Summer  >>  2005

Figure 1 >> Several communities in the Golden Gate FPD with the false color IKONOS imagery and the hazard ratings derived from the assessment
Burning Solutions

Solving America’s Worsening Wildfire Dilemma

Manager, Business Solutions Division
The Sanborn Map Company
Colorado Springs, Colo.

Managing PartnerAnchor Point Group LLC
Boulder, Colo.


Colorado was burning.

On June 11, 2002, no fewer than 12 major wildfires were ripping across the state, sending the stench and haze of smoke across the West and into cities as far away as Chicago. The biggest blaze, the 138,000-acre Hayman Fire, would grow so large so quickly that it would demand the oversight of not one, but two of the nation’s 12 Incident Management Teams. In a single afternoon, the Hayman’s flames would make a stunning 60,000 acre run, a stark reminder that America had a profoundly serious wildfire problem.

Others around the country had already had similar wake-up calls: Two years earlier, it was the folks in Los Alamos, New Mexico, who had seen more than 400 of their homes burn. The next year, residents of San Diego experienced a firestorm that wiped out nearly 2,000 homes and killed more than 20 people.

Still, if there is a silver lining in this dry cloud, it’s the fact that in recent years, government land managers and politicians are in agreement: A century of forest mismanagement and overzealous suppression policies, which removed healthy fire cycles from the ecosystems, have made America’s forests into tinder boxes.

The sheer magnitude of the problem is the most immediate obstacle to its solution. The ‘wildland-urban interface’ comprises more than 100 million acres of overgrown land filled with millions of homes and many new communities. Just as daunting is the fact that all forests are not alike — a grass fire burns quite differently than a thick ponderosa pine forest. So assessing accurate wildfire hazards and risks on a large scale has been generally a complex, man-power-intensive, inconsistent, time-consuming and costly exercise.


Fighting wildfires has become an expensive fiscal responsibility. In five of the past six years, suppression costs at the federal level alone have surpassed $1 billion annually, an unprecedented and problematic fact. With combined state and local costs believed to be more than double that amount, the bottom line of fighting fire is an enormous figure.

Some of the worst fire seasons on record have occurred in the past five years alone and in highly populated communities. For the first time, reliable studies are demonstrating the long-term fiscal impact of wildfires. A September 2004 study of the Hayman Fire found that, more than a year after the fire, Hayman-impacted counties like Jefferson and Douglas, as well as agencies like the Colorado Department of Transportation, were still paying hundreds of thousands of dollars for a variety of persistent costs, from mud slides to road repair to watershed contamination.

The fiscal realities have led to a deep rethinking about prevention and preparedness planning, best exemplified by Washington’s 2001 $1.8 billion National Fire Plan and the 2003 Healthy Forests Restoration Act (HFRA). Both directives carry accompanying funding for prevention programs.

The HFRA includes a congressional directive whereby the 23,000 communities at risk of a wildfire event must develop a comprehensive Community Wildfire Protection Plan (CWPP). Additional money was provided through FEMA firefighter programs to assist those communities with their CWPPs, and in less than two years, the CWPP has evolved into a nationally-recognized process to save America’s threatened communities from devastating wildfires.

Figure 2 >> IKONOS false color imagery with the image segmentation results that identify homogeneous fuels on the landscape (yellow)
While the CWPP is an earnest and solid start to solving the wildfire problem, it has its own dilemmas. The methodology of the CWPP requires the convening of key community decision-makers for the establishment of community hazard reduction priorities and recommendations. Also in the CWPP directive is the development of community risk assessments based on such key information as accurate fuel models and the likelihood of a wildfire event.

In 2003, two Colorado companies teamed up to develop an effective, scientifically sound and financially reasonable answer for governments, landowners, and corporations — The Sanborn Map Company Inc. (Colorado Springs, Colo.) and Anchor Point Group LLC (Boulder, Colo.). They approached this problem with a logical premise: Complement sound geospatial technologies with robust fire science and operational fuels and fire planning knowledge, and provide a solid basis for mitigating risk and preparing communities for fire incidents. In the fall of 2004, that unprecedented effort was finally accomplished.


GIS and remote sensing, especially for fuels mapping, have been applied successfully at the regional level and strategic scale for several years. In 2002, just as the new ‘prevention’ thinking was getting a foothold in federal and state governments, the State of Florida Division of Forestry successfully completed a statewide fuels mapping and wildland fire risk assessment project that quantified the fire situation across the state. The results provided a sound data platform for Florida fire professionals to plan their long-term strategies for reducing hazardous fuels and risk in fire-prone areas.

The Florida project raised the bar of fire prevention to an unprecedented level of accuracy and consistency by leveraging geospatial technologies, along with fire-science and operational expertise for fire planning. In the end, Florida became the first state to satisfy the requirements of the implementation plan for the federal government’s 10-Year Comprehensive Strategy (Task E, Goal 4), a state and federal multi-agency effort laid out in May 2002 for wildland fire prevention planning across the nation.

Still, while GIS and remote sensing play a critical role in providing detailed data for fuels, structures and physical features, the critical ingredients to providing real wildfire solutions are on-the-ground, wildland-urban, interface-based risk assessment and fire planning. Executives at Anchor Point Group and Sanborn Map Company saw the extraordinary opportunity to blend their technological and subject-matter expertise to provide an unparalleled level of quality fire-risk assessment and planning solutions.

Figure 3 >> The final fuels map developed for Golden Gate FPD
The partnership couldn’t have come at a better time. Government agencies at all levels are struggling to define the science and methodology behind the CWPP. While several organizations such as National Association of Counties, National Association of State Foresters, and International Association of Fire Chiefs, among others, have attempted to set guidelines for CWPP development, to date there are still no hard-science standards coupled with real fire solutions. As a result, with approximately 600 CWPPs completed nationwide to date (about two percent of the total needed), quality and consistency are already becoming issues.

In an effort to bring continuity and consistency to the CWPP, the Anchor Point/Sanborn team turned to local Colorado agencies to demonstrate its revolutionary methodology. With funding from the Colorado State Forest Service and Jefferson County, the Anchor Point/Sanborn team executed a pilot project for a small community outside Denver and in the fall of 2004, completed an unprecedented CWPP.


The Golden Gate Fire Protection District (FPD) is in the heart of fire country in Colorado, located at the doorstep of the Rocky Mountains just west of Golden, 15 miles from downtown Denver. The area studied by Anchor Point/Sanborn encompasses Golden Gate Canyon State Park and the Golden Gate FPD, a total of 49 square miles with about 300 homes located in the center of the wildland-urban interface.

The project utilized high-resolution imagery combined with field surveys to derive a highly accurate fuels map for the area of interest. The fuels map is the cornerstone of the fire-risk assessment. IKONOS 4-meter resolution multi-spectral imagery was used, collected on August 13, 2004, covering the 360-square kilometer area. This imagery is orthorectified to a precision level product (4m CE90, 1:4,800 NMAS) and contains both1-m panchromatic and 4-m 4-band (R, G, B, NIR) multi-spectral information, creating a spectrally balanced and seamless mosaic.

High-resolution imagery supports accurate integration with other GIS data sets, facilitating use for the fuels mapping to be conducted in this project and for future assessment and fire planning efforts. Golden Gate FPD comprises several communities shown in Figure 1 (page 27), with the false color IKONOS imagery for the area and the hazard ratings derived from the assessment.

As part of the imagery classification process, in collaboration with Sanborn, Anchor Point was sent to the study area to calibrate what appeared in the imagery with the actual vegetation types on the ground. While at the study area, the Anchor Point team identified and classified training sites that were put into the automated classification process.

The fuels map was then generated by applying image classification techniques. The initial output was a general vegetation map, which was then combined with Canopy Closure to generate the draft surface fuels model. Surface fuels models are traditionally mapped to 13 Northern Forest Fire Laboratory (NFFL) fuel models described originally by Albini and further described by Anderson (Anderson 1982). A total of 13 fuel models fall within four basic groups. See Table 1.

While doing the ground-truthing and field-accuracy assessment, Anchor Point fuels specialists conducted a manual fuels assessment. Survey points were captured utilizing a standard, digital methodology to facilitate data transfer. The manual fuels data were used to refine the classification and to independently support the accuracy assessment.

This project allowed the Anchor Point/Sanborn team to test new methods for determining surface fuels with the benefits of finer spatial resolution and feature delineation. Sanborn leveraged techniques it has pioneered for land-cover and impervious mapping using high-resolution imagery, but prior to this project, they had never been applied to fuels mapping.


The study area was mapped to the 13 Albini/Anderson NFFL fuel models.

Figure 4 >> The Fuels Interactive Map page of the Golden Gate CWPP GEOBOOK
Due to the diversity of ground cover within fuel-model polygons, the classification was performed on polygons derived from image segmentation, rather than from a per-pixel classification.

A benefit of the polygon classification is the utilization of derivative bands that characterize the general area. Use of texture, slope and aspect on a polygon level is more beneficial than on a per-pixel classification, but a polygon classification lacks the spectral diversity of the vegetation within a polygon. A general per-pixel classification was performed to minimize the loss. Figure 2 shows the IKONOS false color imagery along with the image segmentation results that identify homogeneous fuels on the landscape (yellow).

The fuels mapping deliverables consisted of two ArcView shape files representing Canopy Closure and Fuel Models. The final edited fuels map polygons were dissolved based on the fuel model, and a one-acre minimum mapping unit was applied. The final map consisted of 5,318 polygons averaging about 15 acres in size. Figure 3 shows the final fuels map that was developed for Golden Gate.

The second deliverable was a percent canopy closure map which was calculated from the pixel classification. The percent canopy was calculated for each polygon and grouped into four classes: 1-20 percent, 21-40 percent, 41-60 percent, and > 60 percent.

A fuels accuracy assessment was also undertaken through methods that were originally developed in support of land cover mapping methods for USGS and have become the standard approach for conducting accuracy assessments in the mapping industry. Quantitative accuracy assessment of maps produced from remotely sensed data involves the comparison of a map with reference information. In the end, Anchor Point and Sanborn achieved an overall accuracy of 83.2% for the Golden Gate map.


Using the highly accurate fuels data, a hazard and risk assessment was undertaken at the community scale by Anchor Point fire planners to develop the CWPP and mitigation recommendations, which were then used in community meetings and public outreach providing Pre-Attack Plans for first responders, and Annual Work Plans for long-term ecosystem management programs.

To aid in delivery and distribution of the CWPP to interested parties, the results were encapsulated in Sanborn’s GEOBOOK map book product. The GEOBOOK is an intuitive GIS application that is provided in an easy-to-use digital book format with no software royalty fees.

The real power of the GEOBOOK lies in the Interactive Map pages where real time GIS querying and mapping are available. Operating much like ESRI’s ArcView GIS, the Interactive Map page provides the reader real-time access to the CWPP GIS data. The GEOBOOK provides a comfortable and intuitive mechanism for non-GIS audiences like local fire fighters to access the power of GIS data within the context of a sound, scientific fire plan. It can be customized to support content such as key photographs, individual home assessments, local fire policies and contacts, Pre-Attack Plans, general methodology information, FireWise guidelines, and more.

The Fuels Interactive Map page of the Golden Gate CWPP GEOBOOK provides a range of GIS tools to browse and query the fuels data in concert with other datasets, such as community infrastructures, fire behavior analysis results, and proposed mitigation recommendations. Book readers can quickly zoom to a community area using pull-down menus, or query the map for areas of interest. See Figure 4.


Prolonged drought and dismal snow pack in the Pacific Northwest have led most fire analysts to agree that fires in this area will be the big stories of 2005. As fire-prone communities all across the West head into the 2005 season, many of them are earnestly looking to the future, knowing that the time may come when they must prove their preparedness. The Sanborn and Anchor Point teams’ efforts have left little doubt that geospatial technologies in fire management and planning are the keys to the development of effective, scientifically based CWPPs.



Anderson, Hal E. 1982. Aid to Determining Fuel Models for Estimating Fire Behavior. USDA Forest Service Gen. Tech. Report INT-122, p. 20.

Story and R. Congalton. 1986. Accuracy Assessment: A User’s Perspective. Photogrammetric Engineering & Remote Sensing, 52(3): pp. 397-399.

Congalton, R. 1991. A Review of Assessing the Accuracy of Classifications of Remotely Sensed Data. Remote Sensing of Environment, 37: pp. 35-46.

Lynch, Dennis 2004. What Do Forest Fires Really Cost? Journal of Forestry, September 102(6): pp. 42- 49.

Sensors & Systems | Monitoring, Analyzing and Adapting to Global Change | Stay in tune with the transformation. Subscribe to the free weekly newsletter.