Thesis
Modeling debris-flow risk using hyperspectral remote sensing of wildfire burn severity
University of Iowa
Master of Arts (MA), University of Iowa
Summer 2023
DOI: 10.25820/etd.006940
Abstract
Increasing wildfire activity and rainfall strength in the western United States has elevated post-wildfire debris flow risk, particularly in Wildland-Urban Interfaces (WUI). The USGS has developed an operational post-wildfire debris flow likelihood (M1) model that uses remotely sensed data to derive watershed-averaged burn severity (dNBR) as a primary input. Here we assessed the methodology for the M1 model's burn severity input and determined the capacity for hyperspectral products to use Geo-CBI-optimized dNBR inputs using the 2017 Thomas Fire, and Montecito, California debris flows as a case study. This study modified the M1 model to use the full distribution of dNBR values, creating a per-pixel probability model of debris flow likelihood. We show the M1 model's current operationalization of reducing a watershed's dNBR to a central tendency does not capture the variability or extremes of burn severity and resulting initiation probability within a basin. Using all dNBR values, the per-pixel likelihood model considers extremes lost in single-value modeling, capturing the full range of debris flow probability. Using spectral bands more correlated to Geo-CBI measurements, AVIRIS captured increased vegetation health, increasing the magnitude of environmental change and burn severity reported compared to Sentinel-2b. The per-pixel model can be implemented into the existing USGS framework, creating new emergency management tools, and with emerging hyperspectral satellites, such as the Environmental Mapping and Analysis Program (EnMAP) and Earth Surface Mineral Dust Source Investigation (EMIT), nearly global hyperspectral data with comparable revisit times are now accessible. As wildfire and debris flow risk increases in the future, opportunities exist to improve burn severity analysis and debris flow probability modeling.
Details
- Title: Subtitle
- Modeling debris-flow risk using hyperspectral remote sensing of wildfire burn severity
- Creators
- Samuel K. Z. Taylor
- Contributors
- Susan K Meerdink (Advisor)Eric Tate (Committee Member)Dave Bennett (Committee Member)
- Resource Type
- Thesis
- Degree Awarded
- Master of Arts (MA), University of Iowa
- Degree in
- Geography
- Date degree season
- Summer 2023
- Publisher
- University of Iowa
- DOI
- 10.25820/etd.006940
- Number of pages
- x, 131 pages
- Copyright
- Copyright 2023 Samuel K. Z. Taylor
- Language
- English
- Date submitted
- 07/24/2023
- Description illustrations
- illustrations, tables, graphs
- Description bibliographic
- Includes bibliographical references (pages 65-74).
- Public Abstract (ETD)
- Wildfire size and rainfall intensity are increasing in the western United States, which has made post-wildfire debris flows more likely to occur. Meanwhile, humans have developed into these areas of increased debris-flow risk. The United States Geological Survey (USGS) has developed a post-wildfire debris flow likelihood (M1) model to assess the risk people face. The M1 model uses remotely sensed data to determine burn severity following a wildfire, which is then used to evaluate the likelihood of debris-flow initiation. Here we created a per-pixel probability model of debris flow likelihood and determined the capacity for hyperspectral products to improve burn severity analysis. The 2017 Thomas Fire, and Montecito, California debris flows were used as a case study. I show the M1 model’s current approach of reducing a watershed’s burn severity to an average does not capture the variability or extremes and resulting initiation probability within a basin. The per-pixel likelihood model considers extremes lost in single-value modeling, capturing the full range of debris flow probability. Hyperspectral imagery captured increased vegetation health, ultimately increasing the magnitude of environmental change and burn severity reported. The per-pixel model can be implemented into the existing USGS framework, creating new emergency management tools. With emerging hyperspectral satellites, such as the Environmental Mapping and Analysis Program (EnMAP) and Earth Surface Mineral Dust Source Investigation (EMIT), nearly global hyperspectral data is becoming increasingly more accessible. As wildfire and debris flow risk increases in the future, opportunities exist to improve burn severity analysis and debris flow probability modeling.
- Academic Unit
- Geographical and Sustainability Sciences
- Record Identifier
- 9984454434002771
Metrics
5 File views/ downloads
18 Record Views