Development of forest fire management decision support systems

点击翻译按钮获取中文摘要

I'm requesting NSERC funds to develop and field test 1) a forest fire detection patrol routing model, 2) an initial attack airtanker deployment model and 3) a large fire containment optimization model. My students and I will develop novel models of fire management systems that; 1) will be of interest to operational researchers, fire researchers and managers and 2) can be used to enhance forest fire management in Canada and elsewhere.***1) Forest fire detection patrol routing model: Patrol aircraft are often used to search for forest fires but there are, to my knowledge, no models that are suitable for predicting the probability that airborne detection observers will detect small un-reported fires burning near their flight lines. We will use Global Positioning System (GPS) data that tracks detection aircraft to develop a detection probability model that we will use to develop a daily detection demand model that will be used to determine how important it is for a detection patrol aircraft to visit each location in a protected area each day. We will then use vehicle routing methods to help decide how to satisfy the detection demand each day, and field test our approach in northwestern Ontario.***2) Initial attack airtanker deployment model: Initial attack airtanker systems are complex multi-server spatial queueing systems with fire arrival rates that vary over time and space and service times that depend on waiting time as fires grow while they wait in the queue. My students and I have developed queueing models of airtanker systems but there remain important challenges, two of which are; 1) the development of new service process models and their incorporation in more realistic queueing models of initial attack airtanker systems and 2) the development of daily airtanker deployment optimization models. We will use GPS data provided by the Ontario Ministry of Natural Resources and Forests (OMNRF) to develop a new initial attack airtanker system model. We will then use stochastic optimization methods to develop a decision support system (DSS) that can be used to help determine how best to deploy the airtankers each day, and field test it in northwestern Ontario.***3) Large fire containment optimization model: Large forest fires are managed by Incident Management Teams (IMTs) that often have only 3 to 4 hours to develop a containment strategy that will satisfy the local land manager's objectives under considerable uncertainty concerning weather and other factors, safely and at a reasonable cost. During the past 5 fires seasons I was assigned as a researcher, to an OMNRF IMT, to study their planning and decision-making problems and I have, in collaboration with others, developed a simple deterministic "proof of concept" large fire containment optimization model. I will incorporate what I have learned about large fire management in a more realistic containment optimization model that I will field test in collaboration with an IMT.**