Ship Underwater Radiated Noise and Automatic Identification System Track Association

The rise in shipborne underwater radiated noise (URN) has been recognized as a form of pollution that necessitates regular monitoring. Current monitoring procedures require vessel cooperation and costly infrastructure. We identify a knowledge gap regarding the monitoring of URN from vessels of opportunity, where recordings may capture noise from multiple vessels simultaneously. To address this, we propose a framework that uses clustering techniques to differentiate between narrowband tonal components produced by various vessels, as recorded by a single omnidirectional hydrophone. These tonal lines are then associated with nearby ships using data from their automatic identification system (AIS). Our approach involves feature extraction, allowing for the classification of tonal lines into clusters that likely originate from the same vessel. These clusters are then matched with AIS tracks based on their temporal correlation. Our method eliminates the need for manual intervention for vessel screening and tagging. The methodology was tested using data collected from a recorder placed near the approach route to the Haifa port in Israel. The results demonstrate over high accuracy in tonal clustering and above 99% true positive in cluster–AIS associations.