Add alternative geotif read to use interferograms directly off the portal.

[ditafaith]

Hi @andwatson, I didn't find any files such as .diff, .par, *.unw from the portal. Mostly all files are provided in tiff format, thus, can't be used in this notebook.

[andwatson]

Hi @ditafaith, the files used in the practical are slightly modified versions of the data on the portal. From memory (it's been a while), I took the tif files from the portal, cropped them, converted to UTM, centred the coordinates at the centre of the crop, set a local reference for the phase, and then saved them to float32.

I'm afraid I don't think I have access to this pre-processing script anymore. If you want to use this to model other earthquakes, I'd suggest copying it locally and writing something to read tifs from the portal directly. You could also look into something like GBIS if you want an actual inversion.

[ditafaith]

Thanks @andwatson for updating the codes. Let me try to read the data. Further, in my poor understanding, GBIS is designed to read a matlab file meanwhile my licsbas results looks like not compatible one.

[andwatson]

Yea GBIS is written for Matlab, but it should be possible to structure the LiCSBAS outputs in a way that it will read. Someone may also have already written the code to do that - might be worth asking someone COMET-related at Leeds Uni.

[ditafaith]

Thanks @andwatson. Is there any available tool to downsample the data? As you know that this is critical stage prior to further process.

[andwatson]

Depends what you're doing/working with.

For an inversion, you may want to look into techniques like quadtree downsampling (GBIS might do this from memory), which balance preserving the signal with reducing computational load.

If you're using outputs from LiCSBAS, then you could just run your time series with more multilooking.

If you're just making inputs to use with this notebook, then there's probably python functions for doing basic downsampling that will be good enough.

Out of curiosity, is this work for research? I would say that this notebook is essentially just a teaching tool, and that if you're aiming to do something like modelling earthquake signals then you may want to look at more advanced methods. Maybe something like pyrocko if you're working mostly in python?

[ditafaith]

Thanks @andwatson for answering my inquiries.

If my understanding is correct, you meant this notebook fit for educational purpose and made for licsbas's output. Also, the downsample stage can be done in multilook process through the licsbas. Yes, I do it for research and still lack of knowledge. May be python version of gbis will available shortly.

[andwatson]

No worries.

Yea this is a teaching tool. Also its made to work with LiCSAR outputs (i.e. interferograms), as opposed to LiCSBAS outputs (i.e. time series), although it potentially can depending on the signal you're looking at.

I wouldn't expect a python version of GBIS. GBIS itself was written a good number of years ago and is not under active development outside of some phd students modifying it for their own projects.

I don't know if its too late now, but the COMET InSAR workshop may be useful if you haven't attend before:

https://comet.nerc.ac.uk/comet-insar-training-workshop-2024/insar-training-workshop-2024/

Could be worth an email to see if there's space to join. This notebook was made for this workshop. It would also provide opportunities to discuss earthquake/fault modelling with other people who've done so.

[ditafaith]

Glad it to having this information. I'll join this event. Anyway, did you convert tif file to float using gdal ? This notebook is good step for beginner like me to understand earthquake science prior to use other tools.

[andwatson]

I think I did the tif to float conversion in Matlab, but I'm pretty sure you could do it with gdal as well.

Enjoy the course!