Mapping the Abundance and Distribution of Adélie Penguins Using Landsat-7: First Steps towards an Integrated Multi-Sensor Pipeline for Tracking Populations at the Continental Scale

The last several years have seen an increased interest in the use of remote sensing to identify the location of penguin colonies in Antarctica, and the estimation of the abundance of breeding pairs contained therein. High-resolution (sub-meter) commercial satellite imagery (e.g., Worldview-1, Quickbird) is capable of colony detection and abundance estimation for both large and small colonies, and has already been used in a continental-scale survey of Adélie penguins. Medium-resolution Landsat imagery has been used successfully to detect the presence of breeding penguins, but has not been used previously for abundance estimation nor evaluated in terms of its minimum colony size detection threshold. We report on the first comprehensive analysis of the performance of these two methods for both detection and abundance estimation, identify the sensor-specific failure modes that can lead to both false positives and false negatives, and compare the abundance estimates of each method over multiple spatial scales. We find that errors of omission using Landsat imagery are low for colonies larger than,10,000 breeding pairs. Both high-resolution and Landsat imagery can be used to obtain unbiased estimates of abundance, and while Landsat-derived abundance estimates have high variance for individual breeding colonies relative to estimates derived from high-resolution imagery, this difference declines as the spatial domain of interest is increased. At the continental scale, abundance estimates using the two methods are roughly equivalent. Our comparison of these two methods represents a bridge between the more developed high-resolution imagery, which can b