2019_programme: REMOTELY CLASSIFYING FISH SPECIES OVER WIDE AREAS USING MULTISPECTRAL IMAGING
- Session: 03. Acoustic Monitoring of Ocean Environments and Processes: Biology, Ecology, Geophysics and Man-made activities
Organiser(s): Ratilal Purnima, Miksis-Olds Jennifer
- Lecture: REMOTELY CLASSIFYING FISH SPECIES OVER WIDE AREAS USING MULTISPECTRAL IMAGING [invited]
Paper ID: 1022
Author(s): Pednekar Shourav
Presenter: Pednekar Shourav
Presentation type: oral
Abstract: Classification of fish populations distributed over extensive undersea regions is essential for determining species diversity, reducing by-catch in fisheries and understanding the impacts of ocean warming on marine life. This is difficult to achieve with conventional species identification methods such as capture trawl and high-frequency echosounding, which are restricted to the vicinity of slow-moving research vessels. In recent years, Ocean Acoustic Waveguide Remote Sensing (OAWRS) technique has been used to assess fish abundance and study group behavior over mesoscales in near real time. Here, we demonstrate the ability to remotely classify mixed-species aggregations and simultaneously map population densities over wide areas using multispectral OAWRS. We do this by using low frequency transmissions at or near fish swimbladder resonance, where spectral characteristics of acoustic scattering vary significantly between fish species. We classify capelin and cod populations over a region of spatial overlap near the northern Norwegian coast in the Barents Sea. Dynamic changes in the composition and short-term behavior of capelin and cod groups are observed as they interact as a predator-prey unit. The multispectral imaging technique can be applied to remotely classify vast fish shoals and study multi-species interactions in regions important for biodiversity and fisheries.
- Corresponding author: Mr Pednekar Shourav
Affiliation: Massachusetts Institute of Technology
Country: United States
e-mail: