2023_programme: Analysis of hydroacoustic time series by state space modelling
- Session: 23. Ambient Noise Sources
Organiser(s): -
- Lecture: Analysis of hydroacoustic time series by state space modelling
Paper ID: 1967
Author(s): Galka Andreas
Presenter: Galka Andreas
Abstract: Hydroacoustic time series, recorded in the ocean, may contain signal components originating from sources such as ships, marine mammals, explosions, etc. The aim of the present paper is to detect, separate and characterise such signal components, with particular emphasis on investigating hydroacoustic signatures of ships. We approach this aim by employing a class of parametric models from time series analysis, known as state space models. The models are fitted to the data by Kalman filtering and maximisation of the likelihood, using suitable algorithms for numerical optimisation. The application of Kalman filtering to such situations corresponds to solving an inverse problem.\nWe demonstrate that by state space modelling parametric estimates of the power spectra of the signal components can be obtained, which have attractive properties, as compared to classical non-parametric methods, furthermore filtering and noise reduction can be accomplished. We also discuss how the framework of state space modelling can be generalised to deal with phenomena which typically arise in the investigation of ship signatures, such as time-varying signal levels, combs of lines and Doppler effects.
Download the full paper
- Corresponding author: Dr Andreas Galka
Affiliation: Bundeswehr Technical Center for Ships and Naval Weapons, Maritime Technology and Research (WTD71)
Country: Germany
e-mail:
