Surveying coastal ship traffic with LANDSAT

Abstract

A semi-automated algorithm was developed to detect ships in LANDSAT 7 images. The algorithm combines multispectral and pattern recognition methods to discriminate ships from ocean clutter. Automated processing enables us to process a large number of images and gather a statistical picture of ship traffic patterns. As a test case we applied the algorithm on 54 LANDSAT images in the area of Jacksonville, FL, from the period 1999–2003. The area and time period are the same as an earlier ship traffic study by Ward-Geiger et al. using ship reports in the Mandatory Ship Reporting System (MSRS). The similarities between the two studies suggest that LANDSAT is a good alternative for surveying nearshore ship traffic.