I just checked the availability of our needed variables for the piControl run Example: thetao, and there are only the two GFDL models that saved e.g. thetao and zos!
I think that means that pre-training on piControl runs from CMIP data is infeasible for us. We should probably start thinking about using the historical and scenario data?
I just checked the availability of our needed variables for the piControl run Example: thetao, and there are only the two GFDL models that saved e.g. thetao and zos!
I think that means that pre-training on piControl runs from CMIP data is infeasible for us. We should probably start thinking about using the historical and scenario data?