Aerial survey sampling of contour transects using double-count and covariate data

By combining double-count and line-transect sampling data we obtained estimates of animal and group densities. The sample design consists of flying aerial surveys along elevational contours with two observers in the same platform searching the same region on one side of the airplane. Detected distances are not grouped, and factors such as size and activity of the group, and the extent of the horizon, that may have biased the detection ability of the observers are accounted for by use of covariates. We have adapted line transect methodology to contour transects because the mountainous terrain makes it impossible to fly straight transects. Separate detection functions of each observer are modeled by logistic curves. The theory is illustrated with data from a 1996 survey of brown bears in Kodiak Island, Alaska.