WaveDINO: Learning-Based Atmospheric Correction of Unwrapped InSAR Interferograms Validated by GNSS: Results at Laguna del Maule and Campi Flegrei Volcanoes

2026-06-15 · Source: Computer Vision and Pattern Recognition · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computer Vision & Pattern Recognition, Environmental Science & Earth Systems · Depth: Expert, quick

Summary

WaveDINO introduces a novel learning-based method for denoising unwrapped Interferometric Synthetic Aperture Radar (InSAR) interferograms, specifically addressing atmospheric phase delays and other corruption effects that hinder volcanic deformation monitoring. This wavelet-based multi-scale denoising framework is conditioned on frozen DINOv3 foundation-model features and terrain information. It employs a hybrid training strategy, combining physically motivated synthetic deformation with real atmospheric noise to achieve robust performance. Validated by independent GNSS measurements at Laguna del Maule (Chile) and Campi Flegrei (Italy), WaveDINO consistently outperforms competing models, reducing mean GNSS misfit by approximately 3% and 19% at these sites, respectively, and surpassing weather-model-based corrections.

Key takeaway

For research scientists and geophysicists monitoring volcanic deformation, WaveDINO offers a significant advancement in InSAR data processing by effectively mitigating atmospheric noise. You should consider integrating learning-based, hybrid-trained models like WaveDINO into your analysis pipelines to achieve higher accuracy and reliability in deformation measurements, especially where traditional weather models fall short. This approach can refine your understanding of subtle ground movements.

Key insights

WaveDINO employs hybrid training and DINOv3 features for superior learning-based atmospheric correction in InSAR.

Principles

• Hybrid training improves model robustness against real noise.
• Multi-scale wavelet denoising effectively separates signals.
• Foundation model features enhance contextual understanding.

Method

WaveDINO uses a wavelet-based multi-scale denoising framework, conditioned on frozen DINOv3 features and terrain, trained with synthetic magma-source deformation superimposed on short-term interferograms.

In practice

• Apply WaveDINO for precise InSAR deformation monitoring.
• Integrate DINOv3 features in geospatial analysis tasks.
• Utilize hybrid training for robust noise reduction models.

Topics

• InSAR
• Atmospheric Correction
• WaveDINO
• DINOv3
• Volcanic Deformation
• GNSS Validation

Best for: AI Scientist, Research Scientist, Computer Vision Engineer

Related on AIssential

• GNSS-FM: A Self-Supervised Foundation Model for Daily GNSS Displacement Time Series — GNSS-FM introduces self-supervised pretraining to overcome labeled data scarcity in GNSS time series analysis, outperforming supervised baselines.
• Beyond Backscatter: InSAR coherence from detected SAR images — Deep learning can accurately regress InSAR coherence from detected SAR images, bypassing complex coregistration.
• Kaggle Winners Walkthroughs: Yale/UNC-CH - Geophysical Waveform Inversion with Team greySnow — Combining physics-based models with machine learning enhances Full Waveform Inversion for subsurface imaging.
• Machine-Learning Emulation of Satellite Greenhouse Gas Retrievals: Stability over Time — Machine-learning emulators for GHG retrievals require time-aware training to maintain accuracy and stability over extended periods.
• Physics-Guided Spatiotemporal Learning for Coastal Wave Peak Period Estimation from Video — Physics-guided deep spatiotemporal learning from video can accurately and consistently estimate coastal wave peak periods, overcoming traditional monitoring limitations.
• WILD-SAM: Phase-Aware Expert Adaptation of SAM for Landslide Detection in Wrapped InSAR Interferograms — WILD-SAM adapts SAM for landslide detection in InSAR data by integrating phase-aware spectral alignment and wavelet-guided prompt enhancement.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.