Old Ships and New Tech: Finding Great Lakes Shipwrecks with Machine Learning

Project Description There are an estimated 6000 shipwrecks on the bottoms of the Great Lakes, of which thousands have yet to be located. The traditional methods to locate shipwrecks require many hours of searching the bottom along narrow swaths. The advent of multibeam sonar technology has allowed much more rapid data collection. Within the Great Lakes, large areas of the lake bottoms have been recently surveyed by the National Oceanographic and Atmospheric Administration (NOAA) using multibeam sonar in support of new and planned National Marine Sanctuaries. The amount of data collected in these surveys has overwhelmed the ability of the available experts to analyze the data in real time, so that large portions of the lake bottoms are recorded but not analyzed for shipwrecks.

Applying a machine learning approach to identifying shipwrecks within these data would benefit NOAA, Great Lakes archaeologists, and maritime archaeology broadly. Developing a technique that utilizes the multibeam sonar data, including backscatter data, to identify shipwrecks will create a more complete record of the cultural heritage resting on the lake floors. There are sufficient known shipwrecks in the data to train and test a model that can then be deployed throughout the remaining data to identify previously unidentified shipwrecks. These shipwrecks are claimed by the federal government under the Abandoned Shipwreck Act and the responsibility of state and federal governments to manage. Effective management begins with identifying the shipwrecks. Once identified, the shipwrecks can be analyzed by an archaeologist for their potential to inform our understanding of the past. A model developed for the Great Lakes could be employed more broadly to identify shipwrecks in other regions, greatly improving our ability to manage and interpret humanity’s shared history.

Shipwrecks include information about technological changes; in many cases they were the space shuttles of their time, carrying humans into unknown and inhospitable environments and capturing the attention and talents of the most innovative thinkers of the day. Shipwrecks are also time capsules that can tell a largely complete story about the resources and priorities of people in the past. They are also touchstones for many people, allowing a connection to their ancestors as well as the struggle between humans and larger forces. Shipwrecks are also an excellent means to study chemistry, biology, and geology as they become part of the surrounding environment, and offer opportunities to engage the public in these topics as well as mathematics and other STEM fields. All of this begins with finding the sites.

Expected outcomes would be one or more conference presentations and a co-authored, peer-reviewed publication.
Collaborators will include Dr. Samuel Griegs (Indiana University of Pennsylvania) and Dr. Benjamin Ioset (National Marine Sanctuary Foundation).