Most fishing vessels operate hidden from public view

Satellite tracking reveals up to 76% of fishing vessels are hidden from public view, complicating efforts to combat illegal fishing.

Many fishing vessels do not publicly broadcast their location, either legally or illegally. This study used satellite imagery to track them.

Human industrial activity is well documented on land. Almost every road, building, and industry such as forestry and agriculture, are mapped at a high resolution and updated regularly. In the ocean, however, industrial activity is often either not publicly visible or not mapped at all.

To broadcast their location, seagoing vessels usually have either a vessel monitoring system (VMS) or an autonomic identification system (AIS). While VMS is proprietary, which means that it is not made to be publicly shared by default, AIS is supposed to be publicly broadcasted. However, not all vessels are required to use AIS devices, as regulations vary by country, vessel size, and activity. Vessels can also manually turn off their AIS device or manipulate the location they broadcast, especially when they engage in illicit activities. Other hidden spots also exist because of poor satellite reception in some regions, or because AIS data received by terrestrial receptors are restricted by local governments. All these factors create what are called “dark” (i.e. not publicly tracked) vessels, and are the reason human industrial activity at sea is poorly documented.

This is also true of other maritime industrial activities, such as the transport of goods (80% of all traded goods are shipped by sea) or energy production, such as offshore oil extraction or wind turbines. Although we can expect fixed infrastructure data to be more easily available, they are also sometimes restricted for commercial or bureaucratic reasons. Of course, these maritime industries have huge negative impacts on the environment. A third of fish populations are exploited beyond biologically sustainable levels, and almost half of critical marine habitats have been destroyed due to human industries. There is a high need for better monitoring of human use of the oceans and seas.

In this article, authors present a detailed worldwide map of major industrial activities at sea, created from satellite imagery combined with machine learning to identify and classify objects with high accuracy. This method allows them to observe vessels even if they navigate hidden from monitoring systems. By superimposing the ‘real’ satellite data with the publicly available AIS data, they aim to expose how many vessels and infrastructures are ‘hidden’ and where they are located.

The data used in this research ranged from 2017 to 2021. The satellite imagery was done using SAR (Synthetic-aperture radar, unaffected by light levels and most weather conditions, including clouds) from the Copernicus Sentinel-1 mission of the European Space Agency, as well as the Copernicus Sentinel-2 mission for visible and near-infrared imagery. From satellite imagery, they then used a series of machine learning models and algorithms to detect and classify objects, either as fishing vessels, transport vessels, oil or wind infrastructures. AIS data were obtained from satellite providers ORBICOMM and Spire. They superimposed the public GPS coordinates of AIS data with the satellite imagery data to check if detected vessels were publicly declared or not.

The results show that, on average, there are about 63,300 vessels in operation at any given moment, from which about half are fishing vessels. What is particularly concerning is that three quarters of those fishing vessels are not publicly mapped (72-76%). This does not mean that all dark vessels operate illegally; they might be monitored by local agencies that do not make their data publicly available.

The proportional distribution of fishing activity on each continent is also wrongly assumed from public AIS data. Asia shows, by far, the most vessels in activity (67% of all vessels), many of which are absent from public data (88%).

Because the satellite imagery better depicts the global distribution of industrial fishing, it can also reveal potential hotspots of illegal fishing activity. For example, in two well-known marine protected areas — the Galápagos Marine Reserve and the Great Barrier Reef Marine Park — more than 5 and 20 dark vessels can be observed per week, respectively.

There are some limitations in this methodology. One is that the satellites used do not sample most of the open ocean. However, most industrial fishing activity happens relatively close to shore (<60km), and fishing vessels that venture the farthest (>20km) more commonly use public AIS broadcast (60 to 90%, compared to the 25% of total vessels that do). Another limitation is that authors did not classify objects within 1km of shore due to ambiguous coastlines and rocks, but industrial fishing is negligibly present this close to coasts. Finally, satellite SAR imagery still has resolution limitations, currently at 20 meters. This makes the localisation of vessels smaller than 20m harder, even if their algorithms could mostly predict their presence correctly. Unfortunately, these limitations also mean that the number of fishing vessels might be substantially higher than what is presented in this article, as a great number of small artisanal fishing vessels are still not detected.

Satellite photography is not meant to be a real-time monitoring system like VMS or AIS, but rather a complementary method to observe a broader picture of human industrial activity. From the data presented in this article, animal advocates can deduce some potential new strategies and tactics. Asking authorities for better transparency and monitoring of fishing vessels and adjusting the estimates of vessels currently in operation are two such minor advocacy pathways. More broadly, advocates can use this sort of information to focus their work on regions with the most vessels in operation, and focus advocacy on areas prone to illicit fishing.

Original source: https://faunalytics.org