The role of ML in Sea Farming and Aquaculture
09 December 2021

Machine learning is starting to make waves in Sea Farming and land-based Aquaculture, what are the opportunities and risks?

The Aquaculture industry is not new to innovation, in fact it is a core element that has driven the growth of the industry to where it is today, a significant contributor of global food supplies, creating economical value around the world.

Machine learning (ML) is slowly being adopted as the industry looks to advanced analytics to help solve some of its most complex challenges, and to help the industry become more sustainable and operationally efficient.

Machine learning is also being adopted around the world by other industries, such as oil and gas and manufacturing. One thing each of these industries have in common is the volume of data available. Modern sea farms and land-based fish farms increasingly are able to collect large volumes of data, including environmental data such as weather, temperature, water salinity and oxygen levels, along with data on their fish which is collected from IoT sensors. Operational data is also collected, including fish mortality, escapes, and machinery condition.

Some of the challenges faced by sea farming and land-based fish farming today include;

  • Sea lice and other prevalent diseases in sea farming,
  • Mass mortality is a common challenge for both sea and land-based fish farming,
  • Operational costs, in particular energy costs for land-based aquaculture,
  • Managing water quality in land-based aquaculture is a challenge that greatly effects fish welfare, such as hydrogen sulphide which leads to toxification1
  • And both subsets look to improve operational efficiency, such as reducing man-power reliance in remote locations.


Before we explore how machine learning looks to help the industry, first let’s explain what machine learning is. 

Machine learning is a branch of artificial intelligence (AI) that explores how computer algorithms can improve automatically through experience and the use of data to make predictions and draw conclusions, imitating how humans learn.  

Machine learning can quickly learn from years of historical data from a wide range of sources and discover trends and root causes to outcomes. Combined with real-time data it is able to refine its learnings and predict future outcomes with accuracy.

Three ways machine learning can benefit sea-farming and land-based farming?

Time sensitive events

Fish farms, be it in the sea or on land, are dynamic ecosystems whereby conditions can change rapidly and have significant consequences, such as a mass mortality of a fish. Machine learning offers hope to operators by providing continuous predictive analytics of live data to detect minute deterioration in fish health, welfare, and environment, raising alarms so that operators can interview early.

Scalable Insights

Analysing and acquiring mathematical insights from data is not a new concept. However traditionally the ability to scale such efforts has been very difficult, due to computing power, limited data sets and the manual approach used by data scientists.Machine learning can speed up the process, making advanced analytics scalable.The inclusion of live data means that any algorithms can not only provide insights to historical events but can provide valuable continuous insights into future outcomes. At a corporate level, with the technology in place across a whole network of fish farms, companies will be able to have reliable forecasts for yields which will help with planning and management across the whole value chain.

Optimize Resources

Automating machine learning (Auto ML) with self service solutions reduce reliance on data science individuals and allow operators to run their own AI models to monitor their facilities and the health and welfare of their fish. Priority matrices provide operators with insights so they can optimize their resources, attending to the most critical and important issues first, and combined with IoT sensors, manpower is no longer required to collect manual samples and monitor the condition of shoals of fish. This also assists with safety in a sea-farming environment.
 
The application of automated machine learning in sea farming and land-based fish farms is very promising and could well be a critical next step in helping the aquaculture industry become more economical. However, in order to learn from its data, it is necessary to have quality data in the first place. This is the first hurdle that companies will have to overcome on their journey to implement advanced analytics.



Get in touch with our team to learn more



References:
  1. https://phys.org/news/2021-03-fish-farms-onshore.html


 
Social MediaShare this article