Accuracy of DEMs and deformation maps computed by SNAP?

Hello everyone!

I’m working with 2 Sentinel-1 images in the IW mode. I have been able to obtain the DEM of the place of study and the deformation between the two measurements, I have exported my results to a kmz format file in order to open them in Google Earth.
The elevation measurements are contained between 657,56 meters and 1520,36 meters while the deformation values are contained between 0.056 meters and -0.009 meters, well, now my question is: what is the accuracy of that values? Does the measurements really have an accuracy of centimeters in the case of the DEM and of milimeters in the case of the deformation map? How could I know the accuracy of the deformation measurements compared to the reality?

Thanks for your time.
Eduardo Ramos.

1 Like

Short answer: no, the real accuracy is much worse due to the effect of the changing atmosphere and other factors. Here’s a good reference for SAR Interferometry:

BTW it’s not clear how you could have generated both the DEM and a deformation map from a single image pair?

First of all, thank you for your fast answer mengdahl.

Well, if I have a phase difference between two images (with the help of an external DEM that SNAP downloads by itself) why I couldn’t generate a DEM or a deformation map? Maybe the more images you use, the more reduction of speckle and noise you have and maybe more accurate will be the measurement in the elevation axis (because I think the more images you use you reduce the spatial resolution too).

According to my previous question, I have thought that I could measure the accuracy of my generated DEM comparing it to an accurated DEM downloaded from the geographical institute of the country of the study place, but I have no idea of what I could do in the case of the deformation map, any suggestion?

Yes, if you have an external DEM you can produce both another DEM for comparison, or a deformation map. Just note that the atmospheric disturbances are probably much larger than the deformation signal. If you produce a large number of deformation maps and average them, the atmospheric disturbances mostly average out and you are left with a map of deformation in the image (if any). Deformation maps can be compared against ground-based levelling datasets.

Hello mengdahl,
I couldn’t answer you before. Now I have the doubt of what procedure you follow to obtain a DEM without using an external DEM, I am asking it basically because in the Range-Doppler Terrain Correction step, after the “Phase to elevation” step, SNAP asks you to enter a DEM (one external or one that the own software downloads).
I have been looking for ground-based levelling datasets on the net in order to compare them with my deformation map results and I have found nothing about it…

It is likely that the atmospheric artifacts are at least an order of magnitude larger than the deformation signal, unless you’re working in a desert where the deformation during 12 days approaches the radar wavelength…

Sorry mengdahl, but I can’t find the relationship between my question and your answer :slight_smile:

The DEM generation operator can be found under Radar/Interferometric/Products

So, do you use an Ellipsoid Correction (Average Height Range-Doppler) instead of a Terrain Correction (Range-Doppler Terrain Correction)?