Does Depth Matter Examining Factors that Could Influence the Acoustic Identification of Odontocete Species on Bottom Moored Recorders
Substantial advancements have been made in the identification of odontocete species based on the properties of their whistles and clicks. However, the suitability of species classifiers trained using data from the sea surface to analyze recordings obtained at depth is currently unknown. As a result,...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Text |
| Language: | English |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://www.dtic.mil/docs/citations/AD1014254 http://oai.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=AD1014254 |
| Summary: | Substantial advancements have been made in the identification of odontocete species based on the properties of their whistles and clicks. However, the suitability of species classifiers trained using data from the sea surface to analyze recordings obtained at depth is currently unknown. As a result, it remains unclear how depth, distance of animals from the recorder and sound propagation influence classification results. If classifiers perform differently on data recorded at depth than at the surface, it may be necessary to re-train them to ensure accurate results. Similarly, if the behavior of animals or signal propagation affects the identification of species using echolocation clicks, this must be understood and integrated into analysis methods. In this project, we examine how species-specific signaling cues are affected by recording depth by using both surface-deployed and bottom-moored vertical arrays of hydrophones and autonomous recorders to obtain recordings at different depths in the water column from a variety of free-ranging odontocete species. We also use localization and sound propagation modeling to address the following questions: Does the depth at which dolphin whistles are recorded affect the signal properties observed? If so, does the performance of species classifiers developed using whistles recorded at the surface change when applied to data from bottom-moored recorders? Do reported species-specific click characteristics remain consistent across recording depths? If they do, are the cues consistent across behaviors, such as diving, surface milling and travel? We are at the beginning of a two-year effort to examine these questions. The results of this effort will lead to a greater understanding of the strengths and weaknesses of the acoustic species identification tools being employed for marine mammal monitoring and mitigation by the U.S. Navy. |
|---|