Skip to content
All posts

How to become data-driven in aquaculture

An intelligent system that is constantly and automatically learning can transform fish health for aquaculture farmers.

Data-driven-aquaculture

PwC recently wrote in its Seafood Barometer 2021 report that there is "low added value from standalone hardware investments, but when combined with data platforms and advanced analysis—the added value potential is huge."

Aquaculture farmers navigate volatile conditions and biological complexities every day. More and more, farmers are turning to data to support fish health: Since 2017, PwC noted the industry is moving from reactive to proactive decision-making by investing in data platforms and advanced analytics. A majority (91 percent) of the 55 aquaculture industry leaders surveyed for the report are looking to invest heavily in new technology to improve farm biology, sustainability, and profitability.

Becoming a data-driven organization is no small task, and it means much more than simply using a dashboard or having a data warehouse. To reach analytics maturity – i.e. the ability to manage and use data to make business decisions – data needs to become a core part of a farm’s culture.

Farms typically report numbers to meet regulations or measure progress. A data-driven farm goes beyond this, moving from reporting to analysis, action, and value:

  • Using data to make proactive, not reactive decisions.
  • Using data to answer, “why did this happen?” instead of “what happened?”
  • Using data to influence business decisions and prescribe actions.

This starts with using the right system to automatically collect, organize, analyze, and translate data to real-time insight.

Data collection

Too often, farms collect incomplete, inconsistent, or poorly formatted data. An investment in quality data collection, whether it’s through better sensors, cameras or other technology, both saves employee time and minimizes the risk of human error. Higher-quality data also means higher-quality insights down the line. This answers questions like:

  • How many times did we go over the government lice limit?
  • How many mortalities did each site have?
  • What were the average lice levels last year?
Mortality numbers on Manolin's farm data timeline, app.manolinaqua.com.
Mortality numbers on Manolin's farm data timeline, app.manolinaqua.com

Combining data

Farmers are inundated with tools to manage farm information. The data needed to make daily production decisions may be scattered between Excel sheets, emails, papers and databases. When information exists in silos, value is lost. Pooling all farm data into one place gives employees the most comprehensive, up-to-date picture of farm health at any given time answering questions such as:

  • What were the environmental conditions when these lice numbers dipped?
  • Did this mortality event occur after we treated for lice?
  • How did mechanical treatments impact feeding this year?
Lice overview on app.manolinaqua.com.
Lice overview on app.manolinaqua.com

Measuring data

Even the highest-quality dataset will mean nothing unless it’s accessible and searchable. Tools that slice, dice, analyze, and share data give farmers control over their information. With filtering, grouping and aggregation, important trends and patterns will emerge to help inform decisions, answering questions such as:

  • How is each generation comparing to the previous?
  • Do certain cages consistently perform better?
  • What are the factors leading to the highest fillet quality?
Machine learning models fuel Manolin's risk overview, app.manolinaqua.com.
Machine learning models fuel Manolin's risk overview, app.manolinaqua.com

Applying models

Fish health changes every day, and farmers could spend a lifetime digging deeper into analysis. But time is of the essence. Machine learning models will automatically ingest a constant stream of data and find an essentially infinite number of correlations and nuances between factors - insights impossible to find with the human eye. The models learn from information and adjust in real-time, not in retrospect. This gives farmers the tools to think ahead about questions such as:

  • What is my risk of PD this summer?
  • How will my neighbors’ sea lice situations impact me?
  • What is the optimal treatment option for this cage, in these conditions, and at this time?

The right foundation connects important information to those who need it sooner. In the end, though, humans are the ones that ask questions and make decisions – a data-driven farm combines the accuracy of computers with human intuition and experience.

Atlantic salmon

A culture shift

Often, data-driven strategies stop at the executive level. A truly data-driven company culture only gains traction when everyone, from the fish health biologists to financial planners, is empowered to use data to do their job better.

Solutions must be flexible. Data will have many different uses throughout an organization. Solutions cannot become a barrier to access: if it makes an employee’s job more difficult, he or she is not going to use it. Strategies need to meet the needs of all employees with an easy-to-use, flexible application.

Solutions must work with industry expertise. The aquaculture industry is full of unknowns, and many datasets are incomplete or poorly formatted. This doesn’t mean they’re not useful. The key is to augment this information with traditional industry knowledge. Data-driven solutions need to alert decision-makers, so they can combine insights with their expertise and experience.

Organizations must constantly test and improve. Each team needs to have confidence in and understand the value of data, so they’re empowered to collect better-quality information and use it to improve their work. And leaders should use this information to iteratively improve—strong data-driven strategies invest in a foundation but constantly revisit based on results.

In aquaculture, just a few hours can make or break fish health. Centralized data storage and alerts are key for farms located in remote areas, where information can take valuable time to get to those who need it. This can also mean the difference of millions in lost revenue: a 2019 report by Nofima and Kontali estimated that slaughtering 1 million fish at 3.5 kg rather than 5 kg costs a farm NOK 30 to 45 million (US$3.5 to 5.3 million).

A data-driven farm can not only reduce the risk of larger disasters but also continually optimize: more effectively using feed, expanding the feeding windows, minimizing time in the water and more.

With an intelligent system that is constantly and automatically learning, the potential impact is constantly growing.

This article originally published on TheFishSite.com. For more data, content, and insights, subscribe to Manolin’s newsletter.

Interested in working with us to help fuel aquaculture sustainability? Get in touch here.