Vegetation Cover and Structure Loss in Four Northern California Wildfires: Butte, Tubbs, Carr, and Camp

This study examines the effect of vegetation cover near structures on the loss rate for single family residences (SFR’s) in four recent northern California wildfires: the Butte fire (September, 2015), the Tubbs fire (October, 2017) , the Carr fire (July, 2018) , and the Camp fire (November, 2018). In total, 19,508 destroyed and 5,208 surviving SFR’s were included in the study. The proportion of pre-fire vegetation cover within 25 meters of a central point representing each structure was estimated using high-resolution infrared aereal imagery. For each fire, structures were grouped into 10 vegetation cover classes, based on the proportion of cover, and loss rates were calculated by class. Linear regression was applied to estimate the effect of vegetation cover on loss rates. Loss rates were found to increase in proportion to vegetation cover in three of the four fires. For the two fires in the Sierra Nevada foothills (Butte and Camp) the slope of the loss rate regression line was similar, around 0.9. That is, the probability of loss increased by 0.9 % for every 1% increase in vegetation cover in the 25-meter zone. For the Carr fire, the loss rate slope coefficient was estimated to be 0.4, about half of the value for the Sierra fires. Structure loss rates in the Tubbs fire were uncorrelated with vegetation cover. The effects of wind speed, vegetation type, and housing density on loss rates were also examined. Loss rate regression lines tended to shift upward in concert with the maximum wind speeds encountered on each fire. In the case of the Tubbs fire, high winds appeared to completely negate the influence of vegetation cover on loss rates. Structures located in conifer vegetation types had a higher loss rate when compared to those located in hardwood types for the Sierra fires, but that pattern did not hold for the Tubbs or Carr fires, located in more coastal mountain ranges. Loss rates did not differ significantly by Wildland Urban Interface zones as mapped by the University of Wisconsin’s SILVIS lab.