Reference Point Modification

[4restw...@gmail.com]

Hello Yunjun and Community,

I'm interested in changing my reference point after I've already completed MintPy processing through the velocity step. Which step do I need to begin processing at again if I change my reference point? I know that initial point selection occurs before inversion, but it would be great if I did not need to repeat this step.

Thanks!

Forrest

[Zhang Yunjun]

Hi Forrest,

The impact of spatial referencing from time-series to time functions estimation (velocity) is known, however, it's unknown (to me) of its impact on the network inversion.

If you want to plot a map or time-series with a different reference point rather than the one used in the processing, view.py and tsview.py have the --ref-yx/lalo option for it, so no need for re-processing.

If you do want to re-process the data, then I would recommend to re-start from the inversion. It will be difficult to remember all the inconsistencies in the setup in the long turn, based on my persona experience. I ended up re-process it anyway ;)

Yunjun

[4restw...@gmail.com]

Thanks Yunjun this is very helpful! A follow-up question - should changing the reference point in a ifgramStack.h5 file change the average unwrapPhase values for the interferograms and/or the values that view.py displays when view.py ifgramStack.h5 unwrapPhase is called? My intuition is that they should be because the reference point is changing, but I've been modifying reference point values and these two things appear to remain static.

Forrest

[Zhang Yunjun]

The reference_point step in reality only updates the attributes by default, without modifying the big 3D matrice for efficiency. mintpy script would take this into consideration and apply this spatial referencing on the fly during calculation and plotting.  So, yes, the avgPhaseVelocity.h5 file and plotting result of view.py ifgramStack.h5 unwrapPhase should be different. How big it is would depend on the old/new reference points of course.

Yunjun