The simulation results are compared to existing experimental data for fire acceleration, mean rate of spread, and fireline intensity. Strong coupling between key compounding parameters (wind speed and slope) are observed for fire spread and intermittency. Scaling relations are derived and presented to delineate regimes associated with plume-driven and convection-driven fire spread.