Wildfire simulations to protect rural communities and avoid dire evacuations
File(s)
Author(s)
Kalogeropoulos, Nikolaos
Type
Thesis
Abstract
Large and fast wildfires can have catastrophic effects on lives, property, and the environment. Changing wildfire regimes create more destructive wildfires globally, increasing the number of people placed in danger. Evacuations from wildfires are a dramatic, disruptive, yet sometimes necessary last step to ensure people are removed from danger when other layers of wildfire protection are insufficient. Evacuations that start too late may become dire, without enough time for people to evacuate safely, exposing them to the growing hazard. It is imperative that dire evacuations be avoided. This thesis provides a definition of dire evacuations and ways to plan against them, using calculations of wildfire spread. The core effort resides in trigger boundaries, imaginary lines around communities at risk of wildfires, which when crossed by the flames denote the last safe evacuation chance. For this purpose, the k-PERIL model was developed, the first to conduct probabilistic trigger boundary calculations, and applied to the communities of Roxborough Park in Colorado, USA, Mati in Greece and Fort McMurray in Canada. With wildfire spread rate having paramount importance on the validity of trigger boundaries, even state-of-the-art wildfire models were compared, first in a series of benchmark cases, then on probabilistic trigger boundaries, to estimate the effect of model choice on safety. An ensemble approach, considering all the models combined, is required for accurate trigger boundary calculation. Moreover, an effect of wildfire that is rarely considered in evacuations is smoke. A ground-level smoke dispersion model was developed, Galini, that accounts for downwind terrain, flaming and smouldering combustion. This thesis proposes a framework to assess the danger of wildfires on communities via ensemble modelling inferred trigger boundaries, using coupled wildfire, evacuation and ground level smoke modelling. This provides quality and actionable means for communities to plan and prepare themselves for evacuations and improve community safety.
Version
Open Access
Date Issued
2025-03-28
Date Awarded
01/10/2025
License URL
Advisor
Rein, Guillermo
Sponsor
Engineering and Physical Sciences Research Council
National Institute of Standards and Technology
Natural Resources Canada
Society of Fire Protection Engineers
Publisher Department
Department of Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)