Guided transect sampling with a new strategy for second-stage guidance

Abstract

Guided transect sampling (GTS) was recently proposed as a method for surveying sparse and geographically scattered populations if prior information with high spatial resolution is available. With this method, wide survey strips are, during a first stage, randomly selected in the area to be surveyed. In the second stage, the selected first-stage strips are subsampled with a survey line or survey strip based on the prior information. In this article, a new strategy for second-stage guidance is developed and evaluated in a simulation study. The suggested strategy uses an assumed model for the relationship between the prior information and the variable of interest to find an efficient design for selecting the second-stage transects. The performance of the suggested strategy is evaluated in terms of the variance obtained in different realizations of the assumed model. Results from the simulation study indicate that the suggested strategy is an efficient alternative to previously suggested strategies, if the correspondence between the model population and the real population is good. On the other hand, if the difference between the two populations is large, large variances were sometimes obtained.