pyramid_coarsen (or a more generalized function) should allow a custom coarsen func

[ahuang11]

I think the following:

    for key, factor in enumerate(factors):
        # merge dictionary via union operator
        kwargs |= {d: factor for d in dims}
        plevels[str(key)] = ds.coarsen(**kwargs).mean()

Can be generalized:

    for key, factor in enumerate(factors):
        plevels[str(key)] = custom_func(ds, factor)

Potential use case, if I want to use xesmf regrid or like downsample1d operation

I'm happy to contribute a PR if this is desired.

[maxrjones]

Hi @ahuang11 I wanted to let you know that we'll be having a breakout group on pyramids at the Zarr sprint on Thursday. More details are shared in #101.

I'll try to find time to look in more detail at the two issues you posted this week. On this topic, we already support xesmf regridders through the pyramid_regrid function. Is the downsample1d operation referenced in the discussion you linked part of holoviews? This approach would be useful if it could help add support pyresampler's downsampling operations (xref #10).

[ahuang11]

Thanks for letting me know! I'm at a separate conference right now, but I'll try to attend if I have some spare time!

[ahuang11]

Created a PR here #120; I think it could support #10. Let me know what you think!