Performance of Multibeam Echosounder Backscatter-Based Classification for Monitoring Sediment Distributions Using Multitemporal Large-Scale Ocean Data Sets

Performance of Multibeam Echosounder Backscatter-Based Classification for Monitoring Sediment Distributions Using Multitemporal Large-Scale Ocean Data Sets

Snellen, M. (TU Delft Aircraft Noise and Climate Effects; Deltares)
Gaida, T.C. (TU Delft Aircraft Noise and Climate Effects)
Koop, L. (TU Delft Aircraft Noise and Climate Effects)
Alevizos, Evangelos (GEOMAR Helmholtz Center for Ocean Research)
Simons, D.G. (TU Delft Aircraft Noise and Climate Effects)

2018-02-08

Obtaining an overview of the spatial and temporal distribution of seabed sediments is of high interest for multiple research disciplines. Multibeam echosounders allow for the mapping of seabed sediments with high area coverage. In this paper, the repeatability of acoustic classification derived from multibeam echosounder backscatter is addressed. To this end, multibeam echosounder backscatter data acquired on the Cleaver Bank (North Sea) during five different surveys is employed using two different classification methods, i.e., a method based on the principal component analyses and the Bayesian technique. Different vessels were used for the different surveys. The comparison of the classification results between the different surveys indicates good repeatability. This repeatability demonstrates the potential of using backscatter for long-term environmental monitoring. However, the use of different classification methods results in somewhat different classification maps. Monitoring, therefore, requires the consistent use of a single method. Furthermore, it is found that the statistical characteristics of backscatter is such that clustering algorithms are less suited to discern the number of sediment types present in the study area. The Bayesian technique accounting for backscatter statistics is therefore recommended. A strong positive correlation between backscatter and median grain size for finer sediments (<0.5 mm) using a frequency of 300 kHz is observed within the study area, but an ambiguity is found for sediments with median grain sizes >0.5 mm. Consequently, for the situation considered a unique assignment of sediment type to acoustic class is not possible for these coarser sediments.

Backscatter
Sediments
Acoustics
Principal component analysis
Bayes methods
Grain size
Sea measurements

http://resolver.tudelft.nl/uuid:2b02e1d2-0e66-47fa-98f0-0b66fc1d12d9

https://doi.org/10.1109/JOE.2018.2791878

0364-9059

IEEE Journal of Oceanic Engineering

Institutional Repository

journal article

© 2018 M. Snellen, T.C. Gaida, L. Koop, Evangelos Alevizos, D.G. Simons