Yes it is. It has been done numerous times. It works well as long as groundwater depletion leads to compaction of compressible layers within the lithology. You can contact me if you want more infos.

I agree with Dr. Castellazzi. I've read several research articles on the topic. The popular GRACE (Gravity Recovery and Climate Experiment) satellite mission is also relevant for the problem. However, GRACE mission is coming to an end https://grace.jpl.nasa.gov/

I may add that it does work in theory, but there are quite many error sources and influencing factors which you should be aware of whern you want to perform differential interferometry (DInSAR = measuring changes of surfaces):

• Wavelength: Short waves (X-and C-band) interact with smaller structures while long waves (L-band) are able to penetrate smaller vegetation cover. Some studies showed that the longer the wavelength, the higher is the accuracy of the measured surface.
• Temporal decorrelation: If surfaces change between the two image acquisitions (besides their elevation), regarding their vegetation cover for example, the phase signal is no longer coherent. This often leads to low quality of the results in that areas. Best conditions are dry surfaces with only little or no vegetation cover. Calculating subsidence under tropical forests, for example, is nearly impossible, unleass you don't have some distributed open areas.
• Incidence angle and orbit information: Interfeometry works best if the imaging geometry is known exactly. This involves the position of the satellites at the times of acquisiton (stored in orbit files) and the incidence angle of the signal. The better the information on these two factors the lower is the chance of miscalculations. Many standard techniques only measure the change of surface elevation along the line of sight (LOS, the line between the satellite and the pixel) instead of absolute uplift. Accordingly, the steeper your signal comes in, the closer it is to the actual horizontal uplift and vice versa.
• Quality of your DEM: In order to estimate the impact of topography on your radar phase a digital elevation model (DEM) is used by most software. The better the spatial resolution and accuracy of your DEM the more precise are these calculations and you are able to distinguish real phase difference from topographic effects. Later on, when you wan to geocode your images to a ground resolution, a good DEM also helps to reduce geometric distortions caused by side-looking geometry of the system (terrain correction).
• Perpendicular baseline: This is the distance between both satellites at the time when they took the SAR image. If it is too small or too large, the quality of your product decreases. See "critical baseline" or "spectral shift" if you want to know more about that.
• Atmosphere: SAR waves are able to penetrate clouds but still they are affected by humidity in the troposphere. Dry conditions at the days of image acquisition are recommended therefore. Again, long wave signals are more robust to that factor.

Excellent material on that is provided by the EO college:

• https://eo-college.org/resources/insar_basics/
• https://eo-college.org/resources/insar_errors/
• A full example work flow using the free software SNAP (used to be NEST) is given here: https://eo-college.org/resources/insar_deformation/

Thank you all for the very thoughtful comments.