Datasets and code
5 years of data (Jan ’12 – December ’17)
- 694 vessels capable of transshipping fish at sea.
- 46,570 Loitering Events: Transshipment-capable vessels going slow enough and long enough to transship.
- 10,233 Potential Transshipments: Potential encounters in which a fishing vessel was also recorded traveling slow enough, close enough, and long enough to transship.
To obtain our initial data on transshipment, as well as our transshipment report describing the methodology from the dataset, visit this link.
The first global footprint of vessel rendezvous behavior
In 2017, we released an original report that revealed remarkable new insights about what goes on between fishing vessels at sea. In our report, The Global View of Transshipment, our analyses created the first global footprint of rendezvous behavior within the ocean fishing industry. It is the first comprehensive picture of potential transshipment ever published, and it lays the groundwork for significant reductions in Illegal, Unreported, and Unregulated catch entering the global seafood market.
In July 2018, we expanded on our report with two publications. “Identifying Global Patterns of Transshipment Behavior,” a collaboration with our partners at SkyTruth and Google, was published in Frontiers of Marine Science and was the first-ever global assessment of transshipment in a scientific journal (read our blog post from analyst Nate Miller to learn more about the results). “Global hot spots of transshipment of fish catch at sea,” a study written by our partners at SkyTruth and Dalhousie University, was published in Science Advances and built upon the former study to identify for the first time the fishing vessel types and fisheries most involved in transshipment, as well as what proportion of high seas catch is transshipped versus landed directly.
Vessels meet at sea for multiple reasons, such as refueling and exchanging crew, but in the commercial fishing industry, they also meet to transfer catch, or transship. Large vessels with refrigerated holds collect catch from multiple fishing boats and carry it back to port. By enabling fishing vessels to remain on the fishing grounds, transshipment reduces fuel costs and ensures catch is delivered to port more quickly. It also leaves the door open for mixing illegal catch with legitimate catch, drug smuggling, forced labor and human rights abuses aboard fishing vessels that remain at sea for months or years at a time. As a pathway for illegal catch to enter the global market, transshipment prevents an accurate measurement of the amount of marine life being taken from the sea. It obscures the seafood supply chain from hook to port and hobbles efforts to manage fisheries sustainably.
For these reasons, transshipment is illegal in many cases, but it has been largely invisible and nearly impossible to manage, because it often occurs far from shore and out of sight. Until now. Identifying when and where transshipment happens can play a significant role in reducing illegal activity at sea.
Patterns revealed by a global view
Our report emphasized that transshipment is a significant pan-national problem involving ships registered to a diverse array of countries operating on the high seas and in offshore waters far from their home ports.
We identified important patterns in the data noting that:
- Transshipment in offshore coastal waters is more common in regions with a high proportion of Illegal, Unreported, and Unregulated (IUU) fishing than in regions where management is strong such as in North America and Europe.
- Clusters of transshipment activity occur along Exclusive Economic Zone (EEZ) borders of some countries, and inside those zones of nations with high corruption ratings and with limited monitoring capabilities.
These correlations do not provide any proof of specific illegal behavior but they raise important questions and can lead to more informed international efforts by fisheries management organizations to prevent or better regulate transshipment.
Applying machine learning to detect transshipment
With two hundred thousand ships on the ocean, spotting a refrigerated cargo vessel meeting up with a fishing vessel could take an analyst months on end. The analytical tools developed for our report do it automatically in days or hours. Using machine learning, we processed more than 32 billion Automatic Identification System (AIS) signals broadcast from ships at sea to identify vessels capable of receiving transshipments and analyze their movements. Those vessels include fish tenders, live fish carriers, factory/processors and refrigerated cargo vessels. Verifying our results with confirmed fishery registries and open source online resources, we identified 694 vessels capable of transshipping fish at sea between 2012 and December 2017, and 46,570 incidents in which one of these vessels exhibited telltale transshipment behavior patterns such as drifting slowly enough and long enough to receive a transfer of catch.
The vast majority of those occurrences were not accompanied by signals from fishing vessels, and their activity cannot be verified, but, given that many fishing vessels turn off their AIS device when they do not want to be detected, and some fishing vessels do not have AIS, these events must be considered “loitering events.” Drilling down through the 46,570 occurrences, we identified 10,233 instances in which a fishing vessel was also confirmed to exhibit this behavior within close proximity of the transshipment vessel. We have labeled these “potential transshipments.” Our algorithm was verified by matching a subset of “potential transshipments” to known transshipments recorded by fishing registries.
Release of this report and the related datasets represent our first steps toward revealing vessel behaviors that are often hidden at sea. In June 2018, we released the first-ever ‘live’ global view of likely transshipping at sea. The visualisation is a major step towards greater transparency in transshipping.
With these new analytical tools, developed using AIS data, fisheries managers will be able to identify and monitor transshipment anywhere in the world. We are also expanding our view of fishing vessels and rendezvous activity as we incorporate additional data sources, including government Vessel Monitoring System (VMS) data, and continue to mine our data for a deeper understanding of vessel rendezvous patterns and potential transshipment activity.
Our research partners and NGOs are also working with our transshipment data and analyses to make important discoveries that can reduce the illegal activity associated with transshipment and influence policy-making.
The report and associated datasets are free for download. Note: the original report published in February 2017 was revised in August 2017 to reflect an update to terminology and the database.
This work was supported by a grant to SkyTruth from the Walton Family Foundation and made possible by Google through the in-kind use of Google’s cloud computing platforms and technical and project guidance.