Cropping dataset with rotated poles

[ClimateYukon]

Hello,

I have been looking for a while without much success, I am wondering how to transform a xarray dataset showing a rotated pole into normal lat lon. I am trying to compare a few climate models and some of them such CORDEX or RDRS are using rotated poles.

One example of dataset would be :

Is there an easy way to do this? My goal is to crop it with a shapefile later down the process but I figured the "derotation" should be the first step?

Any help appreciated,

Thank you so much

[kmuehlbauer]

Re-gridding to 1D lat/lon always involves interpolation and that is sometimes not really wanted. There is #2281 discussing potential approaches.

You might also use the provided 2D lat/lon coordinates for cropping/masking. As always this depends on your actual use-case. The below example just crops with a simple bounding box.

ds_clip = ds.where(
    (
        ((ds.lat > bbox["bottom"]) & (ds.lat < bbox["top"]))
        & ((ds.lon > bbox["left"]) & (ds.lon < bbox["right"]))
    ),
    drop=True,
)

Another approach would be to project your shapefile coordinates into the rotated pole domain and do cropping and all further computation in that domain. This at least excludes any errors which might be introduced by grid interpolation.

[ClimateYukon]

Thank you! I will give it a try.