Difference between vertical and LOS displacement

Hi all,

Can you please tell me which formula shall I use to get the vertical displacement from LOS displacement?

Is it - Vertical_displacement = LOS_displacement * cos(rad(Incidence_angle)) ?
Or is it - Vertical_displacement = LOS_displacement / cos(rad(Incidence_angle))?

Thank you for help!


could you please upload screenshots of the two plots to demonstrate how they are of different scale?

It’s free here:


1 Like

@johngan ,@ABraun I have a question here.
When calculating vertical displacement or three-dimensional displacement from LOS, which angle of incidence should be selected?
There are three types when deriving the angle of incidence in the data, including:
Incidence angle from ellipsoid
local incidence angle
projected local incidence angle

If the incidence angle here is the angle between the radar beam and the normal to the scattering surface.
Then according to the picture below, what I should export is the local incidence angle, I don’t know if it is correct.

Thank you Johngan…

But why?..I must say, I had many and great problems with the normal steps from interferograms, I asked for the help of “@ABraun” and the problem was not solved, You can check here: Problem with the final steps of interferometry I was totally frustrated. But someone suggested me make the correction after the goldstein filtering, just for try, and since then all my interferograms have logic, I can detect landslide, flows and deformation correctly I’ve make hundreds of interferograms since that moment, and now all my afirmations from interferometry have sense …So can you explain me please. Why the last step must be the geometric correction? is the only way? or is a very good suggestion?

Terrain direction itself does not automatically solve problems, but my guess is that the resampling which is part of the terrain correction smoothes some of the noise within the interferograms.
If this is the case with your data, I would rather recommend to apply multi looking on the interferogram which has the same effect, but does not project the data into a coordinate system.
I agree with @johngan that this should be done after calculating the displacement.

Can you please show how the results change with terrain correction applied earlier in the workflow? Bases on the from the other topic, my guess is still that for some image pairs decorrelation is too high and unwrapping will introduce random patterns which are no longer trrelated to the deformation at all.

1 Like

Thanks @ABraun . Do you mean resampling the second step? the interferogram formation?.

After interferogram formation, apply Radar > SAR Utilities > Multi looking for a test (eventually also a Goldstein filter) and see if the unwrapping produces better results.

Hello. I might have a question about the accuracy of the Los velocity deformation and the actual PS displacement of the time series when the process is done by StaMPS. I am wondering which is the low limit untill I can considerer the displacement a reasonable value. In one article of Colesanti is written down the following:
So I want to find out whats my PS results accuracy.
I am working with Cosmo-Skymed StripMap Himage data, and the results of my LOS velocity displacement are very low( 0.5mm), even though my movements go from positive to negative with great values ( around 20 mm). This big difference make me thought if I actually can use this results, becouse somewhere i read that the method cant measure movement lower than 2mm/year…

Good afternoon, I have a question. During the procedure for the Dinsar technique, is it necessary to use multilooking? Because when I use it and not when I have totally different results. Could you please help me.

multi-looking aggregates pixels and therefore potentially affects the unwrapping process. These strong phase jumps should not happen, so I would rather recommend phase filtering in this case to grant smoother transitions.


Hello, in the literature it appears as decompose to vertical and E-W components. If the following formula applies:

I have 2 doubts:

  1. Should I apply this formula first to ascend and then descend before rasterizing the images? Or should I apply it after rasterizing the images? to combine them. I dont think is necesary to rasterizing the images because we have points (exported from stamps).

  2. both angle, heading angle and incidence angle in SNAP, in what units do they appear?
    If I use my data. I’m not sure if I should change both angles to radians or if I should do some other operation that I’m not taking into account

This question is because I find it very strange that in the literature it is said that Sentinel normally has angles of -15deg and -165deg in heading angle and 29-46 deg in incidence angle.

and in SNAP my data is -1.666621007081088e+02 heading angle.

Please I need clarification if I should convert my data or use it as it is seen in SNAP to apply the formulas.

Thank you.

I’d like to share the experience of the EGMS team on extracting deformation from LOS phase
expressed in this document
section C, page 111 onwards


Nice reference - thank you!

I hope in the next release of SNAP it will be realized :slight_smile:

To combine the ascending vertical displacement and the descending vertical displacement and being able to get 1 single result for each pixel, would I only have to apply the Vector Decomposition? That is, the following formula:
d = √ ddes² + dasc²

Please explain this equation once?

Hello @ABraun, I have a basic question related to this vertical displacement issue. Can we use only ascending OR descending orbit to calculate the vertical displacement from LOS? OR Is it necessary to have both ascending and descending for vertical displacement? Also, the same question is for horizontal displacement? Is it necessary to both ascending and descending orbit for eats-west displacement?

If I have a pair of SAR images : (a) master image in 2015 and (b) slave image in 2022 the displacement value (Dinsar) is measured in cm/year or in cm/7 years?