Sensors
- Sentinel-1 A/B (C-band SAR, ESA): 5.6 cm wavelength; Interferometric Wide Swath mode gives 250 km coverage at 5 x 20 m ground resolution; 6-day repeat at mid-latitudes with both satellites active, 12-day with one. Phase sensitivity to line-of-sight displacement is theoretically sub-millimetre, though atmospheric noise typically limits practical detection to 2-5 mm per epoch. Free archive from 2014.
- TerraSAR-X / TanDEM-X (X-band SAR, DLR): 3.1 cm wavelength; Stripmap mode at 3 m resolution, Spotlight at 1 m. Shorter wavelength improves sensitivity to small uplifts but increases atmospheric phase noise. Repeat cycle is 11 days; tasking on demand allows shorter stacks over priority areas. Commercial data; archive from 2007.
- Envisat ASAR (C-band SAR, ESA, archived): Same 5.6 cm wavelength as Sentinel-1; Image Mode at 30 m resolution. Operated 2002-2012, giving a 10-year archive that predates Sentinel-1 and is essential for establishing pre-injection baselines at sites such as In Salah and Sleipner. No new acquisitions.
- GNSS campaign benchmarks: Continuous or episodic GNSS at surface monuments provides absolute vertical displacement at discrete points, typically to 1-3 mm precision in the vertical. Used to calibrate InSAR line-of-sight projections and resolve the vertical/horizontal ambiguity inherent in single-geometry SAR.
What a poroelastic medium tells a radar
When supercritical CO2 is injected into a porous formation at depth, it displaces brine and raises pore pressure. The surrounding rock matrix responds elastically: it expands. That expansion propagates upward through the overburden and produces a broad, low-amplitude dome of surface uplift centred roughly above the injection well. The shape and amplitude of the dome encode the depth, lateral extent and stiffness of the reservoir, and, crucially, the injected volume.
InSAR measures the change in round-trip travel time between a satellite and the ground surface. A surface rising by one millimetre shortens the path by one millimetre, advancing the phase of the returned signal by a fraction of the radar wavelength. Stack enough interferograms and apply a time-series algorithm such as Small BAseline Subset (SBAS) or Persistent Scatterer InSAR (PS-InSAR), and the cumulative uplift signal emerges from the atmospheric and orbital noise. At Sentinel-1 C-band, the practical detection floor for cumulative displacement over a multi-year stack is generally quoted in the literature at 2-5 mm, though individual epoch noise can be 10-20 mm before temporal averaging.
The reference sites: what the published record shows
Three sites dominate the published InSAR literature on CO2 storage and are worth naming precisely because they set realistic expectations. Sleipner, in the Norwegian North Sea, has been injecting CO2 into the Utsira Sand since 1996. Because it is offshore, surface InSAR is not applicable; the monitoring there relies on seismic time-lapse and gravity. It is the geophysical benchmark, not the InSAR benchmark.
In Salah, Algeria, is the canonical onshore InSAR case. Statoil, BP and Sonatrach injected CO2 into the Krechba formation from 2004 to 2011. Envisat ASAR and Radarsat-2 interferograms published by Vasco and colleagues revealed a surface uplift of roughly 5-20 mm over the injection zone, with a distinctive double-lobe pattern that indicated fracture-zone reactivation above the reservoir. That signal prompted operational changes and remains the clearest demonstration that InSAR can detect not just bulk uplift but structural heterogeneity in the caprock.
Quest, operated by Shell in Alberta, began injection in 2015. Published studies using Radarsat-2 data identified uplift signals of a few millimetres per year consistent with poroelastic models, though the low topographic relief and agricultural land cover there provide reasonable coherence for C-band InSAR. These three sites together define what is achievable and what is not: offshore storage cannot be monitored by surface InSAR; arid, sparsely vegetated terrain gives the best coherence; and even ideal conditions may not resolve injection volumes to better than order-of-magnitude accuracy without independent GNSS and seismic constraints.
Regulatory context: what the EU CCS Directive actually requires
Directive 2009/31/EC on the geological storage of CO2 requires operators to submit a monitoring plan covering the storage complex, the surrounding area and, where relevant, the seabed and water column. Article 13 specifies that monitoring must be capable of detecting significant irregularities, including unexpected migration of CO2 or formation water. Surface deformation monitoring is not explicitly mandated by name, but guidance documents from the European Commission's Technical Working Group on Monitoring and Reporting treat InSAR as an accepted method for detecting surface uplift as a proxy for pressure build-up.
The Directive also requires a corrective measures plan triggered if monitoring indicates leakage or significant irregularities. An InSAR time-series that shows anomalous uplift acceleration, spatial pattern change or displacement outside the predicted model envelope constitutes exactly the kind of early indicator that triggers that plan. Operators therefore need not just a snapshot but a maintained, regularly updated deformation record with defined thresholds. That is an operational monitoring programme, not a one-off academic study.
Honest limits of the method
InSAR over CO2 storage sites faces several constraints that buyers should understand before commissioning a programme. Vegetation kills coherence. C-band SAR loses phase coherence rapidly over agricultural land or forest, particularly in humid climates, making time-series analysis unreliable without a dense network of artificial corner reflectors or a switch to L-band sensors such as ALOS-2 (24 cm wavelength, far better coherence in vegetated terrain, though not listed as a primary sensor here because it is commercially restricted and has a 14-day repeat).
Atmospheric phase delay is the dominant noise source. Tropospheric water vapour variations can introduce apparent range changes of 10-30 mm in a single interferogram, easily masking the 2-10 mm uplift signal expected over a storage site in its early injection years. Mitigation requires either external weather model corrections (ERA5, GACOS) or a sufficiently long time-series to separate the temporally correlated deformation signal from the spatially correlated atmospheric noise. Neither approach eliminates the problem entirely.
The forward model ambiguity is real. A given surface uplift pattern is consistent with a range of reservoir geometries and mechanical properties. InSAR constrains the surface boundary condition; it cannot independently resolve depth, lateral extent and stiffness without coupling to a calibrated geomechanical model built from well logs and seismic data. Treat InSAR as one node in a multi-method monitoring network, not as a standalone mass-balance meter.
From interferogram to compliance record
A practical monitoring programme for a permitted storage site runs in three layers. First, a baseline stack is built from archive data covering the pre-injection period, ideally several years of Envisat ASAR or early Sentinel-1 acquisitions. This establishes the natural deformation background: seasonal soil moisture cycles, any pre-existing subsidence from groundwater or hydrocarbon extraction nearby, and the long-term tectonic rate. Without this baseline, any post-injection signal is uninterpretable.
Second, an operational acquisition plan is agreed with the SAR data provider. For Sentinel-1, the standard 6- or 12-day repeat is usually sufficient for a slowly evolving poroelastic signal, but the ascending and descending geometries should both be used to decompose the line-of-sight displacement into vertical and east-west components. Third, the processed time-series, expressed as millimetres of cumulative displacement per pixel, is compared against the predicted uplift from the approved geomechanical model. Deviations beyond a defined threshold trigger a review. Satellize structures this kind of monitoring programme as a maintained analytic service rather than a one-off delivery, drawing on the same open-constellation infrastructure used in its Tonga crop-estimation work.
The deliverable a regulator actually needs is not a stack of GeoTIFFs. It is a dated, version-controlled report stating: observed displacement at the injection centroid, comparison against the approved model, any spatial anomalies, and a clear statement of whether the monitoring thresholds have been breached. That document has a legal function. The data pipeline behind it is the engineering problem.
Typical figures
| Primary sensor frequency | C-band (5.6 cm wavelength, Sentinel-1); X-band (3.1 cm, TerraSAR-X) for higher spatial resolution |
| Spatial resolution (InSAR product) | 20-100 m per pixel after multi-looking; point-target PS precision at native resolution (~5 x 20 m IW mode) |
| Minimum detectable cumulative displacement | 2-5 mm over multi-year SBAS or PS stack (atmospheric noise limited); single-epoch noise 10-30 mm before averaging |
| Repeat cycle | 6 days (Sentinel-1 A+B); 12 days (single satellite); 11 days (TerraSAR-X) |
| Processing latency (operational) | 2-5 days from SAR acquisition to interferogram; 2-4 weeks for full time-series update with atmospheric correction |
| Archive depth | Envisat ASAR from 2002; Sentinel-1 from 2014; TerraSAR-X from 2007 |
| Coverage per pass | 250 km swath (Sentinel-1 IW); 30 km swath (TerraSAR-X Stripmap) |
| Coherence requirement | Best in arid or semi-arid terrain; poor over dense vegetation without corner reflectors or L-band sensor |
| Displacement components resolvable | Vertical + east-west with ascending and descending geometries; north-south poorly constrained by SAR alone |
| Delivery formats | GeoTIFF displacement maps, CSV time-series per point, PDF compliance report, GIS vector layers |
Analytics Satellize can run
| Pre-injection baseline deformation map | SBAS time-series InSAR on Envisat ASAR and early Sentinel-1 archive | GeoTIFF stack and PDF summary of natural deformation rates before first injection, versioned for regulatory submission |
| Operational uplift time-series | PS-InSAR or SBAS on rolling Sentinel-1 acquisitions with ERA5 or GACOS atmospheric correction | Monthly updated displacement maps and CSV point-series at injection centroid and regulatory monitoring benchmarks |
| Anomaly detection against approved geomechanical model | Residual comparison between observed InSAR displacement and Mogi or nucleus-of-strain forward model prediction | Threshold-breach alert with spatial anomaly map, issued within 5 days of acquisition |
| Displacement component decomposition | Ascending/descending geometry combination to separate vertical and east-west line-of-sight components | Vertical and horizontal displacement GeoTIFFs per epoch, with uncertainty estimates |
| GNSS-InSAR integration | Least-squares fusion of campaign GNSS vertical measurements with InSAR relative field to anchor absolute reference frame | Calibrated absolute displacement map suitable for volume-change inversion |
| Regulatory compliance report | Structured comparison of observed deformation against permit thresholds, with atmospheric correction audit trail | Dated, version-controlled PDF report formatted for submission under EU CCS Directive Article 13 monitoring obligations |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.