Semi-Automated Classification of Side-Scan Sonar Data for Mapping Sabellaria spinulosa Reefs in the Brown Bank, Dutch Continental Shelf

Biogenic reefs support marine biodiversity and play a key role in a healthy marine environment. Protecting and enhancing reef-building species, such as Sabellaria spinulosa, require mapping and monitoring strategies. A multi-scale and multi-sensor mapping campaign, including a multi-beam echosounder, side-scan sonar (SSS), box corer and ROV with an attached video camera, has been carried out in the northern Brown Bank (Dutch Continental Shelf) in August 2023. A semi-automated classification workflow, based on a support vector machine (machine learning), was developed to map Sabellaria reefs using SSS and video data. Elevated Sabellaria reefs were classified with a precision and sensitivity of 52% and 49%, respectively. The classified SSS images were merged into full-coverage percentage maps of Sabellaria reef coverage. Located between the swales of the tidal ridges, it was estimated that the reefs cover an area of 3.8 to 5.7% within the surveyed areas. The maps indicate (1) on the large-scale a preference of Sabellaria spinulosa for settlement to the east of the deepest part of the swale and (2) on the small-scale a preference for the troughs towards the stoss side of the megaripples. The employed survey strategy and the developed classification workflow can be extended to other environmental areas and further developed into a standard monitoring procedure.