Synthetic aperture radar for maritime domain awareness: Ship detection in a South African context

Abstract

Maritime Domain Awareness is an initiative started to help each sea-bordering country improve its understanding of its Exclusive Economic Zone. A country that improves its Maritime Domain Awareness ensures that activities such as piracy and Illegal, Unreported and Unregulated fishing are identified more quickly and the appropriate actions for each are taken. For instance, having better awareness of the ships entering and leaving a country’s Exclusive Economic Zone could prevent illegal fishing that, in some cases, accounts for up to a 40% loss of legal catches. Various sources of data can be used to keep track of ships at sea including ship transponders (such as the Automatic Identification System and Long Range Identification and Tracking systems) and active radar sensor systems such as Synthetic Aperture Radar satellites. With the advent of new, freely available Synthetic Aperture Radar imagery from Sentinel-1, the observation of large coastal areas is becoming more and more feasible for countries such as South Africa. Synthetic Aperture Radar satellites are able to monitor large tracts of the Earth, day or night, in any weather condition. This allows for the tracking of hundreds of square kilometres of sea area in a single image. By combining and processing imagery from these satellites and ship transponders, a better picture of a country’s maritime domain can be captured thereby allowing countries to respond to environmental, commercial or security threats at sea. Furthermore, if South Africa can implement its own SAR asset, Maritime Domain Awareness for the entire country would be improved tremendously. This paper aims to give an overview of the research being done to detect ships in Synthetic Aperture Radar imagery within a South African context. It presents the current and future of Synthetic Aperture Radar satellites, how methods such as the Constant False Alarm Rate and Wavelet Transform are used to detect ships at sea and how techniques such as Simulated Annealing and Cascade Classifiers are currently being used to further improve ship detection accuracies.