Wildland fires occupy the biosphere as both an ecological process essential for maintaining species diversity and a hazard to human lives, infrastructure and activities. Fire managers’ ability to anticipate fire behavior is key to maximizing ecological value of fire while simultaneously minimizing negative impacts. Predictive models are often called upon as tools to support decision making by fire managers. However, prediction of fire behavior is complicated by the fact that small changes in environmental conditions can have large impacts on wildfire outcomes. This is most evident in marginal burning conditions where fine-scale details have been observed to drastically influence resulting fire behavior. Kara will present a series of collaborative studies probing the sensitivities of wildfire behavior to small changes in atmospheric conditions using FIRETEC, a process-based, computational fluid dynamics model of fire-atmosphere interactions. Ensemble simulations are utilized to sample perturbations in the turbulent wind field over homogeneous grass. In order to characterize the sensitivity of fire behavior to these perturbations in wind, several fire behavior metrics are examined, including rate of growth, total area burned and fuel consumption, which approximates fire intensity. Likewise, the characteristics of the turbulent atmosphere are probed to better understand mechanisms that may influence fire behavior. Overall spread is quantified using a variety of metrics and look for clustering and diverging behaviors. The results indicate a sensitivity to perturbations that show less predictability as the burning conditions grow more marginal.This seminar is part of the 2020-2021 Missoula Fire Science Laboratory seminar series.