Remote sensing and geospatial modeling of wildland firefighter safety

Update Item Information
Publication Type dissertation
School or College College of Social & Behavioral Science
Department Geography
Author Campbell, Michael James
Title Remote sensing and geospatial modeling of wildland firefighter safety
Date 2018
Description With increasing wildfire activity throughout the western United States comes an increased need for wildland firefighters to protect civilians, structures, and public resources. In order to mitigate threats to their safety, firefighters employ the use of safety zones (SZ: areas where firefighters are free from harm) and escape routes (ER: pathways for accessing SZ). Currently, SZ and ER are designated by firefighters based on ground-level information, the interpretation of which can be error-prone. This research aims to provide robust methods to assist in the ER and SZ evaluation processes, using remote sensing and geospatial modeling. In particular, I investigate the degree to which lidar can be used to characterize the landscape conditions that directly affect SZ and ER quality. I present a new metric and lidar-based algorithm for evaluating SZ based on zone geometry, surrounding vegetation height, and number of firefighters present. The resulting map contains a depiction of potential SZ throughout Tahoe National Forest, each of which has a value that indicates its wind- and slope-dependent suitability. I then inquire into the effects of three landscape conditions on travel rates for the purpose of developing a geospatial ER optimization model. I compare experimentally-derived travel rates to lidar-derived estimates of slope, vegetation density, and ground surface roughness, finding that vegetation density had the strongest negative effect. Relative travel impedances are then mapped throughout Levan Wildlife Management Area and combined with a route-finding algorithm, enabling the identification of maximally-efficient escape routes between any two known locations. Lastly, I explore a number of variables that can affect the accurate characterization of understory vegetation density, finding lidar pulse density, overstory vegetation density, and canopy height all had significant effects. In addition, I compare two widely-used metrics for understory density estimation, overall relative point density and normalized relative point density, finding that the latter possessed far superior predictive power. This research provides novel insight into the potential use of lidar in wildland firefighter safety planning. There are a number of constraints to widespread implementation, some of which are temporary, such as the current lack of nationwide lidar data, and some of which require continued study, such as refining our ability to characterize understory vegetation conditions. However, this research is an important step forward in a direction that has potential to greatly improve the safety of those who put themselves at risk to ensure the safety of life and property.
Type Text
Publisher University of Utah
Subject Remote sensing
Dissertation Name Doctor of Philosophy
Language eng
Rights Management (c) Michael James Campbell
Format Medium application/pdf
ARK ark:/87278/s66q6vxb
Setname ir_etd
ID 1494240
Reference URL
Back to Search Results