From Bands to Depth: Understanding Bathymetry Decisions on Sentinel-2

2026-01-19

AI Summary

This study addresses the poor robustness and lack of interpretability of Sentinel-2-based remote sensing bathymetry models when applied across diverse regions. To this end, the authors propose Swin-BathyUNet, integrating spectral band ablation analysis with A-CAM-R—an attention visualization technique tailored for regression tasks—to uncover the critical pixels and the dominant role of green and blue spectral bands in shallow-water depth estimation. The research reveals that conditional cross-attention in the decoder enhances model robustness against glare and foam artifacts and quantifies a linear increase in cross-regional prediction error with water depth. Furthermore, through strategic preprocessing and fine-tuning, the model’s generalization capability and prediction reliability are significantly improved.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Deploying Sentinel-2 satellite derived bathymetry (SDB) robustly across sites remains challenging. We analyze a Swin-Transformer based U-Net model (Swin-BathyUNet) to understand how it infers depth and when its predictions are trustworthy. A leave-one-band out study ranks spectral importance to the different bands consistent with shallow water optics. We adapt ablation-based CAM to regression (A-CAM-R) and validate the reliability via a performance retention test: keeping only the top-p% salient pixels while neutralizing the rest causes large, monotonic RMSE increase, indicating explanations localize on evidence the model relies on. Attention ablations show decoder conditioned cross attention on skips is an effective upgrade, improving robustness to glint/foam. Cross-region inference (train on one site, test on another) reveals depth-dependent degradation: MAE rises nearly linearly with depth, and bimodal depth distributions exacerbate mid/deep errors. Practical guidance follows: maintain wide receptive fields, preserve radiometric fidelity in green/blue channels, pre-filter bright high variance near shore, and pair light target site fine tuning with depth aware calibration to transfer across regions.

Problem

• satellite-derived bathymetry
• Sentinel-2
• cross-region inference
• depth-dependent degradation
• water depth estimation

Innovation

• Swin-BathyUNet
• satellite-derived bathymetry
• attention ablation
• A-CAM-R
• cross-region inference