AUTOMATIC CLASSIFICATION OF LOW-FREQUENCY ARCTIC AMBIENT NOISE

PDF

Authors
  1. Greening, M.V.
Corporate Authors
Defence Research Establishment Atlantic, Dartmouth NS (CAN);Datavision Computing Services Ltd, Victoria BC (CAN)
Abstract
A technique is described which can be used to automatically classify Arctic ambient noise data collected using an array of hydrophones suspended below the ice. The technique is currently designed for analysing low frequency (5-60 Hz) noise collected from the central Arctic pack ice. It is capable of distinguishing between nearby or distant active pressure-ridging events, thermal ice-cracking events, biological noise, cable strum, flow noise and microseisms. Different noise types are distinguished by the power spectra, cross spectral matrices and noise directiviety of the received sound. The technique is designed as a real time system which can automatically classify the dominant noise contained in the data over a specified time interval. The technique is applied to data obtained in the Lincoln Sea over five days in April 1988. Application of the algorithms produced a quantitative classification of the noise type which was clearly recognized in displays of the distinguishing measures.
Keywords
Lincoln Sea;Pressure ridging;Thermal ice cracking;Biological noise;Cable strum;Flow noise;Noise classification
Report Number
DREA-CR-95-496 — Contractor Report
Date of publication
01 Aug 1995
Number of Pages
29
DSTKIM No
96-01873
CANDIS No
498523
Format(s):
Document Image stored on Optical Disk;Hardcopy

Permanent link

Document 1 of 1

Date modified: