Wildfire Decision Management Using Soft Computing in a System of Systems Approach

When evaluating the process of fighting wildfires, it is advantageous to consider that the fire and the firefighting process are two separate systems that interact with each other. A fire suppression effort is generally considered to be a System of Systems. The suppression activity is under the direction of the incident command system, which is part of the system of the local fire department, which is part of the emergency services and 911 dispatch systems. If the fire is a multiple jurisdictional effort, then State or Federal management systems make up part of the response parties. When considering that the socio-ecological system has a disturbance (fire) in that system, it becomes apparent that two separate systems are interacting with each other with, which allows modeling of the fire as a dynamic emergent system behavior within the overall socio-ecological system. It can be noted that a fire emergency response rapidly becomes a large-scale socio-technical system of individuals and groups, organizations, and jurisdictions of responsibility and a myriad of other system and stakeholders including the environment that all interact with each other. Without the coordinated action of the stakeholders and other systems, the fire response operations will cease to have a coherent or effective approach to the emergency. A soft computing approach using Model-Based Systems Engineering (MBSE) and Machine Learning (ML) can help determine the requirements of these fire emergencies and provide the most viable solution to mitigate the emergencies while using the optimal capabilities of these stakeholders. Further, since wildfire behaves in a very dynamic manner, they can be affected by multiple variables that affect the firefighting approach. Some of these variables include weather, fire behavior, topography, fuels, and performance of the initial attack. As such, a single decision-making process cannot neatly constrain all these dynamic variables. Fire managers have previously employed deterministic decision-making models and processes in their management plans. Further, managers typically sway from decision models for preferred outcomes during a fire. Although most fire manager decisions take a risk-averse approach, incident command may make what appears to be a risk-averse decision, but given the large amount of uncertainly, the decision may turn out to be something other than risk-averse. A dynamic model-based decision-making approach can be evaluated for such cases studies and various real-world emergency examples. A novel approach to firefighting decision-making models would be to incorporate Artificial Intelligence (AI) within the existing decision models combining existing models for fire behavior, weather, fuel load, and topography with the current approach. This allows the flexibility of intuitive decision making on the part of the incident commander to be quickly supported and balanced by analytical decisions and updates from AI. The analytical approach of AI allows the incident commanders to immediately develop feasible solutions to a dynamic system such as a fire.