LOS displacement numbers seem unnaturally large

Hello,

I am very new to InSAR, so I apologize if I leave out any necessary context and ramble on trying to explain myself.

I am being tasked to look at LOS surface displacement along a very large river system (the area is contained within 4 bursts). My overall objective is to understand LOS surface change over time induced by varying river conditions and validate my displacement maps with GNSS reference stations positioned nearby. I am using Sentinel-1 SLC data downloaded from ASF’s data repository.

I have completed the ASF tutorial (https://asf.alaska.edu/how-to/data-recipes/create-an-interferogram-using-esas-sentinel-1-toolbox/) and the ESA interferometry tutorial (https://step.esa.int/docs/tutorials/S1TBX%20TOPSAR%20Interferometry%20with%20Sentinel-1%20Tutorial_v2.pdf). When I try to follow the steps outlined within the ESA tutorial using my own data, I am able to successfully produce displacement maps, but the displacement numbers at the most coherent locations seem rather large (10s of centimeters) over a 12 day period. Realistically, these locations are shifting very minimally as there is little tectonic activity or subsidence in this location.


I have 3 primary questions:

1.) Why are my LOS displacement values so high for coherent locations?

2.) For coherent pixel spaces, what type of accuracy in LOS displacement measurements should I expect for a single displacement map? For example, assuming that I processed my data correctly, the data is good, and I am investigating coherent enough pixel spaces, does interferometry provide displacement measurements that are accurate to 10 dm, 10 cm, 10 mm, or what? I know that a long time series can provide a more accurate representation of the movement happening at a given location and can therefore average out some of my unrealistically high displacement values, but what can be expected from the accuracy of a given displacement measurement?

3.) What resources would you all suggest for someone who is looking to perform this type of analysis?

Please let me know if I need to elaborate and/or provide more information.

Thanks!

InSAR processing is based on phase change measurements but the change is produced not only the ground movements but is mostly related to atmospheric delays. The atmosphere produced delays are real and the related fake surface movements have amplitude 100-200 mm. That’s a single interferogram accuracy in case you expelling time-series analysis. For large ground movements with visible fringes we still can estimate the real movements but not for small ones. Time-series analysis like SBAS uses least-squares approach to extract the real movements excluding atmospheric phase and seasonal trend analysis is even more powerful detecting accurate ground changes on top of SBAS results.
Also, your result obviously includes linear trend which should be removed first. You can find the interactive examples on PyGMTSAR Python InSAR github page: GitHub - mobigroup/gmtsar: PyGMTSAR