Deep Learning-based Bathymetry Retrieval without In-situ Depths using Remote Sensing Imagery and SfM-MVS DSMs with Data Gaps

TL;DR

The work tackles shallow-water bathymetry by combining refraction-corrected SfM-MVS DSMs with a deep learning-based spectral bathymetry method. It introduces Swin-BathyUNet, a U‑Net backbone augmented with Swin Transformer blocks and cross-attention, built to predict complete bathymetric maps from RGB orthoimages and co-registered, refraction-corrected DSMs, supported by a boundary-sensitive RMSE loss. Validated on two contrasting sites (Agia Napa, Cyprus and Puck Lagoon, Baltic) and extended to Sentinel-2 data, the approach delivers improved depth accuracy, reduced noise, and greater coverage compared with baseline methods, while achieving CATZOC/S-44-compliant predictions. The method reduces reliance on costly in-situ data, increases applicability across modalities, and offers near-real-time inference potential for coastal management and marine operations.

Abstract

Accurate, detailed, and high-frequent bathymetry is crucial for shallow seabed areas facing intense climatological and anthropogenic pressures. Current methods utilizing airborne or satellite optical imagery to derive bathymetry primarily rely on either SfM-MVS with refraction correction or Spectrally Derived Bathymetry (SDB). However, SDB methods often require extensive manual fieldwork or costly reference data, while SfM-MVS approaches face challenges even after refraction correction. These include depth data gaps and noise in environments with homogeneous visual textures, which hinder the creation of accurate and complete Digital Surface Models (DSMs) of the seabed. To address these challenges, this work introduces a methodology that combines the high-fidelity 3D reconstruction capabilities of the SfM-MVS methods with state-of-the-art refraction correction techniques, along with the spectral analysis capabilities of a new deep learning-based method for bathymetry prediction. This integration enables a synergistic approach where SfM-MVS derived DSMs with data gaps are used as training data to generate complete bathymetric maps. In this context, we propose Swin-BathyUNet that combines U-Net with Swin Transformer self-attention layers and a cross-attention mechanism, specifically tailored for SDB. Swin-BathyUNet is designed to improve bathymetric accuracy by capturing long-range spatial relationships and can also function as a standalone solution for standard SDB with various training depth data, independent of the SfM-MVS output. Experimental results in two completely different test sites in the Mediterranean and Baltic Seas demonstrate the effectiveness of the proposed approach through extensive experiments that demonstrate improvements in bathymetric accuracy, detail, coverage, and noise reduction in the predicted DSM. The code is available at https://github.com/pagraf/Swin-BathyUNet.

Figures (18)

• Figure 1: The four major pillars of the proposed methodology.
• Figure 2: Water refraction effects on the geometry of the apparent bottom DSM generation using SfM-MVS techniques.
• Figure 3: Spectrally Derived Bathymetry. With the sun as an illumination source, the formed image content comprises back-scatter components from atmosphere, water surface, water column and water bottom.
• Figure 4: The proposed Swin-BathyUNet architecture and a Swin Transformer Block. W-MSA is the multi-head self attention module with regular windowing configurations. Notation presented with Equation \ref{['equation_swin']}.
• Figure 5: (a) Example orthoimage patches of the Agia Napa area, (b) reference bathymetry, and (c) SfM-MVS refraction corrected bathymetry, relative to the WGS '84. In bathymetry images, white color represents the missing data (gaps).