Acoustic tracking of migrating salmon

Area of Science:

• Ecology
• Bioacoustics
• Computer Science

Background:

• Accurate salmon population counts are crucial for fisheries management and conservation.
• Traditional methods like sub-sampling and manual measurements are labor-intensive and may lack precision.
• Variability in annual salmon runs necessitates reliable methods for assessing population sizes.

Purpose of the Study:

• To develop and validate an automated system for counting and sizing salmon during their spawning migrations.
• To implement a probability hypothesis density tracker for multi-target fish tracking using sonar data.
• To assess the accuracy of the automated system by comparing its results with established statistical models.

Main Methods:

• Utilizing dual-frequency identification sonar to collect data on fish in an insonified area.
• Processing sonar data into intensity images for fish location extraction via image processing.
• Applying a probability hypothesis density tracker to solve the multiple target tracking problem and generate fish tracks.
• Employing image segmentation techniques to calculate fish length from the generated tracks.

Main Results:

• The automated system successfully tracked and processed fish crossing the insonified area.
• Fish length information was accurately calculated using image segmentation on the tracked data.
• The algorithm demonstrated favorable comparisons with statistical models derived from sub-sampling and manual measurements.
• The system was tested on data from the 2010 Kenai River salmon run, validating its performance.

Conclusions:

• The implemented probability hypothesis density tracker provides an effective automated solution for counting and sizing salmon.
• This sonar-based approach offers a more efficient and potentially more accurate alternative to traditional fish assessment methods.
• The developed algorithm contributes to improved fisheries management by enabling precise monitoring of spawning salmon populations.