Frameworks for Representing Large-Fire Decision Processes and Outcomes1

Although the vast majority of wildland fires are suppressed effectively in initial or extended attack, on relatively rare occasions fires become exceptionally large, resulting in unusual resource damages, significant financial impacts and/or loss of life. Understanding how to better manage large fires and to improve methods for controlling their costs and impacts requires a detailed knowledge of the decision making processes that were ongoing at the time of the incident. Fire reviews undertaken post hoc tend to focus predominantly on incident characteristics and not on the broader decision making context within which incident management occurs. The research described in this paper uses analytical approaches from the decision and risk sciences to develop a model for decomposing and reconstructing large-fire decision processes, including influence diagrams, decision tree analysis, multi-attribute utility analysis (MAU), and other models based on decision process tracing methodologies. A preliminary model for incident decomposition and reconstruction is developed based on the concept of an “Event-Frame Model ” by which a sequential set of “Event Frames ” are defined by temporal and contextual factors, and that lead to a visual representation of an incident decomposition. The set of event frames decomposes an incident into discrete units of analysis that can incorporate other models or processes (e.g., decision analysis) to describe decision elements of a fire incident. The paper describes progress to date on the modeling approach, including efforts to apply the model in the context of an actual incident