Climate Change and Wildfire Severity in California
Producing realistic and useful predictions of climate change impacts calls for a systems view that accounts for the combination of direct effects of climate on biogeophysical processes and likely human responses on a regional scale. As a case in point, we estimated the impact of global warming on wildland fire in northern California (Figure 1) by linking general circulation climate model (GCM) output to local weather and fire records, and then projecting fire outcomes with a fire suppression model.
Wildfires are a pervasive risk in the U.S., consuming an average of five million acres per year. The ongoing expansion of low-density residential development into areas covered by flammable vegetation creates wildland/urban interface environments that place more and more people and property at risk from wildfire. Between 1985 and 1994, wildfires destroyed more than 9,000 homes in the U.S. at an average insured cost of about $300 million per year, nearly an order of magnitude greater than during the previous three decades.
The California context is a fertile one for in-depth analysis because it contains a broad continuum of ecosystems here fire plays an important role, and because of the extent of the wildland/urban interface phenomenon. California experiences greater economic losses from wildfire than any other state. Its population growth is strongly concentrated in areas that would be most affected by climatic change.
Climate is one of the main determinants of wildfire regime. By warming and drying vegetation, and by stirring the winds that spread fires, global warming and associated climate change have the potential to increase the severity and extent of wildfires. Researchers applying predictions of general circulation models (GCMs) have consistently found that climate change will lead to increases in the frequency of weather conditions associated with high wildfire hazard and to corresponding changes in weather-related indices of potential fire intensity and rate of spread (Figure 2), increases in fire ignitions, and a lengthened fire season.
However, because prior studies were based on weather indices and not actual fires or fire behavior in specific locales, they could not take into account the complex interaction of wildfires and suppression efforts or the skewed probability distribution of fire severity, in which large and extreme fires are rare and small fires are common. As a result the likely changes in area burned or suppression activity have not previously been quantified.
We employed analytical tools based on the California Department of Forestry and Fire Protection's Changed Climate Fire Modeling System (CCFMS) to extend what was learned from previous studies in new and important directions. Fire and the full suite of relevant climate variables—temperature, wind, humidity and precipitation—are fully integrated into the model. In addition, CCFMS incorporates inputs and provides outputs at a comparatively fine geographic scale which allows analysis by vegetation and/or population density (Table 1 and Figure 3). We tested the sensitivity of results to the outputs of three GCMs, and chose one—the Goddard Institute for Space Sciences (GISS) GCM—that yielded relatively conservative impact results.
|Escaped Fires (number)||Average Size of
Contained Fires (acres)
|Present||Future||% Change||Present||Future||% Change|
|Santa Clara, Grass|
|Amador—El Dorado, Chaparral|
The warmer and windier conditions corresponding to a 2xCO2 climate scenario produced fires that burned more intensely and spread faster than current-day fires. Despite enhanced fire suppression efforts (a.k.a. "dispatch"), the number of "escaped" fires (i.e., those exceeding initial containment limits) increased 51% in the south San Francisco Bay Area, 125% in the Sierra Nevada, and were unchanged on the northern coast (Figure 4). Changes in area burned by contained fires were 41%, 41%, and -8%, respectively. Interpolation of these results to the entire northern California state protection area produced increases of 110 escapes in an average year (a doubling of the current number), and an additional 5,000 hectares burned by contained fires. California already spends $300 million per year on initial attack fire protection; it is conceivable that this might have to be increased by 50% or more to maintain the current escape rate.
By using conservative climate model projections and disregarding hard-to-model secondary impacts such as long-term changes in vegetation dynamics (which would most likely exacerbate wildfire severity by increasing the geographic extent of high flammability fuels such as grass), our estimates represent a best-case forecast.
Fire severity of this magnitude would have widespread impacts on vegetation distribution, forest condition, and carbon storage, and would necessitate costly augmentation of fire-fighting infrastructure, and greatly increase the risk to property and human life.
By generating predictions of changes in escape frequency and area burned that reflect the interaction of fire growth and suppression, our approach goes a step further than previous fire danger index-based techniques in addressing changes in fire statistics of interest to fire and resource planners, and insurers concerned about climate change. These statistics, along with output on the utilization frequency of firefighting resources, can be used by planners and economists to evaluate likely outcomes. Escapes and area-burned statistics for specific geographic areas can also serve as the basis for estimates of impacts on ecosystems, smoke emissions, and economic losses.
For more information, contact:
- Evan Mills
- (510) 486-6784; fax (510) 486-6996
- Margaret S. Torn
- (510) 495-2223; fax (510) 486-7070
- Jeremy S. Fried
- USDA Forest Service, Pacific Northwest Research Station
- (510) 808-2058
This research is sponsored by the U.S. Environmental Protection Agency.