Radiotelemetry error: location error method compared with error polygons and confidence ellipses

Publisher Website
Google Scholar
Abstract
We assert that researchers should use statistics derived from the linear distances between actual and estimated locations of test transmitters to estimate location error in radiotelemetry data. We call this approach the location error method. We used the distribution of such linear distances from a test data set from a study on black bears (Ursus americanus) in the mountains of North Carolina to predict error statistics for another test data set. We then compared the predicted with the actual error statistics. We also predicted error statistics for the second test data set using the error polygon method and Lenth's maximum likelihood estimator method. Linear and areal predictions of error using the location error method closely matched actual error in the second data set. The 90% confidence area calculated from test data contained 90% of the actual locations. The linear error measures taken from the error polygon method averaged twice the length of the 90% confidence distances generated from test data, and the 90% error polygons actually contained 95% of the true locations. The 95% confidence ellipse of Lenth's maximum likelihood estimator method [Formula: see text] was also a poor indicator of the precision of the actual location errors (95th percentile = 187 ha). Investigators should use bearing analysis on known locations only to find and correct biases in the bearing data prior to triangulation.

This publication has 16 references indexed in Scilit: