Open Collections

Arkin, Jeremy Benjamin

University of British Columbia

Due to climate change and historic land management practices, wildfires are becoming more prevalent, with increases in fire sizes, frequencies, and intensities. As a result, there is an increased need to accurately understand the environmental drivers of these wildfires, as well as their impacts. The primary objective of this dissertation was to meet this need by developing a suite of workflows that examine the utility provided by newly available remote sensing data that provide greater levels of detail than conventional data. This research was divided between pre- and post-fire management stages. The post-fire research questions evaluated the ability of remote sensing data acquired from cost-effective, remotely piloted aerial systems (RPAS) to create detailed pixel- and individual tree-based fire severity maps. The data used to produce these maps were acquired exclusively post-fire and included high resolution orthomosaics and dense digital aerial photogrammetric (DAP) point clouds. The pixel-based fire severity maps were produced at 5- and 1-meter resolutions using a supervised random forest classification approach, whereas the individual tree-approach estimated crown scorch height across segmented trees that were classified as having fire damage. The pre-fire questions were focused on characterizing the amount and arrangement of combustible fuels that drive wildfire behaviour, and did so at the plot- and individual tree-level. The plot-level approach estimated a full suite of canopy and surface fuel metrics that are commonly used in fire behaviour models. Due to the complex nature of these fuels, these estimations were carried out separately using RPAS-acquired light detection and ranging (LiDAR) and DAP point clouds, as well as ground-based mobile laser scanning (MLS) point clouds. The individual tree-based fuel characterizations were carried out by automatically segmenting and evaluating continuous portions of crown fuels from RPAS LiDAR point clouds and comparing against measurements collected from more detailed MLS point clouds. Through the investigation of these research questions, I provide detailed information that describes both the strengths and weaknesses of these data as well as a suite of methodologies that can be incorporated in conventional and next generation forest management frameworks.

Text

2022-07-22

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/82192

Forestry

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/