Abstract
The overarching goal of this effort is to provide information for assuring compliance with the Species at Risk Act (SARA) for tidal energy devices installed in the Bay of Fundy, Nova Scotia, Canada. Part of that goal is to monitor one of those species at risk, the harbour porpoise (Phocoena phocoena), to assess whether the devices are having any impact on the population. To do this, it is necessary to collect information on the occurrence of the porpoises over time, which is done by making underwater acoustic recordings and finding out how much of the time the porpoises are present. These recordings are sufficiently extensive, typically spanning months, that manual review is not practicable. An automated method is needed to scan the recordings for sounds of harbour porpoises and report the results in a way that can be used for SARA compliance.
The goal of this project was to build such an automated system. The approach taken was to construct a multi‐stage system that takes as input a collection of recorded sound files and produces as output a series of tables and figures summarizing harbour porpoise presence found from the recordings. The structure of this was a system called FindPorpoises for automated detection and classification of harbour porpoise echolocation clicks, followed by a module for producing the desired tables and figures summarizing harbour porpoise presence. FindPorpoises in turn comprises three internal stages, for click detection, click classification, and click grouping.