Aerial Photo/GPS :High-Tech Desert Quest
Using a combination of GPS, multi-spectral airborne video data and photography, researchers at Joshua Tree National Prak attempt to track down the unpredictable Desert Tortoise.
By Jerry Freilich, Ph.D.

When the U.S. Fish and Wildlife Service listed the Desert Tortoise as ÒthreatenedÓ in 1990, it charged land managers with counting the tortoises, documenting population levels, and protecting their habitat with the goal of reversing the downward trend that got them listed. Joshua Tree National Park is an important home to the Desert Tortoise. Important because human activities such as urban sprawl, trash dumps, livestock grazing, and poaching have been implicated in the tortoisesÕ decline. Joshua Tree National Park includes 800,000 acres, most of which have been completely protected as national park land since 1936. These nearly pristine acres represent a control against which impacted areas can be compared.
      The trick is to find the tortoises in order to count them. At the present time, the only method that works is to walk every inch of the desert looking for the tortoises, their burrows, or their droppings. Because tortoises are locally clumped and rare relative to the vastness of the desert, a great deal of time is wasted looking in areas without tortoises. Then, too, tortoises can stay underground and Ôsit outÕ several consecutive years if winter rains do not materialize and plant food is sparse. Adding to the mess, tortoises may be either abundant or completely absent in adjacent areas which look almost identical to human (but apparently not to tortoise) eyes. Is there some clue as to what makes good tortoise habitat? Or is there some signature which reliably shows tortoises to be absent?
      At Joshua Tree National Park, the Division of Resources Management is coordinating several research projects focused on the problem of identifying the tortoise habitat. The need is immediate. Each spring, teams of observers are sent into the field to find and count the tortoises as mandated by the U.S. Fish and Wildlife Service. Any method that could target areas for search, or simply remove areas from consideration would represent a huge savings in effort. This is a spatial problem calling for GPS solutions. Currently, we are working in different parts of the electro-magnetic spectrum, and in different spatial scales, to search for tortoises. All of these efforts are tied together by global positioning.
      First, on the ground, whenever tortoises are found, we obtain GPS carrier-phase fixes on their locations. Using Magellan ProMARK X receivers, this technology is capable of post-processed accuracy of better than 1-meter. But in practice, we use the less accurate unit-mounted antenna, donÕt bother measuring antenna height, and are quite content with the plus or minus 3-meter accuracy this provides. Every GPS point is entered into our Macintosh-based tortoise database, showing each of our study plots in map view. Points representing tortoises are shown as dots superimposed on an aerial photo. When the mouse pointer touches a dot, its GPS location is displayed. Clicking with the mouse on a dot opens a separate window showing size, sex and weight information on that tortoise, a color video image of the animal, and a color image of that spot on the ground.
      Our goal of identifying tortoise habitat also include variables such as soils, micro-topography, and vegetation. Teams of observers on the ground take soil and vegetation samples on sites with known tortoise density. As with the tortoises, data in the field are GPS fixed to plus or minus 3-meters so that each layer is GPS stamped and GIS compatible.
      Moving off the ground, weÕre trying many different sorts of remote sensing to see which are capable of showing what we need to know. Before each mission, we place aerial targets on the ground, usually made of 1 x 5-meter long strips of white paper weighted down with rocks. As with everything else, the targets are GPS fixed to plus or minus 3-meters. Using a helicopter we were able to place dozens of targets on the ground in the four days before a major airborne video over-flight.
      Our interest is in correlating imagery from diverse sensors and from diverse altitudes. Satellite data, such as Landsat Thematic Mapper, has a spatial resolution of 30-meters per pixel. But deserts have less than 20 percent vegetative cover, and many places in Joshua Tree have barely 10 percent cover. There simply isnÕt enough chlorophyll on the ground to discern much. So the project has gradually moved closer to the ground.
      In order to measure spectral reflectances of desert plants in the field, we used a balloon owned by Marine Ecological Consultants of Carlsbad, Calif. This hand-tethered craft carries a 35mm camera, camcorder, and radiometer 50-meters above the ground. The observer on the ground looks through the eyepiece remotely attached to the camcorder and can fire the radiometer when everything is lined up around a point of interest. Once again, GPS-stamped targets were photographed, usually at the corners of each frame shot by the balloon. The result is a high-resolution radiometric reading showing the spectral reflectance of individual bushes which we can visit on the ground to correlate that signature with plant species, size and health.
      The balloon photos are very detailed, but the area imaged is only about 20-meters on a side. Obviously, this view is too detailed to cover any large area. The goal is, rather, to closely study the multi-spectral signatures. We will then use this knowledge to explain what we are seeing in images taken from higher up. This is precisely the approach being taken by our collaborator, Dr. Jae Lee, a remote sensing expert from Argonne National Laboratory in Argonne, Ill. Working with Dr. Lee, we are calibrating airborne video data with radiometric readings both from the balloon and lab studies in hopes of developing a predictive model for tortoise habitat.
      The next step off the ground is multi-spectral airborne video. We have used the ADAR system operated by Positive Systems of Whitefish, Mont. The equipment produces 4-band digital images with spatial resolution of 0.3 to 1.0 meter per pixel. The color bands and band-widths may be specified, and the aircraft stamps a GPS location on the center of each frame.
      The ADAR system is much like the SPOT satelliteÕs instrument package, but flown from a small plane. Tiny differences in micro-topography and soil influence appear clearly in the ADAR data. ADAR imagery will be great for checking if plant spacing or species diversity are correlated with tortoise density. To date, we have had difficulty distinguishing plant species from ADAR data, but it still may be possible if pictures are shot in the right time window (when the plants are green) and if multi-variate statistical techniques are employed.
      Higher in the air, the next imagery we use is from airborne visual light video and 35mm photography. The rolling video gives a feeling for the topography which can be lacking in traditional still photography. When the entire park was shot at 1.8-meter per pixel resolution, we had zoomed close-ups taken every six seconds, giving a point subsample of high resolution in which resolution is about 20-30 cm. The still photography was GPS stamped and sent to us on CD ROM using Kodak Photo-CD technology.
      Reaching the highest levels of flight for conventional aircraft, our next data layer comes from the Airborne Visual Infrared Imaging Spectrometer (AVIRIS) sensor, flown on a U-2 type aircraft. Although the high-altitude (about 24,000 meters) does not allow for great spatial resolution, AVIRIS is known for exquisite spectral resolution, breaking the infrared spectrum into 224 separate channels. This project, done in cooperation with the U.S. Geological Survey will yield our best information on soil chemistry and geological correlates with our areas of known tortoise clusters.
      Lastly, folks in the military wondered if spy satellite data (the defense people call them National Technical Means or NTMs) when combined with off-the-shelf SPOT satellite data would be useful for vegetation mapping or tortoise habitat evaluation. Here again, GPS matched layers would be rectified and matched with other coverages. We still donÕt know what resolution the spy satellite data can offer, as itÕs currently classified, and we donÕt know exactly how the combination with SPOT data would occur, but these technologies are theoretically excellent for resolving these difficult questions on a landscape scale.
      It is still too early to say whether these high-tech means will be able to locate tortoises for us with statistical accuracy. It is also unclear which, if any, of these technologies will prove most useful. If a predictor of tortoises can be found, our teams of staff biologists and volunteers will know the best places to target. Even knowing this, it will take years to survey Joshua Tree National Park. If the effort to develop a habitat model fails, we still will have learned much about soils, plants and topography. Because the tortoise is widespread and ecologically important, its protection serves as an umbrella for other species with unfamiliar names and whose biology may be completely unknown, so the ecosystem as a whole and we, its managers, will benefit.
      As we go about our daily business in Resources Management, we find it hard to imagine how we ever did things before there was global positioning. Starting with our first handheld receiver in 1991, these instruments have gradually become invaluable to nearly every aspect of our work. We can no more do without them than we could without pocket calculators. Whether mapping fire lines, surveying borders, locating historic structures, or tracking Bighorn Sheep, we feel this technology is the single most valuable new technology since the personal computer.

Note: Reference to product names and companies does not represent endorsement by the National Park Service or the U.S. government.

About the Author:
Jerry Freilich, Ph.D. is an ecologist working for Joshua Tree National Park. He may be reached at 619-367-4528 (phone), or 619-367-6392 (fax).

Back