Mapping Volcano Deformation using InSAR
- Arpit Shah
- Feb 28, 2022
- 10 min read
Updated: 2 days ago
1. Introduction
I tend to recall this comic strip whenever I encounter the word volcano—
Sabu is a giant alien from Jupiter who lives on Earth with Chacha Chaudhary—the clever, quick-thinking protagonist of a widely loved Indian comic series. Together, they combine brawn and wit to rid society of its unsavoury elements. Although Sabu is formidable against thugs, he often struggles against a devious or occasionally stronger adversary. But when the lives of Chacha and his family are threatened, Sabu enters a Hulk-like rage that multiplies his strength. Pran, the creator of the series, signals this impending destruction by illustrating an erupting volcano in the background.
SECTION HYPERLINKS
Credits: RUS Copernicus & EO College
2. Setting the Context
Humans have feared and revered volcanoes for millennia. India has only one currently active volcano—and that too far away from the mainland—yet we have had a striking name for this phenomenon since ancient times: Jwalamukhi (Sanskrit), meaning the mouth that spurts fire.
Much like a human mouth, a volcano “breathes,” albeit on a geological timescale. Its size and shape expand and contract as the pressure from the underlying magma fluctuates. Surrounding rocks deform to accommodate these changes. This deformation trend is of great importance to volcanologists because it is often a leading indicator of eruption potential—the more restless the magmatic system becomes, the more frequent and intense the ground displacement tends to be, whether through uplift (swelling) or subsidence (sinking).
These movements are too subtle for the naked eye. Researchers therefore rely on geodetic techniques such as levelling, GNSS and InSAR (Interferometric Synthetic Aperture Radar) to detect and quantify deformational changes.
In this post, I demonstrate how to map ground deformation and derive displacement rates using InSAR applied to Sentinel-1 SAR imagery across three recent eruption sites:
While monitoring deformation deeper in the subsurface—within the magma chamber, located 1–10 km below the surface—would be far more ideal, current technologies can only penetrate a few metres of overlying rock.
If there is a layer of water in between, then even doing that is nigh impossible. This limitation is worrying, especially because most of Earth’s productive volcanic systems lie underwater. The violent eruption of Hunga Tonga–Hunga Haʻapai on 15 January 2022 was a stark reminder of how little warning we receive for submarine volcanic events.
3. The Processing Chain and Video Walkthrough
For this demonstration, I apply Interferometry on Sentinel-1 SAR Imagery using ESA's SNAP tool to map ground/surface deformation and derive displacement rates (uplift or subsidence).
The processing chain is extensive; here is its graphical representation:
I have prepared a detailed video walkthrough of this workflow using the 2021 eruption at Cumbre Vieja, Canary Islands. If you prefer visual learning, the video will be particularly helpful, although I recommend reading the detailed explanation first.
4. Step-by-Step Guide on InSAR Interferometry to map Deformation
Below is the detailed explanation of each step in the processing chain shown in Figure 3.
You may download the image to view it more conveniently.
(Note: In the illustrations accompanying each step, I alternate between the three volcano sites; the figure captions specify the location.)
Read & Read(2) - Two SAR images—one pre-eruption and one post-eruption—serve as input. SNAP reads them so they can later be compared for change detection.
TOPSAR-Split - SAR scenes are large, often ~1 GB, with a 250-km swath and 5×20 m resolution. Sentinel-1 acquires data in TOPSAR (Terrain Observation with Progressive Scans) mode, well suited for interferometry.
To reduce computation time, I subset the imagery to the area of interest (AOI) and retain only the required polarization (VV). The TOPSAR-Split operator handles this on SNAP.
Apply Orbit Files - A Satellite captures data as it navigates in its Orbit. Orbit files contain precise satellite position and velocity data. Updated (refined) orbit files become available only after a few weeks, so using them ensures accurate geolocation and improves the reliability of deformation measurements.
Consider the visual output of this processing step the same as Figure 5 as there is no discernible change in it after running the Apply Orbit Files operator.
Back Geocoding - the Back Geocoding operator assigns the backscatter readings captured by the satellite’s receiver to their correct geographic locations on the Earth’s surface. This conversion—i.e., the reprojection of data from radar coordinates to geographic coordinates—requires the use of a Digital Elevation Model (DEM). A DEM contains elevation values of the Earth’s bare surface; it excludes the heights of natural and built features such as vegetation and buildings.
In addition to geocoding the imagery (aided by interpolation), this operator bundles both reprojected datasets into a single Product Stack, as seen in the Product Explorer panel in Figure 6 below. This bundling process is known as Coregistration, and SNAP also provides a dedicated Coregistration operator should one wish to perform that step independently.
Coregistering both SAR datasets is essential—only after doing so can SNAP compare them pixel by pixel. This comparison underpins deformation mapping workflows, which fundamentally rely on accurate change detection.
Figure 6: Back-Geocoded Imagery Stack factoring in elevation values from a DEM dataset. This stack contains both pre (19 May 2021) and post (12 June 2021) Imagery datasets over Mt. Nyiragongo Volcano eruption site in Congo.
Enhanced Spectral Diversity - This operator refines the Back Geocoding output by improving the geometric alignment of backscatter values using an optimization algorithm. It is the final step of the Coregistration sequence. The resulting product—a Coregistered ESD Stack—contains two precisely aligned SAR images (pre- and post-eruption), ready for interferometric processing to extract deformation and displacement.
Visually, the Enhanced Spectral Diversity output appears almost identical to Figure 6, but the underlying geometric accuracy is significantly improved, which is crucial for reliable phase comparison.
Interferogram - This operator begins the core task of detecting deformation and estimating ground displacement between the pre- and post-eruption SAR acquisitions, with centimeter-level sensitivity. It generates three outputs: Intensity, Phase, and Coherence.
Intensity – In a single SAR image, the Intensity band (the i_ band in Figure 4) represents the square of the amplitude of the returned microwave signal. In the interferometric product, however, Intensity is computed by multiplying the amplitude of each corresponding pixel across the two SAR images:Intensity(pixel₁) = Amplitude₁ × Amplitude₂.This operation highlights areas where surface scattering characteristics remained similar between both acquisitions.
Phase – The Phase band of a single SAR image (the q_ band in Figure 4) encodes a modified representation of the distance between the sensor and the ground target. What we seek, however, is the difference in phase between the two acquisitions—this is the interferometric Phase, also known as the Interferogram. Phase(pixel₁) = Phase₁ − Phase₂.These phase differences capture minute surface movements—uplift or subsidence.
Coherence – Coherence measures the reliability of the interferometric phase and ranges from 0 to 1. High coherence occurs when both amplitude and phase values of a pixel remain similar across the two images; low coherence indicates surface change or noise. For deformation studies, clusters of low-coherence pixels often signal real ground change, making them valuable diagnostic features.
TOPSAR Deburst - Sentinel-1 acquires data in TOPSAR mode, transmitting microwave pulses in short bursts that illuminate the ground in discrete sections (sub-swaths, as can be seen in Figure 4). This acquisition pattern creates horizontal gaps—visible as black lines—between bursts, indicating areas with no recorded data.
The TOPSAR Deburst operator removes these discontinuities by interpolating and stitching together the adjoining burst segments. The output is a seamless, continuous interferometric image that preserves the geometric integrity required for downstream processing.
Topographic Phase Removal - Differences in interferometric phase arise from two sources:
Topographic variation (e.g., vegetation shifts, harvesting, flooding, wind-induced surface changes)
Actual ground movement (uplift or subsidence)
Since our objective is to map deformation, we must remove the component of phase caused by topography. The Topographic Phase Removal operator isolates and subtracts this contribution using a DEM, leaving behind only the displacement-related phase differences.
The resulting “flattened” interferogram is a clearer representation of true ground deformation:
Multilooking - Because radar signals strike the surface at varying angles, pixel dimensions in the interferogram are not uniform. This geometric inconsistency introduces speckle noise, producing a grainy appearance. The Multilooking operator resamples the data to create square pixels and averages neighboring values, reducing speckle and improving radiometric clarity. While this reduces spatial resolution slightly, it greatly enhances the visibility of phase fringes associated with deformation (refer Figure 11 below).
Goldstein Phase Filtering - Goldstein filtering applies a specialized frequency-domain algorithm to further suppress noise and enhance interferometric phase continuity. Though technical in implementation, its purpose is straightforward:
improve the signal-to-noise ratio,
sharpen phase fringes, and
prepare the interferogram for accurate phase unwrapping.
After this step, deformation patterns become noticeably more distinct.
Snaphu Export - Snaphu (Phase Unwrapping) - Snaphu Import - Phase to Displacement - Phase values in interferograms are “wrapped,” meaning they repeat cyclically every 2π radians. To derive meaningful displacement, these values must be converted to continuous (unwrapped) phase.
This is handled through SNAPHU (Statistical-cost, Network-flow Algorithm for Phase Unwrapping), an external module that:
unwraps the filtered interferometric phase,
resolves ambiguities in regions of rapid deformation,
returns a continuous phase field suitable for displacement calculation.
Once imported back into SNAP, the Phase-to-Displacement operator converts unwrapped phase into physical ground movement (uplift or subsidence), typically expressed in meters along the satellite’s line of sight.
For example, a minimum value of –0.1 m would indicate approximately 10 cm of subsidence at the volcano cone—consistent with post-eruption collapse.
Range-Doppler Terrain Correction - Although earlier steps converted radar geometry to geographic coordinates, the dataset still carries distortions caused by satellite viewing geometry, Earth’s curvature, and terrain variations.
The Range-Doppler Terrain Correction operator:
corrects geometric distortions caused by slant-range viewing from satellite,
aligns pixels to their true ground positions,
orthorectifies the interferogram using a DEM, and
removes water bodies when desired (since they often produce unreliable displacement values).
The resulting terrain-corrected product is geospatially accurate and ready for visualization or export. Earlier, Back-Geocoding had already converted the SAR data from radar coordinates into geographic coordinates on a 2D map. However, Range-Doppler Terrain Correction goes a step further: it adjusts the geometry of each pixel using satellite viewing parameters and a DEM so that the image is correctly orthorectified—i.e., it represents the Earth’s curved surface accurately within a 2D map projection.
Write (& Export) - This operator enables me to save the processed output to my system.
Visualizing the Interferometric Phase and Displacement Output on Google Earth
Interferometric Phase Visualization on Google Earth
Why are the Phase fringes located so far from the Mt. Nyiragongo's cone?
These fringes correspond to persistent seismic activity (earthquakes) that followed the eruption. Deformation mapping through Remote Sensing allows us to detect not just volcanic unrest but also related phenomena such as regional earthquakes, landslides, or any process that alters surface geometry at scale.
Displacement Visualization on Google Earth
Thank you for reading this post, and I hope you found it as fascinating as a blazing volcano! 🤯😁Feel free to share your feedback. If you’d like to see the complete workflow in action, do watch the SNAP-based video walkthrough of the InSAR processing chain.
Let me leave you with a spooky possibility—just in case you need a little less volcano in your life!
ABOUT US - OPERATIONS MAPPING SOLUTIONS FOR ORGANIZATIONS
Intelloc Mapping Services, Kolkata | Mapmyops.com offers a suite of Mapping and Analytics solutions that seamlessly integrate with Operations Planning, Design, and Audit workflows. Our capabilities include — but are not limited to — Drone Services, Location Analytics & GIS Applications, Satellite Imagery Analytics, Supply Chain Network Design, Subsurface Mapping and Wastewater Treatment. Projects are executed pan-India, delivering actionable insights and operational efficiency across sectors.
My firm's services can be split into two categories - Geographic Mapping and Operations Mapping. Our range of offerings are listed in the infographic below-
A majority of our Mapping for Operations-themed workflows (50+) can be accessed from this website's landing page. We respond well to documented queries/requirements. Demonstrations/PoC can be facilitated, on a paid-basis. Looking forward to being of service.
Regards,
