AI-POWERED INSIGHTS FOR ENTERPRISE
Soil Displacement Estimation from Integrated Sensing Technologies in Data-Driven Models Biased by Temporal Coherence of PS-InSAR
This paper presents a coherence-based framework for estimating soil displacement from PS-InSAR data combined with ground sensor measurements. It introduces a method to identify an optimal temporal coherence threshold using Sparse Bayesian Learning (SBL) uncertainty, ensuring reliable displacement time series. The SBL model integrates satellite line-of-sight displacement with soil temperature and humidity to reconstruct denser displacement histories and enable continuous monitoring. The approach demonstrates how low-cost in-situ sensors can effectively complement remote sensing for dynamic soil behavior tracking, overcoming limitations of traditional DInSAR and enhancing spatial/temporal coverage.
Executive Impact: Enhanced Geohazard Monitoring
This research significantly advances ground deformation monitoring for critical infrastructure and geohazard management. By integrating PS-InSAR with in-situ data and applying advanced machine learning, it offers unparalleled accuracy and temporal resolution.
0
Improvement in prediction accuracy for ground displacement
0
Identified optimal temporal coherence threshold for reliable data
0
Increase in temporal density of displacement monitoring with in-situ integration
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
InSAR Processing
Multi-temporal InSAR techniques overcome decorrelation limitations by processing long image stacks to separate various phase contributions, enabling millimetric displacement measurements. PS-InSAR, in particular, identifies phase-stable targets ('permanent scatterers') to reconstruct their displacement histories, offering superior precision and temporal continuity even with atmospheric noise. The SARPROZ POLIMI model is used to estimate residual height, linear/seasonal deformation, and temporal coherence for each scatterer.
Sparse Bayesian Learning (SBL)
SBL is used to model the relationship between environmental variables (temperature, soil moisture) and satellite-observed displacements. It formulates a probabilistic regression problem, employing a dictionary of nonlinear basis functions (Fourier and polynomial series) and estimating weights using the Relevance Vector Machine (RVM) algorithm. SBL provides explicit estimates of predictive uncertainty, which is crucial for defining data reliability. The model is trained, validated, and tested to ensure generalization capability and prevent overfitting.
Coherence-Based Thresholding
Temporal coherence in PS-InSAR measures phase stability over time (0 to 1). This study treats coherence not just as a quality indicator but as an explicit design variable. By analyzing how SBL's predictive uncertainty changes with coherence, an optimal threshold (0.56 in this study) is identified. Below this threshold, uncertainty increases abruptly, making data unreliable for modeling. This ensures only sufficiently reliable PS data is used for training.
Optimal Coherence Threshold
0.56 Minimum temporal coherence for reliable PS-InSAR data.Coherence-Based Monitoring Workflow
PS-InSAR Data Processing
→
Temporal Coherence Estimation
→
SBL Uncertainty Analysis
→
Optimal Threshold Determination
→
Filtered PS Data for SBL Training
→
Continuous Soil Monitoring & Prediction
| Feature | Traditional PS-InSAR | Proposed Integrated Method |
|---|---|---|
| Feature |
|
|
| Data Reliability Filtering |
|
|
| Continuous Monitoring |
|
|
| Cost-Effectiveness |
|
|
| Bias Reduction |
|
|
Vicoforte Sanctuary Ground Movement Monitoring
Description: The study applied the integrated sensing framework to monitor ground settlements at the Sanctuary of Vicoforte, Italy, an important heritage site with complex soil conditions.
Challenge: Continuous and reliable monitoring of soil deformation, especially slow-moving components, is crucial for heritage sites but challenging due to sparse in-situ networks and PS-InSAR limitations (e.g., temporal coherence bias over soft soils).
Solution: Combined COSMO-SkyMed PS-InSAR data (62 images, X-band) with hourly surface soil temperature and humidity from a low-cost ground sensor. An SBL model was trained to predict LOS displacements, with an optimal coherence threshold (0.56) determined by analyzing predictive uncertainty. The model then provided temporally denser displacement predictions.
Results:
- SBL model accurately reproduced seasonal and non-linear displacement components, fitting satellite data with high fidelity.
- Stochastic oscillator showed overall soil stability (values predominantly near 0.25), with peaks during warm months indicating expected deviations.
- Enabled hourly reconstruction of ground displacement, significantly exceeding satellite revisit frequency, allowing early anomaly detection.
Calculate Your Potential ROI
See how leveraging AI for advanced monitoring can translate into significant operational efficiencies and cost savings for your enterprise.
Annual Cost Savings
$0
Employee Hours Reclaimed Annually
0
Your AI Implementation Roadmap
Our proven methodology guides your enterprise through every step of AI integration, from strategy to sustainable operation.
Discovery & Strategy
In-depth analysis of your current operations, data infrastructure, and business objectives to define clear AI use cases and strategic alignment.
Data Engineering & Integration
Preparation, cleaning, and integration of diverse data sources, including satellite imagery and IoT sensor feeds, to build a robust foundation for AI models.
Model Development & Training
Custom development and training of Sparse Bayesian Learning (SBL) models, tailored to your specific displacement monitoring needs and environmental variables.
Validation & Deployment
Rigorous testing and validation of AI models, followed by seamless integration into your existing monitoring systems for operational use.
Continuous Optimization & Support
Ongoing monitoring of model performance, recalibration with new data, and expert support to ensure long-term accuracy and effectiveness.
Ready to Transform Your Monitoring Capabilities?
Leverage the power of integrated sensing and AI to gain unprecedented insights into ground deformation. Book a free consultation with our experts today.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat