Understanding the risk of building damage to wildfires at the wildland-urban interface
Abstract
The encroachment of human settlements into the wilderness has expanded the wildland-urban interface (WUI) areas, increasing probabilities of wildfire occurrence and structure damage. In recent years, the state of California has experienced repeated wildfire events of record setting magnitude and destruction. The 2020 fire siege alone destroyed more than 9000 buildings at the WUI. However the linkages between fire behavior, fuel patterns, and structure damage are less studied. We evaluated effects of daily fire spread, spatial patterns of buildings, and other environmental conditions on the risk of building damage using a machine-learning method. Wall to wall maps of building footprints were augmented with a synthetic deep-learning framework, including Mobile-Unet and generative adversarial network, based on the high resolution NAIP aerial imageries. The probability of structure loss at daily time scale was then calculated from CalFire Damage Inspection Specialists (DINS) survey points and building footprints for all large fires during 2012-2020. Daily fire behavior metrics were derived from MODIS and VIIRS active fire products. We found that fire spread, residential density, landscape patterns, vegetative fuel loading, and fuel moisture were among the top variables controlling the building survival. This study improves our understanding of structure loss at WUI from wildfires, providing guidance for strategies toward fire-adaptive and resilient communities.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFMNH31A..07H