GeoWorld Articles Mapping a Firewall: Modeling and Visualizations Assess Wildfire Threats, Risks and Economic Exposure

Further Reading for Map Analysis Understanding Spatial Patterns and Relationships (Berry, 2007 GeoTec Media)

Feature article for GeoWorld, October 2009, Vol. 22, No. 10, pgs. 20-23

<Click here> for a printer-friendly version of this paper (.pdf).

In the 2007 fire season, San Diego County alone saw 360,000 acres burned, more than $1 billion in losses, more than 1,200 homes destroyed, many buildings and critical infrastructure lost, and significant amounts of commodity agriculture ruined. Suppression costs at the federal level have surpassed $1 billion annually for the last several years, and state and local costs are believed to be more than double that.

The consequences of wildfires have never been greater as more people move into wildfire-prone areas. And there’s an increasing need for fuel treatments, mitigation planning, prevention awareness and recovery preparedness to reduce wildfire risk and impacts to these communities.

But where is the greatest risk? What are the potential economic, social and environmental impacts? What and where are mitigation actions most needed? How can alternatives be quantified, compared and prioritized? Are we spending our budgets effectively and efficiently?

This article focuses on the utility of geotechnology, map-analysis procedures, and Web-based visualization and delivery options to identify areas of greatest jeopardy as well as quantify the dollar impact of wildfire loss and proposed mitigation efforts.

Wildfire Threat and Risk Modeling

Previous wildfire risk models developed a relative scale, such as the low, medium, high and extreme fire-danger levels seen at the entrances of national forests. Although this scale is useful for informing the public and guiding broad fire planning, it doesn’t fully express wildfire risk. Comprehensive risk modeling involves three distinct elements:

1) Wildfire Threat—estimating the probability and intensity of a wildfire occurring at a location.

2) Wildfire Effects—quantifying the impact of the potential loss.

3) Wildfire Risk—combining the threat and effects into a measure of probable loss over time.

The Wildfire Threat portion integrates numerous mapped data layers such as weather factors, historical fire occurrence, surface and canopy fuels, terrain, and suppression effectiveness based on historic fire protection (see Figure 1). A previous GeoWorld article (Quantifying Wildfire Risk, December 2005) described the fundamental approach and data layers involved in spatially modeling wildfire threat.

Note that Wildfire Effects is subject to change based on the characteristics and priorities of the specific geographic area. As such, Figure 1 only provides an example of common fire-effects inputs. The current enhanced model had input from an actuarial statistician and a risk-modeling expert with considerable experience in risk mapping for the insurance industry. The modifications incorporated advanced techniques, such as dynamic elliptical windows for calculating wildfire probability based on fire-behavior parameters, adjustments for urban-area partial windows and refinements for handling non-burnable areas.

This expanded perspective fully integrates remote sensing, current fire-science research, actuarial statistics and GIS expertise. The solution involves vector and raster data layers and processing procedures as well as integration with the standard LANDFIRE Program datasets. As a result, the output maps are useful to a broader group of users, ranging from traditional wildfire professionals to county land-use planners, insurance industry agents and all levels of government decision makers.

Figure 1. A flowchart depicts the key components of Sanborn’s Wildland Fire Risk Assessment System.

Visualizing Wildfire Risk Outputs

Traditionally, GIS has been used to display the outputs of models, such as wildfire risk, using desktop software applications. Recent advancements have led to the delivery of thematic maps using Web-mapping interfaces on the Internet (although there are few examples for wildfires).

With the advent of 3-D globes and related public Web-mapping capabilities (e.g., Google Earth, Microsoft’s Virtual Earth (Bing! Maps) and ArcGIS Online), the public and professionals now have an expectation of Web-mapping capabilities and availability. This explosion in Web mapping with multi-resolution imagery backdrops has made the consumer “spatially aware” and set the baseline for delivering Web-mapping products.

Figure 2. A Virtual Earth visualization of Wildfire Threat maps provides interactive access and processing for a variety of fire professionals, land planners and the general public.

Figure 2 shows an example of a wildfire threat map superimposed on the terrain for Boulder, Colo., using public Web-mapping capabilities in Microsoft’s Virtual Earth map interface. Capabilities exist to integrate thematic risk maps with the underlying imagery-based map interface, including enhancements that show real-time weather information, such as cloud cover or NEXRAD data.

Figure 3 shows wildfire risk outputs from the Southern Wildfire Risk Assessment superimposed over the perimeter of the recent Highway 31 Fire in South Carolina. The prototype uses ESRI’s Silverlight interface for ArcGIS Server combined with ArcGIS Online imagery services. The integration of active or real-time data provides greater context for using wildfire risk assessment data, providing tactical utility in addition to conventional planning uses.

The maps can be served via the Internet and accessed by a variety of users: the general public, county planners, and emergency-response and wildfire professionals. By varying the transparency of the wildfire risk output layers, the relative visual prominence of the underlying land cover and features can be adjusted.

This easily accessed format facilitates a user’s ability to “fly through” the wildfire threat information, zoom in to an area of interest and assess the relative patterns within a context of its surrounding conditions and features. The threat values can be expressed as traditional wildfire danger ratings (low, moderate, high) or as the discrete probability of a wildfire occurring.