BALTIMORE, MD, January 26, 2022 – New audio is available for media use featuring Matthew Thompson, a Research Forester with the Wildfire Risk Management Science Team at the Rocky Mountain Research Station, USDA Forest Service. He is also a member of INFORMS, the largest association for the decision and data sciences.
In this audio content, Thompson discusses the latest ways to predict and combat wildfires. All sound should be attributed to Matthew Thompson. There are 4 questions and responses. These responses were provided on January 25, 2022.
Question 1: What are the conditions that create an environment for wildfire risk?
Time Cue: 00:35, Soundbite Duration: 00:40
Common ingredients include an ignition source as well as the interaction of topography, vegetation, and weather to affect fire behavior. But wildfire risk is multi-faceted so it’s critically important to clarify the question of “risk to what.” If we are talking landscapes, watersheds and habitat for example may be at elevated risk due to overly dense forests that burn with higher intensity. If we are talking housing loss, the primary risk factors are the condition of the home and everything around it, for example debris in gutters or firewood piles. Or, if we are talking fire personnel, then we have other risk factors to consider such as limited accessibility and falling trees.
Question 2: What are the major challenges for preventing and combatting wildfires?
Time Cue: 01:22, Soundbite Duration: 00:41
Worsening landscape conditions, a changing climate, expanding human development, and an increasingly strained response system, to name a few. But to my mind the greatest challenges may relate more to changing predominant mindsets and management paradigms. To the former, fire science points to a need to return fire to fire-prone landscapes and to use it intentionally and judiciously as a tool, challenging the notion that fire is always something that should or even can be combatted. To the latter, I’ve seen how organizational and cultural barriers to adoption of analytics can stifle innovation and inhibit learning, although that is shifting as the “moneyball for fire” idea is gaining steam.
Question 3: How can Operations Research be used to predict wildfire behavior?
Time Cue: 02:11, Soundbite Duration: 00:34
Here is where I see machine learning playing an ever more important role. To date applications of machine learning have focused more on descriptive analytics, for example, what factors explain historical patterns of area burned. However, because of trends in high performance computing, remote sensing, growing use of machine learning by fire scientists, and the growing salience of the wildfire problem attracting more data scientists and analysts, I expect we will see more predictive applications in the future, for instance predictive models of fire occurrence, fire weather, and fire behavior.
Question 4: What are the best OR strategies to combat wildfires?
Time Cue: 02:51, Soundbite Duration: 00:45
Much of the OR work relates to preparedness and mitigation, for example optimally locating operations bases, dispatching fire personnel, routing fire detection aircraft, or scheduling forest management activities to reduce hazardous fuels. When it comes to the incident response phase, however, especially for the large, complex, and long duration wildfires that we are increasingly seeing, a lack of descriptive analytics on response effectiveness is proving to be a bottleneck. Meaning, we are challenged to develop predictive and prescriptive models that are operationally relevant and empirically credible. A valuable role for OR in this space is to help improve upon and develop new performance metrics to evaluate operational capabilities, safety, efficiency, and effectiveness.
Media Contact
Ashley Smith
Public Affairs Coordinator
INFORMS
Catonsville, MD
[email protected]
443-757-3578