Find your next learning step
An interferogram measures the atmosphere as well as the ground. Smooth colour bands can therefore look like deformation even when the surface did not move.
Open the chapter contents · Remote Sensing root · Gold Standard of Education · How X Works
Chapter contents
Choose the route closest to the question, or read from Chapter 1 for the complete argument. The chapters remain open on the page; these grouped routes are an index.
Route 1 · Build the phase budget
Route 2 · Recognise atmospheric patterns
Route 3 · Correct without erasing signal
Route 4 · Validate the interpretation
Route 5 · Check and continue
CHAPTER 1 OF 12 · BUILD THE PHASE BUDGET
1. An interferogram contains sky as well as ground
Interferometric Synthetic Aperture Radar compares phase from two acquisitions. After removing reference geometry and topography as well as possible, the residual can still contain surface displacement, orbital error, ionospheric delay, tropospheric delay, noise and unmodelled topography. An atmospheric phase screen is the spatial phase pattern attributed to propagation through the atmosphere at one acquisition epoch.
The atmosphere is not harmless because radar phase is extremely sensitive to path length. A smooth pressure or water-vapour field can create a broad ramp; turbulent water vapour can create patchy structure. Either can resemble geophysical deformation.
The correct question is not whether a map looks smooth. It is whether the phase budget and independent evidence support the deformation interpretation.
CHAPTER 2 OF 12 · BUILD THE PHASE BUDGET
2. Delay becomes phase through radar wavelength
For line-of-sight path change Δr, the usual displacement-phase magnitude is Δφ=4πΔr/λ, with sign depending on the product convention. A delay difference of only a fraction of the wavelength can therefore produce several radians of phase.
Tropospheric delay has a dry component related mainly to pressure and a wet component driven by water vapour. The neutral atmosphere is non-dispersive to a good approximation at common radar frequencies. Ionospheric effects are dispersive and become especially important at longer wavelengths such as L-band.
An interferogram records the difference between two acquisition atmospheres, not a permanent property of the scene.
CHAPTER 3 OF 12 · RECOGNISE ATMOSPHERIC PATTERNS
3. Work a false two-centimetre ramp
Take a C-band wavelength λ=5.6 cm. Suppose atmospheric path delay differs by 2.0 cm from one side of the scene to the other. The phase difference is 4π(0.020/0.056)=4.49 radians, or about 0.714 cycle.
If every residual phase value is converted directly to line-of-sight displacement, that atmospheric ramp becomes an apparent 2.0 cm ground-motion gradient even though the surface did not move.
| Quantity | Calculation | Result |
|---|---|---|
| Radar wavelength | Given | 5.6 cm |
| Differential atmospheric delay | Given | 2.0 cm |
| Phase ramp | 4π×0.020/0.056 | 4.49 rad |
| Cycle fraction | 4.49/(2π) | 0.714 cycle |
| False line-of-sight displacement | λΔφ/(4π) | 2.0 cm |
The sign must be checked against the processor’s phase convention before physical interpretation.
CHAPTER 4 OF 12 · RECOGNISE ATMOSPHERIC PATTERNS
4. Stratified delay can correlate with elevation
Pressure, temperature and water vapour often vary with height. In mountainous terrain, residual phase may correlate with elevation because the radar path traverses different atmospheric columns. A simple phase-versus-height regression can reveal or reduce a stratified component.
Yet correlation is not causation. Landslides, glacier motion, aquifer compaction and volcanic deformation can also vary systematically with elevation. Removing the fitted height term may erase real signal if deformation and topography are confounded.
Use weather fields, multiple interferograms, stable reference areas and physical expectations before treating elevation-correlated phase as purely atmospheric.
CHAPTER 5 OF 12 · RECOGNISE ATMOSPHERIC PATTERNS
5. Turbulent delay has spatial memory
Water vapour turbulence produces correlated patches rather than independent pixel noise. Nearby pixels can share similar delay, and the correlation length depends on weather, acquisition geometry and atmospheric structure. Spatial smoothing can therefore make the artefact look even more geophysical.
A regional mean does not gain the usual independent-pixel precision when all pixels share a coherent atmospheric screen. Uncertainty must account for spatial covariance or use independent acquisition times.
| Pattern | Plausible atmospheric mechanism | Evidence check |
|---|---|---|
| Broad scene-wide ramp | Orbit or long-wavelength atmosphere | Compare orbit correction and neighbouring frames |
| Phase correlated with height | Stratified troposphere | Test weather fields and stable high/low terrain |
| Patchy kilometre-scale structure | Turbulent water vapour | Compare independent dates and meteorology |
| Range-direction dispersive streaking | Ionosphere | Use split-spectrum or frequency-sensitive tests |
CHAPTER 6 OF 12 · CORRECT WITHOUT ERASING SIGNAL
6. Changed context: the same ramp can be signal or atmosphere
Suppose a broad phase gradient repeats with similar spatial shape across several interferograms that share one acquisition date. That common-date structure points to an atmospheric screen associated with that epoch. If instead the gradient accumulates consistently through time and agrees with GNSS motion, deformation becomes more plausible.
Now place the scene on a volcanic flank where true displacement also follows elevation. A one-interferogram height regression cannot separate the two. The changed context requires temporal redundancy, weather information or an independent deformation model.
Transfer the reasoning by asking which dimension—time, space, frequency or external measurement—breaks the ambiguity.
CHAPTER 7 OF 12 · CORRECT WITHOUT ERASING SIGNAL
7. Time series help because atmosphere and deformation differ
Persistent Scatterer and Small Baseline approaches use networks of acquisitions rather than one pair. Slowly accumulating deformation has temporal structure, while atmospheric delay often varies more rapidly from date to date. Estimating a screen for each epoch can reduce atmospheric contamination.
The separation is imperfect. Seasonal deformation can correlate with seasonal water vapour. Long-lived atmospheric anomalies and sparse acquisitions can leak into velocity. Filtering choices impose assumptions about temporal and spatial scales.
Report the acquisition network, temporal filter, spatial filter, reference region and uncertainty. A smooth velocity map can inherit a smooth bias from the processing model.
CHAPTER 8 OF 12 · CORRECT WITHOUT ERASING SIGNAL
8. Corrections need an independent information source
Numerical weather models, GNSS zenith delays, spectrometer water-vapour products and local meteorological measurements can estimate tropospheric delay. Elevation-based models use the InSAR data themselves and are vulnerable when signal correlates with terrain. Ionospheric split-spectrum methods exploit frequency dependence.
Map zenith delay into radar line of sight using the incidence geometry and a defensible mapping function. Match correction time and location to the satellite acquisition; a daily average weather field may miss a short-lived gradient.
Validate corrections on held-out interferograms or stable areas. A correction that merely reduces spatial variance can also suppress real deformation.
CHAPTER 9 OF 12 · VALIDATE THE INTERPRETATION
9. Referencing can move the artefact rather than remove it
Setting a reference pixel to zero removes one additive constant. It does not remove a spatially varying atmospheric screen. If the reference lies inside a local vapour anomaly, the entire map inherits that anomaly with opposite sign.
Disconnected unwrapping components may carry separate phase offsets, while atmospheric ramps act within and across components. Coherence indicates phase stability but cannot declare a stable atmosphere. Coregistration error and topographic residuals remain separate terms.
Choose a reference region with independent stability evidence, inspect several pixels and state the component used. Repeat the analysis with plausible alternative references.
CHAPTER 10 OF 12 · VALIDATE THE INTERPRETATION
10. Practical evidence checks
Inspect wrapped phase, unwrapped phase, coherence, connected components, topography and amplitude together. Compare ascending and descending tracks while respecting their different line-of-sight sensitivities. Look for common acquisition dates in a temporal network and closure inconsistencies around interferogram triangles.
Compare with GNSS, levelling or physical process models. Examine residual phase against elevation and weather variables without assuming the regression term is entirely atmospheric. Estimate spatial covariance from stable areas and carry it into regional uncertainty.
Preserve before-and-after correction products. If the correction changes the claimed displacement materially, that sensitivity belongs in the conclusion.
CHAPTER 11 OF 12 · CHECK AND CONTINUE
11. Questions readers should ask
Does high coherence eliminate atmospheric error?
No. A coherent radar signal can still contain a smooth propagation delay.
Is every phase–elevation correlation atmospheric?
No. Real deformation can correlate with terrain. Independent temporal or geophysical evidence is needed.
Can one stable reference pixel remove the screen?
It removes one constant, not spatial variation.
Why do more acquisitions help?
They provide temporal redundancy that can separate persistent motion from date-specific atmospheric structure, though the separation remains model-dependent.
CHAPTER 12 OF 12 · CHECK AND CONTINUE
12. Continue through the verified remote-sensing ecosystem
Begin with the Remote Sensing root, the Gold Standard of Education and the How X Works hub. Read Interferometric Coherence, Decorrelation and Phase Reliability before treating a phase pixel as stable, then use InSAR Phase-Unwrapping Residues and Disconnected Components to audit cycle errors and component offsets.
For other false spatial changes, continue to Co-Registration Offsets and False Change at a Boundary and Incidence Angle, SAR Backscatter and False Change Across Orbits. A Shared Baseline and the Uncertainty in Comparing Two Changes helps preserve dependence in change comparisons.
Authoritative technical reading: the European Space Agency’s InSAR Principles processing and interpretation guide develops the full phase budget and atmospheric limitations. The open review Atmospheric Effects on InSAR Measurements and Their Mitigation compares tropospheric and ionospheric effects and correction strategies. NASA Earthdata’s unwrapped-interferogram workflow shows where phase becomes a deformation product.
Top · Contents · Previous · How X Works