Accuracy of Sentinel-1 IW SLC Data for displacement

Hello,
I have a question I tried to measure the displacement with S1A IW SLC Data. I used this Graph:

Unwrapped the Phase with SnapHu, imported the Data again and converted the Phase to Displacement. It looked ok until i tried to validate my Data with other Images (not made out of S1 Data). My Results are sometimes up to 15cm off from the other Result.

I don’t know what i can do to make my results better I tried changing the filtersize or making the subset smaller. Is there a way to enhance the accuracy? Or is this normal?

Thanks in advance!

what do you mean by “other images” or “almost same time”?
Small changes in the imaging geometry (ascending vs. descending) or the acquisition time can make a huge difference, especially when atmospheric conditions are different.
Instead of processing all data in a graph, I would recommend keeping better track of the intermediate results. Many things can go wrong at the different steps, so you would be able to see when things get different.
Additionally, InSAR isn’t something that can be trusted at the cm scale throughout the whole image. Only areas whith high coherence are trustful and, as stated above, still can contain atmospheric distortions.

Thank you for your answer.
The other Images are displacement-maps provided to me for comparisons made with TerraSAR-Data.
Acquisition time left image: 11.08.2015 - 22.08.2015
Acquisition time right image: 12.08.2015 - 24.08.2015
I know that atmospheric conditions can affect the result, but the results are always different, even in areas with high coherence. This was a rather extreme example, but I can’t understand why they are always different, as far as I understand this the results should be rather similar given the comparable conditions?
I tried the process step by step and in my view it looks ok the whole way its just that the results don’t match up. Looking at the S1A displacement solo even looks ok.

so one pair is right looking and the other one is left looking?
In that case, large differences are totally reasonably because traditional DInSAR measures displacement along the line of sight (not vertical displacement).

Please have a look at these discussions:

@Len Could you upload the graph you put in your first topic?
This could serve as standard processing chain and save time to many of us.
Thank you.

Please note that there’s a sub-forum for graphs that has not seen much activity lately… :wink:

https://forum.step.esa.int/c/gpf-graphs