The rapid increase in sensor capacities and computational power are opening new avenues for the aquaculture industry. Mirroring developments in agriculture, interconnected Internet of Things (IoT) sensors, big data analytics, and Artificial intelligence (AI) promise to revolutionise aquaculture supply and value chains. However, the application of the Precision Agriculture framework and tools is very challenging, requiring detailed knowledge on a three-dimensional system in the harsh ocean environment: it is not easy to observe what is happening in a 20 m deep cage containing about 150,000 individuals, while making sense of these observations in a chaotic environment pose very real challenges!
GAIN comes on the scene at this very exciting time: we are assessing the performances of new, market-ready sensors for non-invasive monitoring of fish distribution and behaviour, as well as key environmental variables (e.g. water temperature, dissolved oxygen). We are processing these complex data using machine learning and big data analytics, discovering patterns and anomalies which can facilitate the optimisation of feeding and other husbandry operations (e.g. net changes).
In late June, I left a very hot Venice and traveled north to Norway, to the midnight sun village of Inndyr, in Nordland County, to discuss the preliminary results of our comprehensive monitoring programme with Giulia and Ronald (GIFAS), who have installed and are looking after the sensors, and with Fearghal (IBM Research), expert in AI and IoT. Inndyr is located in the Gildeskål municipality which, in Viking times, was a renowned meeting point for the whole region. Aquaculture is the backbone of the local economy, to which our industry partner GIFAS (Gildeskål Forskningsstasjon AS) provided a relevant contribution in its thirty years of activities, which were celebrated this year. Far from being seen as an environmentally unfriendly activity, in this area aquaculture provides jobs, fosters educational activities and also represents a touristic attraction. Domus Pisces is a building owned by Nordland county and used by both the local highschool (Meløy VGS avd. Inndyr) and GIFAS. GIFAS uses it both for running cleaner fish tank-based trials and for promotion of aquaculture. For the latter purpose, GIFAS has an aquarium containing salmon, a hall exhibiting informative, aquaculture-themed posters and a small souvenir shop. In addition, the promotional centre of GIFAS runs tours to its research and commercial sites and these services are used by tourists, school and universities, amongst others.
Giulia and Ronald welcomed myself and Fearghal at GIFAS headquarters in Inndyr. We sat down in a cosy meeting room and started looking at data collected in the last five months. Our observing system was installed at GIFAS salmon production site Rossøya, about 10 minutes by boat from Inndyr.
We equipped one of the 90 m circumference cages that you can see in the picture above with ABM, a close-to-market system for detecting individual fish position and estimating its weight and swimming speed. The 20 m-deep cage was stocked with about 150,000 fish: we have been following their growth and behaviour day by day since February 2019.
ABM was selected because it could represent the silver bullet for dealing with the three-dimensionality issue: it provides data along the vertical water column and, with up to 50,000 detections per day, statistics concerning fish average weight are based on a representative sample size. Fish distribution is displayed every five minutes on a dashboard, allowing barge operators to inspect their feeding behaviour (feeding fish tend to congregate near the surface), as seen in Fig.1.
We also deployed sensors for measuring water temperature (below, left), dissolved oxygen (below, right) and water current every 10 minutes. The upper-portion of the cage is surrounded by an impermeable barrier to prevent lice entry. This may also reduce the water flow and circulation, thus affecting dissolved oxygen levels which may affect fish appetite and, ultimately, growth.
The data are flowing to a cloud platform designed by IBM: Fearghal and his colleagues are crunching these numbers in order to extract useful information to improve feeding efficiency, provide accurate predictions of salmon growth, and disseminate early-warning on anomalous patterns. We are eager to see the results: keep following the blog and you’ll be the first to know!