Fires set by people are a real problem for wildland fire managers on all types of land ownerships, including tribal lands. Because they usually occur closer to valued property and resources, human-set fires also tend to be more damaging than fires ignited naturally. Human-ignited wildfires fall into two categories – incendiary, or intentionally set fires, and those started accidentally.
U.S. Forest Service and National Institute of Standards and Technology (NIST) scientists recently found that tools designed to forecast incendiary wildfire hotspots could lower the value of the damages and the costs of suppressing wildfires on tribal lands. The findings by Forest Service Southern Research Station project leader Jeff Prestemon and co-authors David Butry and Douglas Thomas from NIST were recently published online in the International Journal of Wildland Fire.
The primary insight of earlier research by these scientists is that human-ignited wildfires are often clustered in space and time, patterns that could enable effective forecasts of future wildfires. However, the authors point out that economically effective forecasts also require that the costs of forecasting be less than the benefits that the forecasts would provide.
Wildfires in general occur in clusters for a variety of reasons, most commonly because fuel and favorable weather conditions also tend to cluster. More critically, because human wildfire-igniting events cluster where human activities cluster, wildland managers and law enforcement have the opportunity to take actions to reduce future wildfire occurrences.
It’s been known for some time that human-ignited wildfires tend to cluster, but Prestemon and Butry were the first to document that arson wildfires tend to cluster in both time and space in patterns distinct from those of both natural and accidentally set wildfires. “In cases of human-set wildfires, fire clusters or hotspots can result from serial or copycat wildfire setting behavior or other factors,” said Prestemon.
Using a model to forecast hotspots could reduce both the damages from fires and the costs of suppressing them. For the study on tribal lands, the researchers examined a total of six models, some for incendiary fires only, others including other human-ignited fires. They developed forecast models that included information on previous weeks’ intentional and accidental wildfire occurrences on 23 tribal land units. All models were then tested for their economic effectiveness by using existing data on wildfire damages, suppression, the cost of law enforcement actions to respond to a cluster forecast, and the costs of tool development and deployment.
Tool development cost is the most critical element in determining economic effectiveness. “We estimated a high initial cost of developing a tool at $50,000 per model per tribal land unit,” said Prestemon. “But in principle, the models could be developed to be used by all the units, with the cost shared among them, which would bring that expense to a much lower level, perhaps as low as $2,000. This would make it more economically justifiable to deploy such tools for the land units we examined.”
The researchers found that:
- Both incendiary and other human-ignited wildfires on U.S. tribal lands demonstrate statistically and operationally significant clustering at the weekly time scale;
- Statistical models recognizing the clustering had forecast skill, which means that they predicted human-ignited wildfire clusters accurately in out-of-sample conditions, and the simplest models showed the highest forecast skills;
- Because the incidence of intentionally set wildfires responds more strongly to law enforcement efforts than those of accidentally set fires, a hotspot tool for intentionally set fires is more likely to yield benefits than one for all or other human-ignited wildfires;
- Prospective hotspot tools are more likely to yield net benefits in areas with larger and more frequent wildfire clusters; and
- When choosing a wildfire forecasting tool based on economics, managers should consider the underlying wildfire risk. More sophisticated models might be more appropriate for high-risk, wildfire-prone areas.
Overall the researchers found that a hotspot tool designed to help reallocate law enforcement resources probably will yield benefit for incendiary fires but not for accidentally set fires, which are not responsive to law enforcement.
“However, law enforcement is not the only tool available to bring down occurrences of human-ignited wildfires,” said Prestemon. “Previous Forest Service research on tribal lands has shown a significant response to educational forms of wildfire prevention, so using a forecast tool coupled with prevention education tools for non-incendiary human wildfires might yield more positive net benefits.”
For more information, email Jeff Prestemon at firstname.lastname@example.org.