ClimateImpactLab / downscaleCMIP6

Downscaling & bias correction of CMIP6 tasmin, tasmax, and pr for the R/CIL GDPCIR project
MIT License
128 stars 31 forks source link

Documentation on running downscaleCMIP6 #655

Open sg2400 opened 1 year ago

sg2400 commented 1 year ago

I would like to use the downscaleCMIP6 code to create bias corrected and downscaled daily 1/4-degree CMIP6. Could you please point me to any step-wise documentation to set up the code on a workstation, or any particular notebook in the repository that I can use for a small scale example, e.g. one GCM in a small regional area for a few years?

I did install dodola in a conda environment, but I am not sure what to do next.

Thanks

brews commented 1 year ago

Hey @sg2400! That's a great question. We've wanted to write a small demo for some time (e.g. see #169). Most of what's in this repo (and dodola) is for running the method at a large scale in the cloud. Let me talk with the team and we'll see if we can write a smaller laptop-sized example with a subset of real data or maybe some toy data.

sg2400 commented 1 year ago

Thank you. That would be really helpful.

brews commented 1 year ago

Small update on this. I'm blocked from making progress because QPLAD (the downscaling step) depends on a fork of the xclim package that has not merged yet. Unfortunately, this fork and the newer versions of xclim are incompatible because of breaking changes in their dependencies. This means it can be very difficult for users to build the environment to run an example. I'm going to need to work around this or we'll need to wait for the fork to merge into xclim.