Counting Wildlife from Space; The How, When, Where, and for Which species to use hyper-spatial resolution satellite imagery to count wildlife in wild places.
Scott
Bergen, Living Landscapes Program, Wildlife conservation Society/ Global Conservation Program 2300 Southern Boulevard, Bronx, NY 10460 USA, sbergen@wcs.org
Eric
W
Sanderson, Living Landscapes Program, Wildlife conservation Society/ Global Conservation Program 2300 Southern Boulevard, Bronx, NY 10460 USA, esanderson@wcs.org
(Presenting)
With the advent of commercially available hyper-spatial satellite imagery (IKONOS, Quickbird-2, Orbview3, Worldview-1), we have used this imagery (Quickbird-2, 0.6 m) for the purposes of counting wildlife in wild places. Early results from an experiment conducted at the Bronx Zoo, NY found that an animal’s brightness and size, and surrounding vegetation significant affect a species identification frequency. Estimates based on these factors provided insight to areas with a high probability of counting wildlife in wild places. Since ground observational data was collected at the time of satellite image acquisition, we are able to verify individual animals and their locations within the imagery. Here we review different identification and classification methodologies that we found capable of counting wildlife. Under ideal environmental conditions, with large mammals (>100kg) in low vegetation habitats, satellite imagery can count wildlife accurately, in fact more accurately then ground survey in conditions where large aggregation occur. At Ruaha Nation Park, Tanzania, we developed object oriented change detection techniques to identify Africa savanna-miombo wildlife. At the National Elk Refuge, Jackson, WY, we counted several thousand elk and bison with high accuracy (>90%). At Peninsula Valdes, Argentina (coastal Patagonia), we were able to count southern elephant seals and were capable of distinguishing pups, sub-adult, females, beta and alpha males. The use of hyper-spatial satellite imagery is a needed tool for counting large mammal species occurring in remote places in large numbers (>1,000).
NASA Carbon Cycle & Ecosystems Active Awards Represented by this Poster: