Over the past two decades, a decline in the abundance of eels has been observed in parts of North America. This has raised concern about injuries and mortalities that may occur when outmigrating adult (i.e., silver-phase) eels pass through hydropower turbines. As a result, the Federal Energy Regulatory Committee has recently included measures to protect eels as part of the licensing requirements for hydroelectric facilities. Our study was commissioned by Madison Paper Industries, who own and operate two adjacent projects on the Kennebec River, Maine. The work was conducted in cooperation with Kleinschmidt Associates in the intake canal of the Anson Project, a small hydroelectric facility with five turbine/generator units. The goal of the study was to test the feasibility of developing a hydroacoustic monitoring system for the detection of downstream migrating adult eels. Such a system could potentially be used as an early warning system to initiate actions (altering generation or opening waste gates) that would allow eels to pass without going through the turbines.
We used two split-beam systems (BioSonics DTX, Simrad EK 60) and a DIDSON (dual-frequency imaging sonar) system to evaluate the effectiveness of different types of sonar technology. In an intial test, live yellow-phase and silver-phase eels, and other indigenous species were tethered in the acoustic field to determine the distance from the transducer at which eels could be detected and distinguished from non-eel targets.
DIDSON images allowed confident identification of large eels (>90cm), based on shape and anguilliform swimming motion, out to a maximum range of approximately 20 m. While the split-beam systems were able to detect eels at longer ranges (27 m), the features used for positive split-beam identification of eels (sawtooth pattern of the echo trace and varying echo width) were more subtle and sometimes more ambiguous than the DIDSON images. We monitored the natural eel migration over a 2-week period and detected and identified well over 400 eels.
The information provided by the DIDSON data on target shape and motion made this type of acoustic system a good candidate for the development of a classification algorithm that could be used for an automated eel-counting system. Target shape and motion are important as the classification system needs to be able not only to positively identify eels, but also be accurate enough to reject more than 99% of debris and other targets, which sometimes outnumbered eels by a ratio of ~100 :1. Target tracking and classification tests done by Pacific Eumetrics, Ltd., with DIDSON data collected in this study have shown promising results.
Reference
Interim American Eel Downstream Passage 2005 Pilot Study Report. Report prepared for Madison Paper Industries,
Madison, Maine.