This document discusses several projects using data to help plan small pelagic fisheries and oceanic tuna fisheries. It describes analyzing historical catch and market data to determine optimal fishing locations and times. It also analyzes the relationship between catches and environmental factors to predict fishing yields. Machine learning models are developed using both open and private vessel tracking and catch data to predict the best daily catch locations without exposing sensitive private data. Visualizations of catch data, ocean conditions, and vessel locations are provided to fishing operators to help with planning.

1. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
1
Small pelagic fisheries web portal
Historical catch and market analysis (SINTEF Ocean)
- Analysis of historical catch data
- Dependencies between prices,
season, location and species.
- Used for planning when and
where to fish for the various
species, to optimize value.
- Data (2012 - ):
- Small pelagic catches
- Trade information

2. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
2
Small pelagic fisheries web portal
Historical catch and environment analysis (SINTEF Ocean)
- Analysis of how historical catches
has depended on environmental
factors.
- Investigate covariance between
catches and e.g. zooplankton
concentrations.
- Data (2012 -):
- Small pelagic catches
- Earth observations
- Meteorological simulations
- Oceanographic and biomarine
simulations.

3. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
3
Small pelagic fisheries web portal
Marine environment forecasts
- Forecasts for the marine
environment.
- Supports choice of fishing
grounds for the next days.
- Data (the last days):
- Earth observations
- Meteorological simulations
- Oceanographic and
biomarine simulations.

4. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
4
Oceanic tuna fisheries planning
Tuna fisheries and Copernicus meteogeobiochemical data in
the Indian Ocean
Jose A. Fernandes, Igor Granado, Iñaki Quincoces
Conceptual diagram of data and components flow

5. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
5
Oceanic tuna fisheries planning
Conceptual diagram of forecasting model based on pipeline of
supervised classification methods (Fernandes et al., 2010)
Jose A. Fernandes, Igor Granado, Iñaki Quincoces
Example of species distribution probabilistic forecast based on
Copernicus environmental data and fishing events
Satellite data Vessel data
Probabilistic
forecasting
Performance:
• Absence accuracy: ~ 80% (what to avoid)
• High biomass false positive: ~25% (where to go)
• Vessels fuel reduction achieved by a tuna company
vessels during DataBio project is between 4% and
30% with a 19% reduction on average
Fernandes. J.A., Quincoces, I., Fradua, G, Ruiz, J., Lopez, J., Murua, H., Inza, I., Lozano, J.A., Irigoien, X., Santiago,
J. Fishery pilot B1: Planning of oceanic tuna fisheries - Arrantza B1 kasua: Atun tropikalaren arrantza plangintza.
DataBio general assembly 02 (Helsinki), 27-29 June, DOI: 10.13140/RG.2.2.22519.32165.
Fernandes J.A., Irigoien X., Goikoetxea N., Lozano J.A., Inza I., Pérez A, Bode A. (2010) Fish recruitment
prediction, using robust supervised classification methods. Ecol. Model. 221(2): 338-352.

6. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
6
Open catch data:
• All catches last 20 years
• Coarse in location and time
(catch zones /landing time)
• No catch value
WP4: Team Fish: Machine learning of best
catch locations in open and private data
Private catch data:
• One fishery company
• Precise location and times
(catch position and time)
• Catch value
ML Objective: Predict best daily catch location
• Focus on best vessels/captains for a specific fishery
• Model comparison for open vs private data
=> Same fish species and vessel class (cod & trawlers)
=> Test same MPC model (LOESS) on both data sets
• Later: Best model per dataset and optimal combination
MPC: Multi-Party-Computation using CYBERNETICA Sharemind
Fisheries analytics and prediction models can be trained on the union
of open and sensitive data sets from multiple users:
… without exposing the private data sets to each-other
… collation & linking with open data can be done once
… less total work resultinging in better models for fisheries

7. This document is part of a project that has received funding
from the European Union’s Horizon 2020 research and innovation programme
under agreement No 732064. It is the property of the DataBio consortium and shall not be distributed or
7
WP4: Team CODeFish: Open + Private Catch
Data Analytics & SINTIUM Visualisation
• Open Norwegian
catch data (20years)
• Catch data drill down
• Species, tools, time,
weight, volume
• Copernicus (CMEMS):
• Currents (animated)
• Sea Surface
Temperature
• Live AIS
• Vessel positions
• Machine learning
• Fishing activity from
live AIS by Global
Fishing Watch ML
model
• Catch prediction from
private data (whitefish
data)