Sentinel-1 RTC Median over whole country

[JPPereira93]

Hi!

I've been using stackstac for some time and recently I had a project where i need to create a median composite of sentinel 1 rtc images over a whole country. I have been having some challenges with the workflow because just simply processing up to 24 items leads to huge RAM usage (up to 80%) in a 256 gb machine when applying .sel method for example.

Is there a way to optimize this? How is the array 900gb with only 24 scenes? Is it due to the bounds of the whole AOI? Each sentinel-1 rtc image is around 1,5gb

I'm trying to do some pre processing operations before applying stackstac.to_dataset, for example, clipping each item to the area of interest that i have.

What i am trying to do:

but it outputs an error that it isnt in the correct format to apply stackstac.stack, therefore i was trying to simply concatenate all the datarrays that i had from that clipping.

Thank you