But How Good Are They?

While the BEAR teams revise the satellite maps based on field observations, their primary concern is conducting their own emergency assessment. They have neither the time nor the resources to provide any kind of systematic feedback to the satellite mappers to help them improve their maps. In search of something more scientifically defensible, the RSAC has partnered with scientists at the University of Maryland and obtained funding from the National Interagency Fire Center’s Joint Fire Science Program to collect their own field data on burn severity conditions, which they are using to check the accuracy of the satellite classification, a process called validation.

Sohlberg is one of the scientists who spent some time in the field this summer. Since BAER teams are often on the ground before fires are totally contained, Sohlberg had to become certified as an entry-level firefighter in order to do this work. A lot of the training was in the classroom, said Sohlberg, but they also had to learn how to dig fire lines, operate pumps and hoses, and how to maintain their tools. They had to participate in timed drills to see how quickly they could set up an emergency fire shelter using the distinctive shiny Mylar bags. Said Sohlberg, “We were out there to collect data, but we always had to remember to be on guard; and we had to carry full personal protective equipment: fire-resistant pants and shirt, 8-inch logging boots, fire shelter, compass, water, goggles, helmet and leather gloves. Although we were never in contact with active fire, there were still lots of hazards—damaged trees that are prone to falling, called snags; rolling rocks; and smoldering stump holes.”

In the field, the team uses a mapping-grade Global Positioning System (GPS) receiver that has a built-in data dictionary. Once the GPS device has pinpointed the scientists’ location, they can access the data dictionary, which prompts them with a series of preset questions about the burn severity at the site. Is there any ground cover left? How deep is the ash, and what color is it? How long does a drop of water sit at the surface before soaking in or running off? Are there any green trees left? Are there any needles or leaves left on the trees? What is your overall assessment of the burn severity at this location? Since a satellite maps the whole area, the team tries to collect data to match up with all the different burn categories within a fire perimeter. If the physical geography of the terrain is highly variable, for example because of topography or ecology, the team tries to collect data for each burn category in each of the different physiographic regions within the fire perimeter. That’s a lot of different categories.
 

Physiographic differences can be tricky for satellite classification. Says Parsons, “You have to be initially cautious about the satellite classifications, particularly in areas of diverse terrain or varying geology. Fire may quickly pass through a rocky area with already sparse vegetation, or grassland, and blacken everything. Or it might burn over a forested ridge, consuming acres of trees and all the understory vegetation. From a satellite, all these areas might look the same—just black—and might be classified as high severity. But the actual severity as far as the impact on the watershed would be very different. In the first case, the post-fire runoff and erosion would probably not be substantially increased over pre-fire conditions, and the vegetation would probably recover in a season. The effects could hardly be called severe. In the second case, the impacts would be much more severe.”
 

The RSAC has provided satellite-based burn severity maps for BAER teams for two fire seasons: 2001 and 2002. Parsons says that their experience suggests that the remote-sensing-based assessments are most accurate in more forested ecosystems. In that kind of terrain, they can often use the satellite maps from the RSAC without any major modification. In shrubby or rocky ecosystems, they require more fine tuning from ground-truth. Even so, Parsons says, her main criteria for the project’s success is less in how technically accurate the maps are and more in their utility to field crews. “We have produced satellite maps for at least 70 fires in the 2002 fire season, and the word we have gotten back from BEAR field teams is overwhelmingly positive. They may not be 100 percent accurate at fine scale, but they have been very useful in pointing teams to areas of greatest concern.” Using those criteria, the project can already be called a success.

That won’t stop the scientists from continuing to tweak their approach. Over the winter, Forest Service and University of Maryland scientists will look for ways to increase the accuracy of the satellite maps using data collected this past summer at the Missionary Ridge Fire in Colorado, the Rodeo-Chediski Fires in Arizona, the East Fork Fire in northern Utah, and the Biscuit, Winter, Toolbox and Eyerly Fires in Oregon. NASA is supporting that work as part of its Natural Hazards program. The team will also be preparing for another season of data collection in the summer fire season of 2003.

The group also plans to take advantage of its involvement in NASA’s newly created Wildfire Response Team to coordinate the logistics of acquiring imagery from airborne sensors. The team has their eye on the MAS (MODIS Airborne Simulator) on NASA’s ER-2 aircraft and the AIRDAS (Airborne Infrared Disaster Assessment System), developed by NASA’s Ames Research Center and the Forest Service to fly on a variety of aircraft. They also hope to make use of another Terra satellite instrument, called ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer). ASTER offers unique capabilities in sensing shortwave infrared radiation, which makes it particularly useful for seeing through smoke. Increasing the number of data sources increases the team’s opportunity to produce useful images and burn severity maps to BAER teams and also increases opportunity for cross-comparisons of all the remote-sensing data, which serves as means of validation in itself.
 

This partnership is not the first for the RSAC, NASA, and University of Maryland. The three originally teamed up as part of the MODIS Land Rapid Response Project, in which daily imagery and fire detections from MODIS were relayed from NASA’s Goddard Space Flight Center, to the University of Maryland, and on to the RSAC for use by the National Interagency Fire Center as it allocated firefighting personnel and resources across the country. That collaboration ultimately led to the installation of a MODIS Direct Broadcast Receiving Station at the RSAC and paved the way for the current work. With solid validation data, the satellite-based burn severity maps produced in coming seasons will only get better, saving BAER crews precious time as they assess the dangers left behind in the wake of wildfires.

For more information, visit:
  USDAFS Burned Area Emergency Rehabilitation, BAER
  USDAFS Remote Sensing Applications Center

Ongoing BAER activities at summer 2002 fire sites:
  Missionary Ridge Fire
  Hayman Fire
  Biscuit Fire
  McNally Fire

 Satellites Do It Faster, Cheaper