The Centre for Fisheries Monitoring and Communication System (CFMCS), a subordinate agency under the Russian Federal Agency for Fishery (Rosrybolovstvo), has developed a system for forecasting fish stocks using mathematical modeling and artificial intelligence.
The first result of testing the system in operational conditions was the discovery of the ivasi sardine, which this year left the traditional fishing region for fishermen.
"The program analyzed the possible areas of accumulation of ivasi sardines," said Ilya Shestakov, Head of Rosrybolovstvo.
"Outside the Russian economic zone, they found fish and sent a fishing fleet there. We plan to use this system in the future, including for forecasting the approaches of other types of aquatic biological resources. The model will be finalised and improved."
The system aggregated satellite and hydrological data. At the next stage, specialised software processed the information and, based on the trained model, built maps of probability zones.
The maps were sent daily to the vessels participating in the experiment, providing captains with a source of information for fishery planning. The materials can be compared with the data of ship observations, echo sounders, and other sensors for decision-making.
A key feature of the system is its flexibility and learnability. For each new fish species, the model goes through a learning phase on the corresponding historical data.
CFMCS specialists are now working on adapting the model for pollock and mackerel. There are also plans to forecasting the situation of the saury fishery.
The development is based on a dynamic habitat model that is trained on large amounts of historical data for each fish species.
To assess the probability of commercial fish aggregations, the model analyzes key parameters including: temperature; salinity; sea level; currents; concentration of zooplankton and chlorophyll; wind speed and direction; cloudiness; the phase of the moon; and the impact of earthquakes on the behavior of fish.
