Part 2 - Evaluation of imagers and visual inspections
Vegetation was a significant factor in the detection of rabbit warrens
(p=<0.001). There were 22 warrens present within the survey
area. All warrens identified by visual assessment (n=14) were identified
in the imagery from the Vayu. Three warrens identified by visual
assessment were not identified in the Zenmuse footage. However, both the
Zenmuse and the Vayu detected more rabbit warrens and entrances than
visual inspection (Table 3, Fig. 2a). A pairs plot shows the correlation
between counts under each imager over all vegetation classes (1:1 lines
added to show agreement, Fig. 3). The methods seem broadly in agreement.
The Zenmuse detected significantly more entrances than both visual
inspection and the Vayu (P=0.049). Several entrances and warrens
detected using the Zenmuse were later visually identified as false
positives (n=21 entrances, n=19 warrens) (Fig. 2b). The Vayu detected
active warrens beneath vegetation and detected more warrens than visual
inspection (Fig. 2d, Table 3) with no false positives. Only 10 of the 22
warrens were detected by all three methods (Vayu, visual and Zenmuse),
and 13 of the 22 were detected by both the Vayu and visual inspection.
Two warrens were detected by thermal imager (both types) that were not
detected by visual inspection with an additional four warrens (three
single and one four-entrance warren) only detected by the Vayu (see
Appendix).