We used AI and satellite imagery to map ocean activities that happen out of sight, including fishing, shipping and energy development

Humans are racing to harness the ocean’s vast potential to power global economic growth. Worldwide, ocean-based industries comparable to fishing, shipping and energy production generate at the very least US$1.5 trillion in economic activity every year and support 31 million jobs. This value has been increasing exponentially over the past 50 years and is anticipated to double by 2030.

Transparency in monitoring this “blue acceleration” is crucial to forestall environmental degradation, overexploitation of fisheries and marine resources, and lawless behavior comparable to illegal fishing and human trafficking. Open information also will make countries higher in a position to manage vital ocean resources effectively. But the sheer size of the ocean has made tracking industrial activities at a broad scale impractical – until now.

A newly published study within the journal Nature combines satellite images, vessel GPS data and artificial intelligence to reveal human industrial activities across the ocean over a five-year period. Researchers at Global Fishing Watch, a nonprofit organization dedicated to advancing ocean governance through increased transparency of human activity at sea, led this study, in collaboration with me and our colleagues at Duke University, University of California, Santa Barbara and SkyTruth.

We found that a remarkable amount of activity occurs outside of public monitoring systems. Our recent map and data provide probably the most comprehensive public picture available of commercial uses of the ocean.

Operating at nighttime

Our research builds on existing technology to offer a far more complete picture than has been available until now.

For example, many vessels carry a tool called an automatic identification system, or AIS, that robotically broadcasts the vessel’s identity, position, course and speed. These devices communicate with other AIS devices nearby to enhance situational awareness and reduce the possibilities of vessel collisions at sea. They also transmit to shore-based transponders and satellites, which could be used to monitor vessel traffic and fishing activity.

However, AIS systems have blind spots. Not all vessels are required to make use of them, certain regions have poor AIS reception, and vessels engaged in illegal activities may disable AIS devices or tamper with location broadcasts. To avoid these problems, some governments require fishing vessels to make use of proprietary vessel monitoring systems, however the associated vessel location data is generally confidential.

Some offshore structures, comparable to oil platforms and wind turbines, also use AIS to guide service vessels, monitor nearby vessel traffic and improve navigational safety. However, location data for offshore structures are sometimes incomplete, outdated or kept confidential for bureaucratic or industrial reasons.

Shining a lightweight on activity at sea

We filled these gaps through the use of artificial intelligence models to discover fishing vessels, nonfishing vessels and stuck infrastructure in 2 million gigabytes of satellite-based radar images and optical images taken across the ocean between 2017 and 2021. We also matched these results to 53 billion AIS vessel position reports to find out which vessels were publicly trackable on the time of the image.

Remarkably, we found that about 75% of the fishing vessels we detected were missing from public AIS monitoring systems, with much of that activity going down around Africa and South Asia. These previously invisible vessels radically modified our knowledge concerning the scale, scope and placement of fishing activity.

For example, public AIS data wrongly suggests that Asia and Europe have comparable amounts of fishing inside their borders. Our mapping reveals that Asia dominates: For every 10 fishing vessels we found on the water, seven were in Asia while just one was in Europe. Similarly, AIS data shows about 10 times more fishing on the European side of the Mediterranean compared with the African side – but our map shows that fishing activity is roughly equal across the 2 areas.

For other vessels, that are mostly transport- and energy-related, about 25% were missing from public AIS monitoring systems. Many missing vessels were in locations with poor AIS reception, so it is feasible that they broadcast their locations but satellites didn’t pick up the transmission.

We also identified about 28,000 offshore structures – mostly oil platforms and wind turbines, but in addition piers, bridges, power lines, aquaculture farms and other human-made structures. Offshore oil infrastructure grew modestly over the five-year period, while the variety of wind turbines greater than doubled globally, with development mostly confined to northern Europe and China. We estimate that the variety of wind turbines within the ocean likely surpassed the variety of oil structures by the top of 2020.

Supporting real-world efforts

This data is freely available through the Global Fishing Watch data portal and might be maintained, updated and expanded over time there. We anticipate several areas where the knowledge might be most useful for on-the-ground monitoring:

– Fishing in data-poor regions: Shipboard monitoring systems are too expensive to deploy widely in lots of places. Fishery managers in developing countries can use our data to observe pressure on local stocks.

– Illegal, unreported and unregulated fishing: Industrial fishing vessels sometimes operate in places where they mustn’t be, comparable to small-scale and traditional fishing grounds and marine protected areas. Our data may help enforcement agencies discover illegal activities and goal patrol efforts.

– Sanction-busting trade: Our data can make clear maritime activities which will breach international economic sanctions. For example, United Nations sanctions prohibit North Korea from exporting seafood products or selling its fishing rights to other countries. Previous work found greater than 900 undisclosed fishing vessels of Chinese origin within the eastern waters of North Korea, in violation of U.N. sanctions.

We found that the western waters of North Korea had way more undisclosed fishing, likely also of foreign origin. This previously unmapped activity peaked every year in May, when China bans fishing in its own waters, and abruptly fell in 2020 when North Korea closed its borders due to COVID-19 pandemic.

– Climate change mitigation and adaptation: Our data may help quantify the dimensions of greenhouse gas emissions from vessel traffic and offshore energy development. This information is significant for enforcing climate change mitigation programs, comparable to the European Union’s emissions trading scheme.

– Offshore energy impacts: Our map shows not only where offshore energy development is going on but in addition how vessel traffic interacts with wind turbines and oil and gas platforms. This information can make clear the environmental footprint of constructing, maintaining and using these structures. It can even help to pinpoint sources of oil spills and other marine pollution.

Healthy oceans underpin human well-being in a myriad of how. We expect that this research will support evidence-based decision-making and help to make ocean management more fair, effective and sustainable.