Trial by fire: insurers are using technology to manage wildfire risk more effectively.

Brush fires are common in Southern California from late summer into fall. The native thickets of shrubs, known as chaparral and coastal sage scrub, become dry by late summer and many species are especially combustible. When Santa Ana winds fan the flames, these fires can turn nearby suburban enclaves into infernos of catastrophic proportions.

As property/casualty insurers struggle to fend off increasing losses from brush fire, industry experts are intensifying their focus on the factors driving those losses and the effectiveness of technology available to identify the true nature of wildfire and manage the risk more effectively. There is little disagreement that wildfires are a serious and growing threat.

Consider that the 12 most destructive wildfires in the United States between 1917 and 1983 caused a total of nearly $3 billion in property loss. Wildfires in a single month near the end of 2003 caused $2.3 billion in property loss.

For more than 50 years, government wildfire suppression policies aimed at extinguishing fires as quickly as possible have created an unintended consequence--a dramatic accumulation of brush and dense woodland vegetation, the heavy fuel needed to feed future wildfires. Without periodic natural fires or active measures to reduce fuel, the risk and potential intensity of future fires increase.

The wildfire problem has been compounded by population growth and building density in scenic but highly fire-prone woodlands--especially in California--intensifying the concentration of risk for many insurers of homes and commercial buildings. As the Federal Wildland Policy wryly notes, "These new wildland/urban immigrants give little thought to the wildfire hazard." And so in the past 20 years, wildfires have destroyed more than 11,000 homes and other insured property. The Southern California fires of 2003 alone attest to the severity of this problem.

In response, many insurers are focusing more intently on the wildfire hazard as a serious natural catastrophe requiring hands-on management. Such management has yielded new processes--automated identification of risks that are exposed to the hazard, attentive underwriting guidelines, and proactive loss control programs.

Hazard Score

Since 1997, many property writers have relied on FireLine, a geographic information system that combines 30-meter high-definition LANDSAT satellite imagery to identify fuel types along with topological maps to indicate slope and geographic detection of dead-end roads bordered by fuels that inhibit ready access by fire department equipment. This approach was initially applied in California. In its simplest form, the model calculates a hazard score based on the algorithm Fuel x Slope + Road Access Difficulty = Wildfire Hazard. That score is then complemented by a special hazard interface area determination--a climatic assessment of the likelihood that a risk lying outside, but adjacent to a heavily fueled area would ignite because of its proximity to high heat and windborne burning embers.

The FireLine model is based on research by the National Fire Protection Association, in participation with ISO. Since its introduction, FireLine has been enhanced several times. To date, insurers that write more than 60% of the property insurance in California have used FireLine to assess their exposure to the wildfire hazard. FireLine technology has been expanded to Arizona, Nevada, New Mexico, Oregon, Utah, and Washington--states where home development is spreading to brushfire-prone areas.

Following the catastrophic wildfires that ravaged Southern California in October and November 2003, experts evaluated the methodology behind FireLine. Independent data from government agencies about the geographic areas that burned and the addresses of 2,313 insured properties that were destroyed or damaged provided the statistical base for their evaluation.

As property insurers continue integrating automated solutions into their business processes, they will continue to rely on tried-and-true methods to measure, manage, and reduce risk. The wildfire hazard model based on risk proximity to fuel as well as topography and road-access problems passed a critical test in late 2003 and continues to be the foundation for effective assessment and management of exposure to wildfire.

Contributor W. M. Raicble is assistant vice president of ISO Risk Decision Services.

Key Points

* Insured losses caused by wildfires are growing due to greater population and building density in woodlands.

* A wildfire hazard model based on risk proximity to fuel and roads passed a critical test in light of California's 2003 wildfires.

THE PERIMETER OF 2003'S CEDAR FIRE IN CALIFORNIA is outlined in black and the location of the wildfire fuels identified by FireLine are color-coded. FireLine identified 97.5% of the fuels, including forested areas, brush, and tall grasses and areas that actually burned in the recent wildfires.

When fire perimeters were superimposed on FireLine fuels data, experts were able to assess geographic correlation between the two data sets. Data for the Cedar and Paradise Fires in San Diego County and the series of fires in San Bernardino County, consisting of Grand Prix, Old, and Padua Fires, revealed a near-perfect match.

This map illustrates the correlation between the Cedar Fire and FireLine fuels. Of 424 square miles that burned in the Cedar Fire, FireLine classified 416 square miles, or 98.3%, as fuels. The analyses of FireLine's behavior against the burn areas of the other fires of 2003 showed similar results.

THIS VIEW OF A SOUTHERN-FACING AREA OF SAN BERNARDINO was photographed in November 2003. Arrows indicate the destroyed properties, which were located outside traditional wildfire fuels areas but inside FireLine's hazard interface area. FireLine identified 95.7% of the homes and commercial properties affected by the fires as exposed to the wildfire hazard. Virtually all affected properties--99%--were within 2,500 feet of FireLine's heavy or medium fuels.

ISO conducted field research that revealed wind-borne burning embers carried outside of the actual wildland fuels area caused 295 of the property ignitions studied in the California fires. FireLine's "special hazard interface area" model proved effective in identifying the vast majority of these properties, which would not have been detected using a fuels data set alone.