The fishing industry generates a vast ocean of data. The largest vessels number in the tens of thousands and it is compulsory for these to ping their movements via satellites. Now, analysis tools and AI are being brought to bear on that data, aiming to tackle issues that include illegal fishing, compliance with quotas and the environmental impact fishing has on the natural world.
It’s a challenge that cloud computing and data analytics are perfect for. Just consider the size of the task: according to reports from the Food and Agriculture Organization, there are around 4.6 million fishing vessels operating around the world, logging at least 120 types of data every day. Combined with satellite tracking information, that’s a wealth of information to process.
For starters, big data can play detective. Bigger ships are typically fitted with an Automatic Identification System (AIS) that uses satellite technology for collision avoidance. But the data generated can also detect instances of transhipment – a process of cargo exchange between two ships at sea, which is one of the main ways fish caught illegally gets into the supply chain. You’ve just got to know what to look for.
Global Fishing Watch is an organisation dedicated to ‘advancing ocean sustainability’ by sharing data and near real-time tracking of global commercial fishing activity. For a major report, it analysed 21 billion satellite signals broadcast by thousands of ships between 2012 and 2016, using an artificial intelligence system to detect which were refrigerated cargo vessels.
This information was compared with telemetry data from nearby fishing vessels, moving at similar speeds, and used to identify ships whose movement suggested that it might be meeting with another fishing vessel that wasn’t visible on the system. Over 90,000 potential transhipments were recorded using this system, helping authorities identify illegal fishing practices. 
Global Fishing Watch uses data analysis to produce a wider picture of fishing activity as well. With 70,000 vessels fitted with AIS, algorithms sift through location and movement data provided by satellites. One application of this analysis was to produce a series of ‘heatmaps’, clearly identifying where fishing activity is at its most intense.
Authorities are implementing machine learning to analyse data automatically.
"What's most exciting is what comes next," says David Kroodsma from Global Fishing Watch. "We can now ask questions that we have the data to answer. Where are different species at risk because of bycatch? Because you can now see the overlap between species' ranges and fishing effort. Or, how do subsidies affect fishing? Or, do fishermen respond more to [fuel] prices than to some type of regulation? Or, what parts of the ocean need more protection? We can now have much more informed discussion." 
Fishing quotas are a huge issue in the fishing industry. In addition to the Automatic Identification Systems equipped on larger vessels, data analysis and machine learning technologies are also finding their way onto smaller boats. Electronic monitoring programmes are being developed in New England, for example, which involve fitting automated camera systems onto fishing vessels to assist with reporting. This is a digital upgrade on the standard reporting system, which involves regulatory personnel boarding ships to manually observe if crews are abiding by permitted quotas.
In 2016, a collaborative venture between the National Oceanic and Atmospheric Administration (NOAA), The Nature Conservancy, the Gulf of Maine Research Institute and the Cape Cod Commercial Fishermen’s Alliance, saw boats outfitted with specialised cameras to record information such as fish discards. This is the process whereby certain species of fish, which vessels are not allowed to catch, are thrown back overboard. Authorities are looking at implementing machine learning to analyse the video data automatically, further reducing costs.
“Electronic monitoring is a tremendous tool,” says Brett Alger, national electronics technology coordinator for NOAA Fisheries. “It isn’t necessarily for everyone or every fishery, but we’re working collaboratively, in all of our regions, with fishermen on the ground to understand their needs. I expect it to grow.” Alger points out that a lot of industry groups are “seeing electronic monitoring as a way to be more opportunistic when they go fishing, and to provide more info about what’s occurring in the ocean.” 
In Europe, meanwhile, The SmartfishH2020 project is led by SINTEF Ocean from Norway and includes UK partners such as the University of East Anglia and Marine Scotland.
The objectives of the project are to develop, test, and promote a suite of high-tech systems for the EU fishing sector. These systems will be designed to optimise resource efficiency, improve automatic data collection for fish stock assessment, and provide evidence of compliance with fishery regulations, all while reducing ecological impact.
Today, The SmartfishH2020 project is developing solutions based on a range of technologies, including machine vision, artificial intelligence, and data analysis. It hopes to assist fishermen in making informed decisions during the pre-catch, catch, and post-catch phases of the fishing process to improve economic efficiency and record environmental impact.
The good news is that there’s no shortage of information.
“Vessel monitoring systems, which are a requirement for fishing vessels longer than 12 metres in the EU, and Automatic Identification Systems, which are intended as a navigational aid but are increasingly playing a role in monitoring all sorts of marine activities, generate huge quantities of data,” says Neill Campbell Fishery Analysis and Assessment Team Leader at Marine Scotland. “One position/speed/heading report, per vessel, per two minutes, quickly adds up!”
But The real challenge is bringing these data sets together across service providers and countries to analyse them in a consistent way and develop predictive models of activity. There’s a need for an automated approach, which is where cloud computing, robotics and artificial intelligence comes in.
“AI is playing a role in fisheries research and environmental monitoring,” Campbell adds. “Sub-surface gliders and autonomous surface vessels have already been developed and are in use around the world, and I could see their use expanding, as they’re a much cheaper platform to use for relatively straightforward tasks like collecting plankton samples or temperature data, compared to a fully-staffed research vessel.”
As more projects like this roll out within the fishing industry, it seems likely that AI, data analysis, and advances in monitoring technology will have an increasingly important role to play in enforcing regulation. Not only that, but this high-tech combination has the potential to increase efficiency for fishermen (in what can be a low margin business), while helping nations get a better sense of the impact fishing has on delicate ocean ecosystems.