A statistical-spectral method for echo classification

David A. Demer

doi:10.1093/icesjms/fsp054

Return

A statistical-spectral method for echo classification

2009-04-02

PRE

DOI：10.1093/icesjms/fsp054

Original

Original request for help

Save

Abstract

En 中文

The frequency dependence of sound-scatter intensity is commonly exploited to classify fish, zooplankton, and the seabed observed in acoustic surveys. Although less utilized, techniques based on the statistics of echo amplitudes can also be used to extract information. For example, single-frequency echo statistics have been used to determine whether backscatter originates from single or multiple fish or from rough or smooth seabeds, and to estimate scatterer sizes and densities. The efficacies of the amplitude-based techniques are challenged, however, by the usual requirement to group echo measurements to facilitate meaningful comparisons with model predictions. Groupings of data over space, time, or both, can combine scatter from multiple taxa or species, confounding the comparisons. These methods are improved with a hybrid, statistical-spectral method for target identification (SSID), which incorporates information contained in both the signal amplitudes and phases. The SSID uses multifrequency echo statistics from individual time-space intensities (pixels) to identify general scattering types, before applying model-based identification schemes for target identifications. The effectiveness of the SSID is demonstrated for fine-scale separation of scatter from demersal fish and the seabed and estimating seabed depth, within-beam slope, hardness and roughness, and the height of the dynamic acoustic dead zone.

Keywords:

acoustic

classification

dead zone

demersal

EK60

hardness

ME70

multifrequency

rockfish

roughness

seabed

slope

split-beam

statistical

target

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.