TRACKING WILDLIFE BY SATELLITE 



19 



Table 10. Bias in locations from three fixed platform transmitter terminals (PTT's) before and after correction from 

 NAD27 to WGS84 mapping systems. Sample sizes are shown in Table 8. 



Table 11. Differences (in meters) between the NAD27 and WGS84 projection systems for representative locations. 



Coordinates 



Difference between NAD27 and WGS84 (m) 



Table 12. Correlations between satellite error and topo- 

 graphic features from 44 visual locations of elk in Yel- 

 lowstone National Park fCervus elaphus). None were 

 significant at P < 0.70. 



captive caribou, 80 from captive muskoxen, 7 from cap- 

 tive mule deer, 9 from captive elk, and 17 from captive 

 moose (n = 330). 



The distribution of errors displayed greater variability 



than did the distribution of errors from fixed PTT's. 

 The median error was 543 m, but three errors of >5 km 

 each increased the mean error to 954 m. Ninety per- 

 cent of estimates were < 1,732 m from the true location 

 (Fig. ISA). Location estimates coded as LQ3 had smaller 

 errors: mean error was 656 m; 90% of these were < 1 , 1 58 m 

 from the true location (Fig. 18B; Table 13). In interpreting 

 a single location estimate, biologists can use these distri- 

 butions as a guide to the certainty of the animal's true 

 location. 



These errors were considerably greater than achieved 

 by Argos under the controlled conditions of a fixed PTT, 

 but they were slightly smaller than errors from PTT's 

 deployed on free-ranging animals. After correcting for 

 mean elevation, the overall median and mean errors from 

 44 known locations of elk in Yellowstone National Park 

 were 810 and 1,531 m 334 SE, respectively. Corrected 

 locations from Dall sheep in the Brooks Range yielded a 

 median and mean error of 578 and 802 m 124 SE, 

 respectively (n = 20). Error estimates from these two field 

 data sets may include a small amount of visual mapping 

 error. 



