Inroads into Ilegal transhipments analysis / by Francisco Blaha

I recently reported on an interesting report on the Global Footprint of Transshipments produced by Global Fishing Watch. As is had a flight with shitty movies, I went over the methodology of the report, and I like it. Is well thought and explained.

Is a pity that these guys don't have access to RFMO and the flag state data on observers, and logsheets, and in many cases the RFMOs don't have the manpower, budget and political will by their DWF Members to get a couple of good consultants/researchers to go over the results and track back the vessels documents, wich are at the end of the day the responsibility of the flag state. (Maybe there is a space for support from big NGOs like Pew or WWF?)

For example, one of the cases they present is of 2 vessels I got to know something about.

In page 6 they present data of an encounter between the ORION S (reefer), and the AMERICAN EAGLE (purse seiner), in 2015 at the high seas (ABNJ) between Ecuador and the Galapagos Islands. Both these vessels are Colombian flagged and belong to the same company that operates a dozen purseiners from the port of Manta in Ecuador. In principle, the purseiners transship in port to the carrier, and when the carrier is full it goes to Panama, crosses the canal and comes down to unload in Cartagena.

All the vessels of this company (including the carrier) are EU listed, hence for that fish to go to the EU a catch certificate is to be validated by the Colombian Fisheries Authority (AUNAP) so in principle they should have access to the logsheets and VMS of the vessels, plus it would surprise is they were to allow transshipments at sea.

In any case, the information should be there to verify the finding, and if is not, then that is a big failure by the Authority of an allegedly responsible flag state. Otherwise, they should audit the company that owns both vessels and look for the “mates receipt” that conform the volumes transferred. No captain would ever let fish go without a receipt because his payment depends on that.

The reality is that there are records for every one of those alleged transshipments, but one need the authority, the knowledge where to look and the willingness by the flag states to do something about it.

Back to the report, I pasted below the methodology of the study so that you can assess for yourself their robustness.


AIS Data
We identified the majority of the world’s reefers and then tracked the movements of these vessels using the Automatic Identification System (AIS), a type of transceiver that broadcasts a vessel’s location every few seconds. Originally meant for vessel collision avoidance, AIS can now be picked up by satellites and terrestrial receivers. This data is aggregated into global into global databases such as the one Global Fishing Watch obtains from the telecommunications company Orbcomm. The International Maritime Organization mandates that all vessels larger than 300 tons on international voyages carry AIS, and most countries have adopted similar or stricter regulations for their EEZs. In 2016, more than 300,000 vessels broadcasted an AIS signal, of which about 80,000 were fishing vessels, and a few hundred were refrigerated cargo vessels.

 Development of Reefer Database
Our database of reefers was compiled from the following sources:

  1. Refrigerated cargo vessels, fish carriers, and fish tender vessels were identified using vessel lists from the International Telecommunications Union and major Regional Fisheries Management Organizations (RFMO).
  2. If a vessel participated in multiple encounters with fishing vessels, we conducted a web search and reviewed RFMO registries using information from the vessel’s AIS to determine if the vessel was a reefer.
  3. Additional reefers were found by investigating documentation on registry websites and other online resources and determining alternate identities that we were able to match in our database.
  4. A vessel classification neural network, developed by Global Fishing Watch to predict vessel types based on movement patterns, was used to identify possible reefers.

Vessels that were identified as likely reefers by this neural network were manually reviewed through web searches and RFMO registries.

After developing the list, we verified vessel information using reputable online sources: the HIS shipping databases, MarineTraffic, ShipSpotting, VesselFinder, and FleetMon. Our database of reefers is now available through

We identified a total of 794 reefers. According to the US Central Intelligence Agency World Factbook, 882 refrigerated cargo vessels were active worldwide in 2010.Assuming that the number of reefers has not significantly increased or decreased since 2010, our dataset includes about 90 percent of the world’s reefer vessels. Some industry analysis suggests the number of reefers is decreasing, meaning that this 90 percent figure is a conservative estimate.1 7 Almost all reefers are required to carry AIS. Ninety-eight percent of the refrigerated cargo vessels in our dataset are larger than 300 gross tons and the International Maritime Organization mandates that vessels heavier than 300 tons on international voyages carry AIS.

Most countries have similar regulations for their EEZs. If we are missing reefers in our dataset, they are likely to be either smaller reefers or vessels that do not make international voyages.

Identifying Transshipments: Encounters and Rendezvous Behavior
We identified potential and likely transshipments in two ways: vessel encounters and rendezvous behaviour by reefers. We extracted these signals using our AIS and reefer databases with help from locations of known transshipments from the Indian Ocean Tuna Commission (IOTC).

Vessel Encounters
To identify likely transhipment events, we identified all interactions between two vessels which remained within 500 meters of each other for longer than 3 hours while travelling at less than 2 knots. These parameters balance the need to detect vessel pairs nearby for extended periods of time while recognising that satellite coverage and inconsistent AIS transmission rates may limit our ability to identify long periods in which vessels are in immediate contact (see data caveats below). We filtered our results to include only events where one of the vessels was a refrigerated cargo vessel and the other a fishing vessel. This left us with 5,065 encounters between reefers and fishing vessels, or “likely transshipments,” from 2012 through 2016.

Rendezvous Behaviour by Reefers
Refrigerated cargo vessels exhibit specific rendezvous behaviours during transshipments. We identified these behaviours by analysing known, observer-reported transshipments from the Indian Ocean Tuna Commission (IOTC; 5,874 transshipments between 2009 and 2015). Through our analysis, we identified reefers that exhibited similar patterns of moving less than 2 knots for longer than 8 hours. Distinctive C-shaped tracks and abrupt shifts in course following a period of slow speeds characterised most transhipment events. Following these metrics, we analysed 117 million reefer positions from 2012 - 2016 and identified 86,490 events where a refrigerated cargo vessel exhibited these behaviours, which we identify as “potential transshipments.”

Not all of these rendezvous events are transshipments of fish. Some may represent transfers of fuel or cargo, and others may be the reefer simply waiting until it is scheduled to travel to its next location. Future research will estimate the fraction of these loitering events that are transshipments of fish. For this report, we present these events as a proxy for transhipment of fish at sea, recognising that it is not a one-to-one relationship.

For this report, we call an event where a reefer encounters a fishing vessel a “likely transhipment” and an event where a reefer exhibits rendezvous behaviour a “potential transhipment.” Our set of “likely transshipments” is a subset of “potential transshipments.” In nearly all cases, we are not able to verify whether the transhipment actually occurs. Any reference to transshipments throughout this report is simply where we see likely or potential transhipment behaviour in our data. Also, we identified several thousand instances of reefers meeting up with non-fishing vessels, or meeting up with other reefers. For this initial report, we exclude these events, and focus only on fishing vessel-reefer encounters.

We also did not investigate transhipment between different fishing vessels.

We restricted our analysis to events occurring at least 20 nautical miles from shore to avoid capturing encounters occurring in ports. This distance is still well within the 200 nautical mile limit of EEZs. Future analysis will consider the distance from port instead of distance from shore so as to capture vessels close to shore but far from the port.

One data challenge is due to the limitations of the satellite receivers used to detect AIS signals. Satellites can fail to receive messages from fishing vessels for two reasons:

  1. High vessel density: A satellite can only record a limited number of messages at once, and when there are too many vessels beneath a satellite, some AIS signals are not recorded. As a result, in areas of high vessel density such as the South China Sea or regions off the coast of Europe, we cannot observe a vessel’s movements as accurately.
  2. Satellite coverage: Based on the number of satellites and their orbital patterns, there can be several hours a day when there is no satellite overhead to receive signals.

Fortunately, these limitations are being addressed by the launching of more satellites. In 2012, only two Orbcomm satellites, the satellite provider for Global Fishing Watch, were operating, and now 18 are in orbit. Also, these limitations do not apply along the coastlines of most developed countries, where terrestrial antennas, which are not as affected by vessel density, are present.

In addition, some vessels will not appear in the dataset for the following reasons:

  1. Vessels may intentionally turn off their AIS transmitters.
  2. Vessels may not have AIS at all. Regulations vary by country, and in international waters, vessels under 300 gross tons are not required to use AIS.
  3. AIS transmitters vary in quality, which results in patchier coverage of vessels with poorer quality hardware.
  4. Some fishing vessels use invalid Marine Mobile Service Identity (MMSI) numbers. For this analysis, we ignored these vessels, as they are difficult to identify. Doing so excluded less than one percent of our total encounters.

We have observed fishing vessels turning off their AIS in some areas of significant transhipment, including near the coast of West Africa, outside the Argentinean EEZ, and in some parts of the Indian Ocean. In future analysis, we hope to quantify this disabling of AIS and determine if it related to transhipment.

While some fishing vessels turn off their AIS from time to time, the practice is significantly more rare among reefers. We analysed all the gaps in transmission from reefers that started and ended more than 10 nautical miles from shore and lasted more than 24 hours, and found that these gaps represented only a small percentage of the total time reefers were active. We estimate that reefers in our dataset only show 24 hours or longer gaps in their track approximately 2 percent of the time while at sea. Therefore, we are confident that the AIS data for refrigerated vessels captures the majority of their footprint.

Transhipment is Most Common in the High Seas and Russian EEZ
About 43 percent of the likely and potential transhipment events happen in the high seas, with the remaining 57 percent within EEZs of different nations. About a third of the total events occur in the EEZ of Russia, where transhipment appears to be a standard part of how their fishing fleet operates. After the high seas and Russia, transhipment is most common in the EEZs of Africa and Oceania.