Satellites Do It Faster, Cheaper

  Page
1Page 3
 

Given the time constraints, remote sensing data have great potential to assist BAER teams in their assessments. Over the past decade, scientists at the USDA Forest Service’s Remote Sensing Applications Center (RSAC) in northern Utah have been developing techniques for using airborne and satellite-based instruments to map out burned landscapes. Mark Finco is a senior scientist specializing in remote sensing and GIS for the RSAC. He described the group’s BAER activities as a long-running project that began 8 or 9 years ago. The group’s initial efforts were in the development of a digital infrared color camera that could be mounted on an aircraft platform and flown over fires. Although the camera produces good imagery, it’s expensive, and it takes a lot of effort to map a large area. Over the last three years or so, the RSAC has been investigating how satellite-based images could be used in the BAER analysis.

   
 

 
MODIS Image of the Missionary Ridge Fire
 

 

The Multi-sensor Advantage
In gathering useful satellite images, the RSAC has to be flexible.Satellites travel in fixed orbits and observe specific areas on a prescribed schedule. If a satellite is set to pass over Idaho on a given day, scientists can’t change the orbit just because southwest Oregon is on fire. According to Finco, they use data from whatever sensor has the best look at the area at the time of the BAER assessment. They use NASA satellite sensors like the Enhanced Thematic Mapper Plus (ETM+), aboard Landsat 7, and MODIS (the Moderate Resolution Imaging Spectroradiometer) on the Terra satellite. They have also used the commercial satellites Ikonos and Quickbird, as well as France’s SPOT satellites. With all satellite data, a trade off must be made between how much surface detail can be collected, called spatial resolution, and how much area can be observed at a given time, called spatial coverage.

 

The burn scar of the Missionary Ridge fire appears dark red in this Moderate Resolution Imaging Spectroradiometer (MODIS) image from August 10, 2002. With a maximum resolution of 250 meters per pixel [the image above is 135 km (84 miles) across] and coverage every 1 to 2 days, MODIS provides frequent observations of large areas. The scientists use higher-resolution satellites like Landsat to map the burned area in detail. The white box in the image above indicates the subsetted Landsat scene shown below. (Image by Robert Simmon and Jesse Allen, NASA GSFC)

 

 
Landsat ETM+ image
 

 

Geographer Rob Sohlberg with the University of Maryland describes how multiple satellite sensors are used in satellite mapping of burn severity. “In general, the higher the sensor’s spatial resolution, the less likely it is that it will be observing the area we need at exactly the right time for burn severity mapping of a specific fire.” Higher resolution data are also inevitably more expensiveto obtain. “With a sensor like MODIS,” Sohlberg continues, “the coverage is much better—we see almost the entire surface of the Earth every day—but at coarser resolution.” These coarser- resolution data aren’t ideal for mapping out such detail as which hillside, creek, or ravine is especially damaged, but they can be useful for very preliminary mapping of burned areas, and perhaps more importantly, they can help scientists to decide in which areas high-resolution imagery is most needed.

 

The Enhanced Thematic Mapper plus (ETM+) aboard Landsat 7 has a true-color resolution of 30 meters per pixel—much better than MODIS—but only covers an area every 9 to 16 days. Smoke partially obscures the burn scar in this true-color image, making it difficult to see the patchy nature of the burn. Infrared wavelengths of light measured by Landsat penetrate smoke, and false-color images made using data from the infrared portion of the spectrum reveal the structure of the scar. (Image by Robert Simmon, based on data provided by Andrew Orlemann, USDAFS)

 

 
IKONOS Image
 

 

Annette Parsons has worked for more than 20 years as a soil scientist and mapping specialist for the Forest Service, and has been involved with BAER activities for more than a decade. Parsons was quick to see the potential of the RSAC satellite-based products and is now serving as a liaison between the RSAC and the BAER teams that operate on the ground. Says Parsons, “One of our main goals is to get local forest officials to notify the RSAC as soon as they think a fire might end up needing a BAER assessment. The earlier we know, the better our window of opportunity for finding the best satellite imagery for the area and having it in the team’s hands when they arrive at the field location.” Parsons agrees that MODIS’ coarser resolution provides less detail than Landsat or SPOT, but she says, “You can count on it almost every day. And in very large fires, such as this summer’s Biscuit Fire in Oregon, coarser-resolution data may be all we get for some areas.”

Finco identifies another use for the MODIS products. “All these burn products have the potential to be used for carbon budget investigations.” Fires release into the atmosphere carbon that is stored in trees, plants, and even the soil if the fire is intense and long-lasting. Regrowth draws carbon back in and stores it in plant matter. “People tend to think that a burn perimeter on a map means everything within that perimeter was totally burned. But the vast majority of the terrain is a mosaic of areas showing different burn severity. If you understood this mosaic, it would improve carbon emission estimates.” MODIS data are well suited for this application, especially when the fires are in the several-hundred-thousand-acre category.

 

Detail down to the level of individual trees is revealed by Space Imaging’s IKONOS satellite, which produces color images at 4-meter resolution. This true-color image, acquired June 23, 2002, shows dark gray, burned area in the upper left, and green, unburned area in the lower right. The bright blue-white smoke plume is caused by a smouldering hotspot, and thin smoke covers much of the rest of the scene. (Image copyright Space Imaging, based on data provided by Andrew Orlemann, USDAFS)

 

 
Preliminary Burn Assessment
 

 

Using a combination of sensors allows the RSAC team to make burn severity maps quickly and relatively cheaply—especially compared to the cost of hours of helicopter time, which is the traditional reconnaissance source. It may seem trivial, but satellites have another advantage over sketch mapping from helicopters, which Finco and Parsons are both quick to point out. Leaning out of a moving helicopter as it weaves back and forth over a burned area, looking out to the terrain and down to the map in your lap not only makes it hard to be sure where you are, but also makes it hard to keep down your lunch!

 

Burn severity maps are produced by assigning the burned landscape into one of four categories based on characteristics like damage to trees and other vegetation, soil hydrophobicity, and ash color and depth. This map of burn severity for the Missionary Ridge Fire is based on preliminary satellite data. Before making a final map, scientists must carefully compare the post-fire effects to the unique pre-fire landscape. (Image courtesy Monte Williams, USDAFS)

 

 
Photo of Burned Forest from a Helicopter
 

 

Assessing Burn Severity from Space
When a BAER team hits the ground, they don’t have the time or resources to do a full canvass of the entire area affected by a fire. This is especially true as the size of the affected area increases, or in cases where significant portions of the fire burned in roadless or wilderness areas The team must decide in what areas to focus limited field time. The goal of the RSAC is to have the preliminary satellite-based maps in the hands of the BAER field teams as they are planning these decisions. Says Parsons, “The satellite burn severity map serves as a fast-track emergency product that guides the field team, helping them decide where to go. The field team then refines the maps based on ground observations.”

 

This aerial view shows the differences between low, moderate, and high burn severity. Green trees are in low burn severity areas, brown trees with dead needles are located in moderately burned areas, and black, needleless trunks have been severely burned. Gound level images of similar areas are on page 3. (Image courtesy Annette Parsons, USDAFS)

 

 
NDVI and NDBR
 

 

The way in which the maps are produced depend largely on what sensor provides the data, but most involve using vegetation characteristics as an indicator of soil conditions. Generally, the maps are based on an indicator called NDVI, for Normalized Difference Vegetation Index, or one called NDBR, for Normalized Difference Burn Ratio. NDVI is based on the principle that vegetation absorbs red light and reflects near-infrared light. Changes in the amount of each kind of energy reflected from the surface can signal how severely the pre-existing vegetation has been transformed by fire. NDBR is based on the observation that vegetation reflects energy in the near-infrared, while soil tends to reflect energy from the mid-infrared part of the spectrum. After a burn, the vegetation within the burned area has been reduced and the soil has been exposed. As a result, the near-infrared values are lower than before the fire, while the mid-infrared values are higher. These NDVI and NDBR relationships are used to classify areas in a satellite image into one of four burn severity categories: high, moderate, low, and unburned.

next But How Good Are They?
back BAER

 

Scientists estimate burn severity with satellite data by comparing the signal from different wavelengths of light. Near-infrared and red light are used for the Normalized Difference Vegetation Index (NDVI), which indicates the density and health of vegetation. Dark areas in the NDVI image (left) represent unvegetated or burned areas. Normalized Difference Burn Ratio (NDBR) uses the same math as NDVI, but near-infrared and shortwave wavelengths. NDBR highlights areas of exposed soil [bright areas in the NDBR image (right).] Data collected after a fire are compared with pre-fire data to produce burn severity maps. Before these maps are finalized, teams must survey the burned areas from the ground. (Images by Robert Simmon, based on Landsat 7 data provided by Andrew Orlemann, USDAFS)