Optimal spatial location of forest fuel management activities

Abstract

Recent wildfire events in many parts of the world have galvanized public and private organizations to spend billions of dollars to reduce the risk of fires to natural and man-made structures. In the previous century, much of this effort was concentrated on developing the science and technology needed to suppress wildfires once they were ignited. However, man's ability to control burning wildfires is limited. Therefore, natural resource managers are now emphasizing fire-preventive (pre-suppression) strategies. Among the most common of these preventive strategies are efforts to reduce the amount of fuel present in the landscape. Unfortunately, no standard methodology exits to help managers optimizing the spatial location of preventive fuel treatment activities. This dissertation describes the conception, definition, implementation, and validation of a heuristic model that spatially optimizes the placement of forest fuel management operations under budgetary constraints. It combines geographic information systems (GIS) based spread cost analysis operations and fire management concepts and techniques to locate the areas where fuel treatments should be applied. Its main purpose is to define near-optimal locations for fuel management practices that minimize the risk of having catastrophic wildfire events negatively affecting points or areas of value. The model was implemented as a GIS-linked computer program and optimized to reduce the computing time required to obtain a solution. It was validated through a comparative study among fuel management prescriptions developed by fuel management experts and solutions generated by the computer model. Results from the validation indicated that the fuel treatment prescriptions defined by the model outperformed the fuel management plans defined by fuel experts. On average, experts' prescriptions were from 40 to 50% less efficient (in terms of the amount of time needed for a wildfire to impact a point of value) than the solutions generated by the computer model.