Regridding curvilinear data

[DamienIrving]

I'm a little concerned about the iris.experimental.regrid.regrid_weighted_curvilinear_to_rectilinear function, as it produced some funky results when regriddinghfds * areacello data (but not in other cases, which is weird).
https://groups.google.com/forum/#!topic/scitools-iris/qIO-tZh0tj4

It looks like a good alternative might be ncremap. I'm having a little bit of trouble figuring out exactly how to use it, so I've logged a question at the NCO help forum:
https://sourceforge.net/p/nco/discussion/9830/thread/da8a8a58/

[DamienIrving]

Another option is cdo:

cdo remapbil,sftlf_fx_NorESM1-M_rcp85_r0i0p0.nc hfds_Omon_NorESM1-M_rcp85_r1i1p1_200601-210012.nc remapped/hfds_Omon_NorESM1-M_rcp85_r1i1p1_200601-210012.nc

(Although when I use this instead of iris for regridding hfds data -- albeit not multiplied by areacello -- it doesn't make much of a difference to the end result.)

[DamienIrving]

xESMF has been written on top of ESMPy to make things much easier. It uses xarray data arrays.

[DamienIrving]

I had some problems with the IPSL-CM5A-LR ocean model grid due to disconnected tiles. See the following issue for details: JiaweiZhuang/xESMF#49