Using `interpolate_grid_label()` to regrid data about xmip HOT 4 CLOSED

I was able to regrid the data with xesmf (with the help of this notebook). I plan on writing up a tutorial to add to the Project Pythia CMIP6 Cookbook, which is a slightly modified version of the one in the Pangeo gallery. Thanks again for the help!

from xmip.

Hi @r-ford, thanks for using xMIP! The interpolate_grid_label function is a specific use case for general regridding. It will regrid outputs on different grids 'per model', e.g. if you have variable_a on grid_label 'gn' and variable_b on grid_label 'gr' this will attempt to return both variables on the same grid. This will however still leave the grid between different models incompatible (which I think is what you are experiencing here).

I think what you are trying to achieve is actually simpler. I would suggest to use xesmf(which is what we are using here under the hood) to regrid all models to a common lat/lon grid of your choice, and then you should be able to average them together on the new grid.

Let me know if that works for you

from xmip.

Thanks @jbusecke! That makes sense. xesmf seems like my best option, though I've been running into some issues getting results so far (one example is in this other notebook, where the regridded data is all nan). Do you know of any examples on GitHub or elsewhere of someone regridding this CMIP6 data? Also @brian-rose

from xmip.

There are some common tricks (I think there is a 'masking' or similar argument that might help you here). But generally this case would be a nice demonstration to include either in the xMIP docs or the pangeo gallery.

from xmip.