18 



FISH AND WILDLIFE TECHNICAL REPORT 30 



-8 -6 -4 -2 



2468 -8-6-4-2 

 Longitudinal Error (km) 



6 8 



Fig. 12. Scatterplot of locations for two PTT's, each deployed on a captive caribou then placed on an adjacent fencepost. The caribou 

 was confined to an area of about 50 X 50 m. 



ation in error magnitude could be explained by maximum 

 satellite elevation. Similar significant (P < 0.01) regres- 

 sions were found for elk in Yellowstone National Park and 

 for Dall sheep in Alaska. Maximum satellite elevation 

 usually explained errors in the longitudinal direction bet- 

 ter than in the latitudinal direction. For example, over 73% 

 of longitudinal error for 20 Dall sheep locations was ex- 

 plained by satellite elevation (Fig. 17). 



For mobile animals, the correct elevation cannot be 

 known. Although a mean elevation during all or part of a 

 year may be specified, this only reduces but does not 

 eliminate the discrepancy between its true elevation and 

 Argos's calculations. For animals that inhabit moun- 

 tainous topography and move readily among sites with 

 varying elevations, discrepancies between true and as- 

 sumed elevation will still occur. Thus, the issue of error 

 arising from PTT elevation will still be relevant, even if an 

 accurate mean elevation is specified. 



Biases arising from PTT elevation can be corrected by 

 providing Argos with the correct elevation. Clark (1990) 



recently confirmed that problems arise not from elevation 

 itself, but rather from Argos making calculations using sea 

 level when the PTT is at a different elevation. He placed 

 CML 86 PTT's (Courrouyan 1987) at known locations in 

 Glacier and Yellowstone national parks and examined the 

 locations calculated by Argos using the correct elevations. 

 Standard deviations were not significantly different from 

 those attained during sea-level tests of similar PTT's. 



Overall Error. Users of satellite telemetry will rarely 

 perform extensive tests on all their PTT's under each of the 

 various situations that occur in the wild. Rather, investiga- 

 tors may wish to simply assume an expected magnitude of 

 error and make their inferences accordingly. Here, we 

 summarize our available data on overall error magnitudes. 



Because PTT's on animals had greater error than those 

 placed on buildings and fenceposts, we compiled a data set 

 consisting of errors from known locations when the PTT 

 was worn by an animal. All known locations were at 

 elevations of <500 m. The resulting data set consisted of 

 estimates from seven PTT's, of which 218 were from 



