Assessing the use of harmonized multisource backscatter data for thematic benthic habitat mapping

Benjamin Misiuk,Myriam Lacharité,Craig J Brown

doi:10.1016/j.srs.2021.100015

Benjamin Misiuk, Myriam Lacharité + Show 1 more

Open Access

https://doi.org/10.1016/j.srs.2021.100015

Copy DOI

Abstract

Legacy seabed mapping datasets are increasingly common as the need for detailed seabed information is recognized. Acoustic backscatter data from multibeam echosounders can be a useful surrogate for seabed properties and are commonly used for benthic habitat mapping. Legacy backscatter data, however, are often uncalibrated, rendering measurements relative to a given survey and complicating the use of multisource acoustic datasets for habitat mapping. Recently, ‘bulk shift’ methods have been proposed to harmonize multisource backscatter layers that overlap spatially, but their application to benthic habitat mapping has not been evaluated. Here, four relative backscatter datasets at the St. Anns Bank Marine Protected Area were harmonized to produce a single continuous surface spanning the extent of available bathymetric data. The harmonized surface was used as a predictor in a benthic habitat (‘benthoscape’) classification, which was compared to previous results using individual backscatter coverages. Results were similar to those obtained previously, but the harmonized surface provided increased class discrimination, fewer unclassified areas, and predictions that cross dataset boundaries – eliminating the need for manual reclassification by the user. While this generally increases the efficiency and repeatability of the analysis and the useability of the data, we caution that an inappropriate harmonization model is a potential source of error for the classification.

Highlights

One of the primary benefits of these sonar technologies is their ability to measure both seafloor bathymetry, from which geomorphology can be derived through the production of digital bathymetric models of the seafloor, and seafloor composition, which can be inferred from the signal of the returning echo – a property referred to as acoustic backscatter intensity (De Falco et al, 2010; Lamarche and Lurton, 2018; Lamarche et al, 2011; Lurton, 2010)
We investigate the applicability of bulk shift backscatter harmonization to benthic habitat classification through a comparative study using individual and harmonized backscatter layers
Few dataset compilation artefacts remained after harmonization, even at the boundary of the Reson7111 survey, which was originally of disparate scale, with arbitrary units ranging between ~130 and 180

Summary

Introduction

Acoustic remote sensing of the seafloor has been increasingly adopted over the past three decades for a wide variety of applications, including geological mapping (e.g., Ferrini and Flood, 2006; Hughes Clarke et al, 1996; Misiuk et al, 2018; Plets et al, 2012; Stephens and Diesing, 2014; Todd et al, 1999), marine archaeology (e.g., Passaro et al, 2013; Plets et al, 2011), seafloor environmental change monitoring (e.g., Montereale-Gavazzi et al, 2018; Montereale-Gavazzi et al, 2019; Snellen et al, 2019; van Rein et al, 2011) and benthic habitat studies (e.g., Boswarva et al, 2018; Brown et al, 2012; Kostylev et al, 2003; Lacharite et al, 2018). Sediment-acoustic relationships from MBES are complex due to the influence of angular dependency of the signal intensity caused by ensonification geometry across the MBES swath (Fonseca et al, 2009; Lamarche et al, 2011; Malik, 2019). This is further complicated by the operating frequency of the MBES, and resulting frequency-dependent interaction between the acoustic signal and substrate (Brown et al, 2019). Despite the complexities of signal analysis, progress has been made in developing post-processing methods wherein backscatter can be rendered useful for delineating differences in substrate and habitat (see Lurton and Lamarche, 2015 for a detailed review)

Objectives

Results

Discussion

Conclusion