loading only a small region

[akasom89]

Thanks for developing such a great tool.

I am interested to do a time-series analysis over several years for a small region on my local system.

Is it possible to download a one-degree by one-degree region ONLY using this library(instead of downloading the whole scene for each observation)?

Thanks

[ghiggi]

Unfortunately, despite the conceptual simplicity, this is a difficult task to perform without downloading whole scenes ...
However, you could try using kerchunk or virtualizarr to create on your local system the references for all files and then just read on the fly all the data chunk of interest from the S3 bucket.

[akasom89]

Thanks @ghiggi

You are exactly right. It seems much more challenging than I thought. Do you know of any tutorial or documentation with the same goal?
I would greatly appreciate it if you could share any.

[ghiggi]

Not really sorry.
I would suggest to adapt the code in the VirtualiZarr Usage section: https://virtualizarr.readthedocs.io/en/stable/index.html#usage
Instead of glob.glob('/my/files*.nc') pass the filepaths returned by goes_api.find_files with protocol="s3" and connection_type="bucket".
See tutorial here on how to search files: https://github.com/ghiggi/goes_api/blob/main/tutorials/01_find_utility.py
Let me know how it works ;)