Peter Leimgruber, Ph.D., head of CRC's Conservation GIS Laboratory, is leading a team of researchers, graduate students, interns, and protected area wardens and rangers in studying the ecology of Asian elephants and their habitat throughout Asia and in two protected areas of Myanmar (Burma), specifically the Alaungdaw Kathapa National Park and Htamanthi Wildlife Sanctuary.

Myanmar, about the size of the state of Texas (approximately 678,500 square kilometers [261,970 square miles]), is located in Southeast Asia, bordering the Andaman Sea and the Bay of Bengal, between Bangladesh and Thailand. Its diverse environments range from alpine meadows in the Himalayan Mountains to mangrove swamps near the Bay of Bengal. The Alaungdaw Kathapa National Park lies at the edge of Myanmar's hill region, close to the country's center, and is very steep and rugged. The Htamanthi Wildlife Sanctuary borders Kachin State in northern Myanmar, and traveling there requires three days of riverboat travel on Myanmar's famous Chindwin River. Its habitats are dominated by lowland rain forest and softly undulating hills with little topography.

Leimgruber and his team are using Esri's ArcGIS software and ERDAS IMAGINE from Esri Business Partner Leica Geosystems GIS & Mapping to identify existing habitat, monitor habitat loss, and assess the effectiveness of Myanmar's protected area system with the intent of preventing local extinctions of Asian elephants and applying these methodologies to other species in the future. Esri's software has been used at the National Zoo in Washington, D.C., for the last 12 years. ERDAS IMAGINE was chosen for image processing because of its easy interoperability with ArcGIS software.

CRC's Conservation GIS Laboratory is equipped with 16 Dell and Gateway desktop PC workstations. Several projects have been performed, including a regional assessment using GIS to determine how much habitat is left for elephants in all of Asia, a national forest cover map and mapping of two protected areas in Myanmar, and intense ground surveys of elephants (population locations and number of individuals), combined with satellite tracking (see photos and the exclusive online-only sidebar "Ground Surveys and Satellite Tracking of Asian Elephants") of one elephant herd, to monitor and identify their habitat selection in the protected areas.

According to Leimgruber, "The regional conservation assessment was performed to answer two questions essential for the preservation of wild Asian elephants: (1) Where are the largest untouched sections of wildlands that constitute suitable elephant habitat? and (2) How much of these remaining untouched elephant wildlands are within the Southeast Asian protected area systems?"

The staff at CRC used ArcView and ArcInfo to combine existing data layers (land cover, protected areas, population density estimates, roads, and fire boundaries) at a regional scale in Asia to delineate remaining wildlands, defined as large unfragmented natural habitats that support Asian elephants.

The results of the study were sobering. Only about 51 percent of the geographic range of the Asian elephant can be considered wildland and elephant habitat. Agriculture, irrigation, fires, and roads affect all other areas within the range. Of these remaining habitats, only about 16 percent are under some kind of legal protection recognized by The World Conservation Union.

Myanmar is the country with the largest number of remaining wildlands in elephant ranges. The total amount of this wildland encompasses more area than all wildlands in the combined elephant ranges of Thailand and India, the two countries ranking second and third. The map displaying the results of the assessment is served on the Web using ArcIMS and can be viewed at www.natzoo.si.edu/animals/asianelephants.

One of the biggest challenges for mapping Southeast Asian forests is timing. Images acquired during the rainy season tend to have too much cloud cover to be useful. Images collected during the dry season are difficult to classify because seasonal changes in mixed deciduous forests, whose trees drop their leaves during the dry season, are easily confused with deforestation. The best time period for image acquisition is between rainy and dry season, as close to the end of the rainy season as cloud cover allows. In Myanmar this time usually falls between the beginning of December and the end of January.

According to Leimgruber, "The forest cover was created using techniques developed by Mark Steininger at Conservation International's Center for Applied Biodiversity Science Program (CABS)." Two Landsat satellite images from two different dates were georeferenced and stacked in ERDAS IMAGINE. The spectral characteristics and differences between the two dates were used to create a classified data set. CRC performed this supervised classification by using the signature editor in ERDAS IMAGINE, which measures spectral differences and similarities, to define the following classes: forest cover, not forest cover, deforestation, reforestation, water, and water change. This data is served using ArcIMS and can be viewed at www.natzoo.si.edu/animals/asianelephants.

Performing image analysis on an area as large and rugged as Myanmar makes assessing the accuracy of the final map extremely challenging. It is nearly impossible to obtain enough ground control points to be used for accuracy assessments of the maps created from each Landsat TM image. The CRC team decided to use at least one ASTER image for every Landsat image tile to determine the accuracy of forest cover data. "If the accuracy assessment, which involved comparing the forest cover data with ASTER images, was below 70 percent," notes Leimgruber, "then we attempted to improve the spectral signatures used for the classification and reclassified the image."

Once the classification process was completed, the classified images were mosaicked together using ArcGIS Desktop to create a countrywide map. The map was smoothed in ArcGIS to eliminate noise inherent in creating such a large data set. Next, CRC performed spatial clustering and regional group analysis to create a polygon coverage of the forest cover classifications in ArcGIS. These polygons allow scientists to see how much forestland was logged or converted to other land use, where most of the forest loss occurred, and the average size of a deforestation patch.

According to Leimgruber, "The good news is there is more untouched forest habitat in Myanmar than most places in Southeast Asia." Other than CRC, no one else has performed surveys of these regions because there is very little infrastructure.

For more information on this project, contact Peter Leimgruber, head of CRC's Conservation GIS Laboratory (e-mail: leimgruberp@crc.si.edu), or visit the Smithsonian's National Zoological Park Conservation and Research Center at www.natzoo.si.edu/ConservationAndScience/ConservationGIS.

For more information on spatial solutions, contact Leica Geosystems GIS & Mapping (toll free in the United States: 1-877-463-7327, outside the United States: 1-404-248-9000), or visit www.gis.leica-geosystems.com.