Cross-Comparison of the “BathySent” Coastal Bathymetry to Sonar Measurements and Ratio Model Technique: Pilot Sites in the Aegean Sea (Greece)

Abstract

The proposed novel “BathySent” approach for coastal bathymetric mapping, using the Copernicus Sentinel-2 mission, as well as the assessment and specification of the uncertainties of the derived depth results, are the objectives of this research effort. For this reason, Sentinel-2 bathymetry retrieval results for three different pilot sites in Greece (islands of Kos, Kasos, and Crete) were compared with ground-truth data. These data comprised high-resolution swath bathymetry measurements, single-beam echosounder measurements at very shallow waters (1–10 m), and the EMODnet DTM 2018 release. The synthetic tests showed that the “BathySent” approach could restitute bathymetry in the range of 5–14 m depth, showing a standard deviation of 2 m with respect to the sonar-based bathymetry. In addition, a comparison with the “ratio model” multispectral technique was performed. The absolute differences between conventional Earth Observation-based bathymetry retrieval approaches (i.e., linear ratio model) and the suggested innovative solution, using the Sentinel-2 data, were mainly lower than 2 m. According to the outcome evaluation, both models were considered to provide results that are more reliable within the depth zone of 5–25 m. The “ratio model” technique exhibits a saturation at ~25 m depth and demands ground calibration. Though, the “BathySent” method provides bathymetric data at a lower spatial resolution compared to the “ratio model” technique; however, it does not require in situ calibration and can also perform reliably deeper than 25 m.