Introduction to Satellite-based Radar Interferometry for Land Deformation Monitoring

Satellite-based radar interferometry, commonly known as InSAR (Interferometric Synthetic Aperture Radar), is a sophisticated remote sensing technique that measures ground deformation with millimeter-scale precision across vast areas. By comparing two or more radar images acquired from satellites at different times, InSAR can detect subtle changes in the Earth's surface caused by natural processes such as tectonic activity, volcanic inflation, landslides, and subsidence due to groundwater extraction or mining. This technology has become an indispensable tool for geoscientists, civil engineers, and disaster management agencies worldwide. In this article, we provide a detailed guide on how to use InSAR for large-scale land deformation monitoring, covering data acquisition, processing workflows, interpretation techniques, common challenges, and real-world applications. We also link to authoritative resources for further reading.

Fundamentals of Radar Interferometry

How InSAR Works

InSAR relies on the phase difference between two radar acquisitions over the same area. Synthetic Aperture Radar (SAR) sensors emit microwave pulses and record the backscattered signal. When two SAR images are acquired from slightly different positions or at different times, the phase difference (interferogram) reveals the relative displacement of the ground in the line-of-sight direction. This principle is analogous to shifting interference fringes in optics: each full 2π cycle of phase change corresponds to a displacement equal to half the radar wavelength (e.g., about 2.8 cm for Sentinel-1 C-band). By unwrapping the phase, we convert the cyclic fringes into continuous displacement maps.

Key Satellite Missions for InSAR

  • Sentinel-1 (ESA): C-band (5.6 cm wavelength), provides global coverage every 6–12 days with free and open data. Widely used for operational deformation monitoring.
  • ALOS-2 (JAXA): L-band (23.6 cm wavelength), better penetration through vegetation and less decorrelation over time, ideal for forested or alpine regions.
  • Radarsat-2 and Radarsat Constellation (CSA): C-band, high-resolution options for detailed urban studies.
  • ERS/Envisat (ESA, historical): C-band data dating back to the 1990s, valuable for long-term time series analysis.
  • NISAR (NASA-ISRO, planned launch 2024): L-band and S-band, will provide unprecedented global coverage with 12-day revisit.

Each mission has trade-offs between wavelength, resolution, revisit time, and availability. For deformation monitoring over large areas, Sentinel-1 is often the starting point due to its open data policy and consistent acquisitions.

Step-by-Step Workflow for InSAR Deformation Monitoring

1. Data Acquisition and Planning

The first step is to identify the area of interest and acquire appropriate SAR imagery. Key considerations include:

  • Number of images: Minimum two for a single interferogram, but a time series of many images (20+) is recommended for robust analysis using advanced techniques like SBAS or PS-InSAR.
  • Perpendicular baseline: The distance between satellite orbits at the time of acquisition. Larger baselines introduce topographic fringes but also increase decorrelation. Baseline thresholds depend on wavelength and terrain.
  • Temporal baseline: Time between acquisitions. Shorter baselines reduce decorrelation but may not capture slow deformation.
  • Orbit files: Precise orbits (e.g., from ESA’s Sentinel-1 Precision Orbit Determination service) are required to remove orbital ramp errors.

Major data repositories include the Copernicus Open Access Hub for Sentinel-1, Alaska Satellite Facility for ALOS-2 and other missions, and the NASA Earthdata Search for historical datasets.

2. Preprocessing of SAR Images

Before interferometric processing, raw SAR data must be corrected for various distortions:

  • Radiometric calibration: Convert digital numbers to radar backscatter coefficient (sigma0 or gamma0) for consistent amplitude analysis.
  • Coregistration: Align two SAR images to sub-pixel accuracy to ensure phase coherence. This step uses geometric and cross-correlation methods.
  • Topographic phase removal: Use a digital elevation model (DEM) such as SRTM or TanDEM-X to subtract the phase contribution from topography. Residual topographic errors can be estimated during time series analysis.
  • Spectral filtering: Apply bandpass filters (e.g., Goldstein filter) to reduce noise while preserving phase features. Adaptive filters often work best over heterogeneous terrains.

3. Interferogram Generation

After preprocessing, the two images are combined by complex conjugate multiplication to produce an interferogram. The interferogram contains both phase and coherence information:

  • Phase: The wrapped phase (modulo 2π) contains the deformation signal plus residual errors from atmosphere, orbital inaccuracies, and DEM errors.
  • Coherence: A value between 0 and 1 indicates the quality of the interferometric phase. Coherence below 0.2 often leads to decorrelation and unreliable measurements. Areas of water, dense vegetation, or rapid surface change typically have low coherence.

Multiple interferograms can be generated for different time intervals, forming a network that later feeds into time series inversion.

4. Phase Unwrapping

Phase unwrapping converts the wrapped phase (values between -π and π) into a continuous displacement field. This is a critical step because errors propagate into the final deformation map. Common algorithms include:

  • Branch-cut (Goldstein’s algorithm): Identifies and cuts around inconsistent phase residues; robust for areas with high noise but difficult to automate over large regions.
  • SNAPHU (Statistical-cost, Network-flow Algorithm for Phase Unwrapping): Uses maximum a posteriori estimation with a statistical cost function; widely used and freely available.
  • Minimum cost flow (MCF): Treats unwrapping as a network optimization problem; suitable for large-scale processing.

The choice of unwrapping algorithm depends on the dataset quality and expected deformation patterns. Manual inspection and iterative correction are often necessary.

5. Geocoding and Displacement Mapping

After unwrapping, the displacement values are in radar slant range coordinates. Geocoding transforms them into a map-projected coordinate system (e.g., UTM, WGS84) using the satellite orbit geometry and a DEM. The output is a deformation map that can be overlaid on GIS layers such as infrastructure, geology, or land use. Typical map products include vertical displacement (assuming no horizontal motion) or line-of-sight (LOS) displacement maps. For many applications, vertical subsidence maps are derived by assuming negligible east-west motion, but if both ascending and descending orbits are available, 2D decomposition (vertical and east-west) can be achieved.

6. Time Series Analysis

For long-term monitoring, a single interferogram is insufficient. Advanced InSAR techniques like Persistent Scatterer InSAR (PS-InSAR) or Small Baseline Subset (SBAS) use many images to estimate deformation time series with reduced atmospheric noise:

  • PS-InSAR: Identifies coherent pixels (e.g., buildings, rocks, metal structures) that maintain stable phase over time. This method is excellent for urban areas and infrastructure monitoring. The Stanford Method for Persistent Scatterers (StaMPS) is a widely used open-source implementation.
  • SBAS: Uses interferograms with small spatial and temporal baselines to maximize coherence in vegetated or rural areas. The SBAS algorithm solves for the deformation history via least squares inversion.

Both methods produce displacement velocity maps and time series plots, enabling detection of nonlinear deformation patterns such as acceleration or seasonal cycles.

Practical Considerations and Quality Control

Atmospheric Correction

One of the primary sources of error in InSAR is the temporal variation of atmospheric water vapor and pressure, which delays radar signals. Correcting for these effects is essential for reliable deformation measurements. Methods include:

  • External weather models: Use data from ERA5, MERRA-2, or GPS-derived zenith total delay to model atmospheric phase screens. Tools like TRAIN (Toolbox for Reducing Atmospheric InSAR Noise) automate this step.
  • Stacking or filtering: If the deformation signal is steady, stacking multiple interferograms reduces atmospheric noise. For time series, temporal filtering (e.g., high-pass) can isolate deformation from stochastic atmosphere.
  • Numerical weather prediction (NWP) integration: Some processing chains (e.g., GMTSAR, ISCE) include direct ingestion of weather model data.

Orbital and DEM Errors

Orbital inaccuracies introduce long-wavelength phase ramps across the scene. Using precise orbit files and applying a deramping step (e.g., by fitting a quadratic surface) minimizes these errors. Similarly, DEM errors cause residual phase proportional to the perpendicular baseline. Time series methods like PS-InSAR compensate by estimating DEM errors along with displacement rates.

Decorrelation and Coherence Loss

Decorrelation occurs when the scattering characteristics of the surface change between acquisitions. Common causes include vegetation growth, snow cover, soil moisture changes, or human activity. To mitigate decorrelation:

  • Use longer wavelength (L-band vs. C-band) for vegetated areas.
  • Minimize temporal baselines by choosing image pairs with short time intervals.
  • Apply multi-looking (averaging neighboring pixels) to improve coherence at the cost of spatial resolution.
  • Use advanced coherence estimation filters (e.g., adaptive windowing).

Applications of InSAR in Large-Scale Land Deformation Monitoring

Natural Hazard Assessment

InSAR has been instrumental in mapping surface deformation associated with earthquakes, volcanoes, and landslides. For example, the 2019 Ridgecrest earthquake sequence in California was imaged by Sentinel-1, revealing complex fault slip distributions (reference). Volcano monitoring, such as at Kīlauea, Hawaii, uses InSAR to track magma chamber inflation and flank motion. Landslide-prone areas, like the Italian Alps, are routinely monitored with interferometry to identify precursory movements before catastrophic failure.

Urban Subsidence and Infrastructure

Many cities worldwide experience land subsidence due to groundwater extraction, underground construction, or soil compaction. InSAR provides high-density measurements over urban areas where buildings act as persistent scatterers. Examples include monitoring subsidence in Jakarta, Indonesia (up to 20 cm/year), and Mexico City, where InSAR data guide mitigation strategies. The technique is also applied to monitor stability of critical infrastructure like dams, bridges, and pipelines (USGS resources).

Environmental and Geological Processes

Beyond hazards, InSAR tracks glacier flow velocities, permafrost thaw, and groundwater storage changes. Large-scale projects like the ESA's Sentinel-1 mission have enabled continent-wide deformation mapping, such as the detection of slow-moving landslides across the European Alps or regional subsidence in the Central Valley of California due to agricultural water use.

Challenges and Limitations

While InSAR offers transformative capabilities, users must be aware of its limitations:

  • Line-of-sight ambiguity: InSAR measures displacement only along the radar look direction, which is near-vertical in most satellite configurations. East-west or vertical components must be separated using combined ascending and descending datasets, or additional assumptions.
  • Temporal and spatial decorrelation: Not all surfaces remain coherent over time. Vegetated, wet, or frequently disturbed areas may yield no usable data. L-band satellites help but are less available.
  • Atmospheric delays: Even after correction, residual tropospheric and ionospheric noise can obscure small deformation signals. In regions with strong topography or weather fronts, this is a persistent challenge.
  • Data volume and processing complexity: Processing hundreds of SAR images requires substantial computational resources and expertise. Cloud-based platforms like Google Earth Engine, Alaska Satellite Facility’s Hyp3, and ESA’s Geohazards Exploitation Platform (GEP) are lowering the barrier, but careful validation remains essential.
  • Validation and ground truth: InSAR results should be validated with in-situ measurements (e.g., GPS, tiltmeters, leveling surveys) to ensure accuracy. Misinterpretation can lead to false alarms or missed hazards.

Future Directions

Advancements in InSAR technology continue to expand its applicability. The upcoming NISAR mission promises global L-band observations every 12 days, greatly improving monitoring in vegetated regions. New algorithms leveraging machine learning for phase unwrapping and atmospheric correction are under development. Integration with other remote sensing techniques, such as optical imagery and LiDAR, provides complementary information for comprehensive deformation analysis. Furthermore, open-source software ecosystems (e.g., ISCE, SNAP, MintPy) have democratized access to advanced InSAR processing, enabling more researchers and practitioners to leverage this powerful tool.

Conclusion

Satellite-based radar interferometry is a mature yet evolving technology that enables cost-effective, synoptic, and precise monitoring of land deformation over scales from a few meters to entire continents. By following a systematic workflow that includes careful data selection, robust preprocessing, appropriate phase unwrapping, and time series analysis, scientists and engineers can detect subtle ground movements that reveal underlying geological and anthropogenic processes. Despite challenges such as atmospheric noise and decorrelation, ongoing satellite missions and processing innovations continue to improve reliability. InSAR is now a standard tool for hazard assessment, infrastructure management, and environmental monitoring, making it an essential component of geospatial analysis for a resilient future.