An Unsupervised Acoustic Description of Fish Schools and the Seabed
21 August 2017
Ben Scoulding, along with Sven Gastauer and Miles Parsons have recently written a paper on target classification, habitat descriptors and geostatics.
Fisheries acoustics is now a standard tool for monitoring marine organisms. Another use of active-acoustics techniques is the potential to qualitatively describe fish school and seafloor characteristics or the distribution of fish density hotspots. Here, we use a geostatistical approach to describe the distribution of acoustic density hotspots within three fishing regions of the Northern Demersal Scalefish Fishery in Western Australia. This revealed a patchy distribution of hotspots within the three regions, covering almost half of the total areas. Energetic, geometric and bathymetric descriptors of acoustically identified fish schools were clustered using robust sparse k-means clustering with a Clest algorithm to determine the ideal number of clusters. Identified clusters were mainly defined by the energetic component of the school. Seabed descriptors considered were depth, roughness, first bottom length, maximum Sv, kurtosis, skewness and bottom rise time. The ideal number of bottom clusters (maximisation rule with D-Index, Hubert Score and Weighted Sum of Squares), following the majority rule, was three. Cluster 1 (mainly driven by depth) was the sole type present in Region 1, Cluster 2 (mainly driven by roughness and maximum Sv) dominated Region 3, while Region 2 was split up almost equally between Cluster 2 and 3. Detection of indicator species for the three seabed clusters revealed that the selected clusters could be related to biological information. Goldband snapper and miscellaneous fish were indicators for Cluster 1; Cods, Lethrinids, Red Emperor and other Lutjanids were linked with Cluster 2, while Rankin Cod and Triggerfish were indicators for Cluster 3.
To read the full paper, click here.