Misconceptions about interpreting InSAR data

 Misconceptions about interpreting InSAR data

In this article guest author Reda Meftahi of Survey Intelligence writes about the fundamental misconceptions associated with the interpretation of InSAR data.Reda would strongly encourage the reader to question every provider on the uncertainty removal applied to convert “InSAR” time series into estimated deformation.

What is InSAR?

Interferometric Synthetic Aperture Radar (InSAR) is an effective way to measure changes in land surface altitude. InSAR makes high-density measurements over large areas by using radar signals from Earth-orbiting satellites to measure changes in land-surface altitude at high degrees of measurement resolution and spatial detail. (source)

Synthetic Aperture Radar (SAR) imagery is produced by reflecting radar signals off a target area and measuring the two-way travel time back to the satellite. The SAR interferometry technique uses two SAR images of the same area acquired at different times and “interferes” (differences) them, resulting in maps called interferograms that show ground-surface displacement (range change) between the two time periods.

Advantages of InSAR

InSAR is ideally suited to measure the spatial extent and magnitude of surface deformation associated with fluid extraction and natural hazards (earthquakes, volcanoes, landslides). It is often less expensive than obtaining sparse point measurements from labor-intensive spirit-leveling and global positioning system (GPS) surveys, and can provide millions of data points in a region about 10,000 square kilometers large. By identifying specific areas of deformation within broader regions of interest, InSAR imagery can also be used to better position specialized instrumentation (such as extensometers, GPS networks, and leveling lines) designed to precisely measure and monitor surface deformation over limited areas.

1 – InSAR is a displacement time series – It is not!

Synthetic Aperture Radar or SAR is an active mono-frequency microwave signal emitted by a satellite in orbit around the earth. That signal bounces on the surface of the earth and travels back to the satellite antenna with information about the reflective surface it was reflected from. There are two main attributes in the signal that defines the properties of the reflective surface it encountered: its amplitude and its phase characteristics.

Interferometric Synthetic Aperture Radar or InSAR is the reflected signal phase difference between a 2nd pass and a previous pass of the satellite recorded over approximately the same area. Theoretically any recorded difference in the phase of the signal would be due to a displacement of the targeted surface. If true, that difference could be later converted into a displacement estimation.

It is assumed that only the ground displacement is capable to affect the values of the phase of the reflected signal.

This is simple and effective until we ask a very important question: are there other factors (apart from a ground displacement) capable of affecting the values of the phase of the reflected signal? The answer is yes, and there are multiple ones. This has an important implication, it means that every time we record a difference in the readings of the satellite signal, the chances that they are due only to a ground surface displacement are not superior to 50% independently of the satellite native band resolution.

While it is true, that a surface displacement will cause a phase difference in the radar signal, a phase difference in the recorded signal is not always due to a displacement.

This is very important to understand for it is the foundation of a proper interpretation of InSAR time series. InSAR “displacement” time series are time series of phase change from which a probability of a ground displacement should be extracted. InSAR instead should be considered as just raw phase change data

Therefore, a simple polynomial approach whether linear, quadratic and/or periodic interpolation into a displacement will not resolve the uncertainty of the contained information. This explains why the accuracy of the InSAR predictions is often not accurate when benchmarked with in situ measurements.

2 – Vertical displacement is the dominant information contained in InSAR time series – It is not!

A point that is often underestimated with InSAR data is the conversion of the one-dimensional Line Of Site (LOS) displacement into a three-dimensional displacement.

A clue of the InSAR community’s struggle with the interpretation of InSAR data is the difference in resolution between the native InSAR (LOS) time series and the decomposed vertical and horizontal displacement time series (figure. 2). If we take the example of the European Ground Motion Service (EGMS) with a native LOS resolution of 100 m2, EGMS couldn’t generate a 100 m2 decomposed data: the best resolution of the final decomposed data is 10000 m2.

To transform the one-dimensional LOS theoretical displacement into a three-dimensional true displacement one needs at least two different angles of measurement for the same location obtained from the ascending orbit and descending orbit of a satellite. One can imagine a mass attached by two springs, one stretching to the right and the other to the left. No movement means the system is at the equilibrium and the length of each spring does not change. But if you move the masse to the right, you compress the right spring and reduce its length while you stretch the left spring and increase its length. In reverse by correctly estimating the length of each spring you can estimate the true position of the mass. To resemble reality all displacements are allowed: linear, periodic, both alternatively, and simultaneously.

It is easier to estimate the vertical deformation by assuming a negligible horizontal displacement component and applying a projection of the LOS displacement on the vertical axis and call it “vertical displacement”. The displacement being a true 3D displacement, ignoring the horizontal component can only increase the uncertainty on the accuracy of the deformation interpretation.

  • Ascending and descending signal reflections are not located at the same locations. Which makes pairing selection to perform the decomposition into vertical and horizontal displacement components extremely difficult
  • Phase uncertainty increases the pairing difficulty, resulting into an increased displacement uncertainty

While it is true, that a surface displacement will cause a phase difference in the radar signal, a phase difference in the recorded signal is not always due to a displacement. This is verry important to understand for it is the foundation of a proper interpretation of InSAR time series. InSAR “displacement” time series are time series of phase change from which a probability of a ground displacement should be extracted (see below figure).

Unresolved phase uncertainties of the InSAR LOS time series are a limitation to the interpretation of the true displacement to accurately monitor ground deformation. Therefore, the delivered LOS InSAR data that we will call RAW-InSAR to differentiate it from a proper displacement data is simply unusable when it comes to accurately predict the deformation of small infrastructure: individual houses, small bridges, and roads.

InSAR time series can be converted into true displacement information and used to perform deformation predictions. However, the above uncertainties need to be addressed by the InSAR data provider.

3 – Displacement and deformation are the same thing – They are not!

The third and last misconception is about the fact that displacement does not always imply deformation. Displacement may generate stress that is the cause of deformation. Stress is a result of differential motion of the ground. The extraction from InSAR data of ground oscillations and displacement common trends contribute to understand the ground stress evolution. That is a more reliable attribute from which we can interpret ground deformation.

The lack of resolution due to excessive averaging of the InSAR data is a limitation in the interpretation and prediction of the ground deformation. It is well established in civil engineering that observed damage on structures is often associated with differential displacement. Structures tend to resist better to homogeneous displacement. Low resolution interpretation tends to detect mainly homogeneous displacement. For example: why is a house damaged when its neighbouring house is not affected? They are both submitted to the same averaged displacement according to the InSAR interpretation.

The observed damage on a structure does not appear by magic: it is the result of accumulated stress on that structure caused by the application of a force. The resistance to that force will determine the appearance and intensity of the damage. The difference in direction and intensity of the ground displacement against the prior equilibrium conditions of nearby areas will control the intensity of ground deformation. That in turn, when interacting with the surrounding structures’ resistance, will control the amount of recorded damage. A combination of plastic and elastic responses of the ground will generate various stress intensities on the local infrastructure. Therefore, the measured displacement alone cannot explain why a structure suffers a damage. Structures are sensitive to pressure, and the inherent qualities of a structure will determine how it will be affected by such a pressure.

It is then imperative to convert the observed displacement at the surface into a 3D stress estimation that is a statistical inference of the force applied to a surface. This starts to sound more like a geological problem than an InSAR problem, a problem one would expect to see resolved by a geoscientist rather than a space engineer. Geoscientists are needed to convert InSAR data into geological information to understand the complex dynamic of ground movement to derive the stress generated on targeted infrastructures.


While InSAR is a proven technology and a solid base for ground motion analysis, it carries also strong uncertainties that needs to be properly discussed. Converting RAW-InSAR LOS time series into deformation information is a difficult art that requires solid expertise in geodesy, geology, and geophysics to understand the mechanics of ground deformation, the generation of stress and the resistance of materials.

It is important to understand the interactions of external factors like climate change, temperature, water levels and the geology. The coupling between infrastructure and ground is also an important factor. And finally, the evolution of all the above interacting factors with time.

The interpretation of InSAR data is not here to replace conventional techniques of analysis of ground deformation but on the contrary, it is here to augment their potential and together bring new valuable geotechnical insight to optimise maintenance operations and the safety of infrastructure.

Also read

This article is part of our series on satellite remote sensing technology. You can find the entire series here.

NISAR mission by NASA and ISRO

Reda Meftahi

Related post