by Mark MacAllister
June 21, 2002
Page 2 : About satellite data
Once a collar is attached to an elephant, it immediately begins to transmit important data to satellites orbiting overhead. That data is sent from the satellites to a facility in France, where it is in turn emailed to a researcher in Chatham County, North Carolina, who also retransmits the data to the NC Zoo, as well as to several locations back in Cameroon.
Researchers in Asheboro review the data daily. These data help us understand an animal's migratory habits over time. We can also determine the homerange of an entire herd based on one animal's movements over several months (though at least twelve months of data are needed to determine a homerange accurately). Most importantly, we can begin to estimate those times when a herd may come into contact with human populations--an important first step in reducing harmful interactions between elephants and humans.
Accuracy issues
Each satellite transmission carries a variety of data about a collared elephant. A typical transmission report looks like the "Transmission Report" graphic in the Media Gallery at right.
Of particular interest are the date and time of the transmission, as well as the animal's latitude and longitude (underlined in red). What we know from the report above is (1) the data pertains to Collar #14120, which at that time was deployed on Desiré in southeast Cameroon, (2) the location data was taken on August 21, 2001 at 12:34:39 GMT, and (3) the elephant's location is believed to be 2.783 degrees north latitude and15.933 degrees east longitude.
We say believed to be because the accuracy of the latitude and longitude fields is not certain. Despite the fact that they are orbiting thousands of miles above the earth, the satellites we use are capable of determining latitude and longitude locations to within a few meters of accuracy. On the other hand, factors like cloud cover, atmospheric conditions, vegetation, and battery strength can make a location determination far less accurate, sometimes throwing it off by a kilometer or more. Therefore, of critical importance is the "location class" (LC) figure, which describes the relative accuracy of the data transmitted in that block (see red arrow). A table of LC figures and associated accuracy is shown below:
| LC Figure | Estimated accuracy |
| 3 | Less than 150 meters error |
| 2 | 150 to 350 meters error |
| 1 | 350 to 1000 meters error |
| 0 | More than 1000 meters error |
| A | No error estimate; (3 satellite messages received) |
| B | No error estimate; (2 satellite messages received) |
| Z | Invalid location |
In the example transmission report, the LC reading of "3" tells us that these particular location data are quite dependable, and that the elephant is within 150 meters (plus or minus) of the latitude and longitude noted in the report. On the other hand, had the LC figure been "0" or "A" or "B", the stated location would not have been as dependable; it may have ultimately been accurate, but we would have no way of knowing that until the points are further analyzed.
Given these variances, we must account for different levels of accuracy and data dependability as our analysis continues. When determining a herd's homerange, for example, we must be careful to build homerange maps using the most accurate measurements available. If less accurate measurements are included in the analysis, the homerange will likely appear much different than it truly is.
Next Page : Basic data analysis
Pages: 1, 2, 3, 4, 5, 6, 7, 8
View printer-friendly version