Janice L. Coen
National Center for Atmospheric Research

Computational Science and Engineering Aspects of Wildland Fire Modeling

Wildland fire is a timely application that presents many substantial scientific, computational, and operational challenges. Although current field tools for diagnosing expected fire behavior are simple algorithms that can be run on calculators, researchers and fire managers alike envision a future when we might rely on complex simulations of the interactions of fire, weather, and fuel, driven by remote sensing data of fire location and land surface properties, as a component of planning, education, evacuation, and wildfire mitigation decision support systems.

Many of the challenges that must be faced are common to other CS&E applications (an interdisciplinary nature, physical processes that span a vast range of scales, estimating the consequences of uncertainty, visualization of complex processes, and the need to assimilate nontraditional data sources, often at disparate scales). Others are peculiar to this application, in that many of the fundamental physical processes are not understood, some have a stochastic nature and may never be deterministically modeled, nowcasting of convective precipitation (often a factor in fire behavior) may show skill only on the order of a few hours, the need to include geographic spatial data such as roads, streams, etc., due to the physical role it plays in fire behavior, and the difficulty of gathering data for verification and initialization in this dangerous environment. And despite the need for intense computations, for use as an operational application, it must proceed at better than real time.

Simulations of a coupled atmosphere-fire model will be presented. This presentation will provide insight into the current state of the science, research in progress, and the challenges that lay ahead.

Return to Program