Interpolation across 180 degrees.
Hi everyone!
I have a somewhat complicated question if everyone is not on summer vacation. 
I must remapcon interpolate from several CMIP6 ocean curvilinear grids onto a custom curvilinear grid. The target grid has cells on either side of 180W. I am looking for advice on whether this might work. My past work has been limited to the Scotian Shelf (off the east coast of Canada). I did not have to worry about the 180W discontinuities in the CMIP6 ocean grids.
If I have a target grid cell crossing the source grid discontinuity, will the interpolation fail? I already know that I will have to manually trim out any ocean grid halos in the data. Uwe has kindly shown me how to do this already. I am worried my cdo processing will either fail or have improper results.
Thanks!
Replies (3)
RE: Interpolation across 180 degrees. - Added by Estanislao Gavilan over 2 years ago
Hi Brendan,
I can tell you my experience on this. I have interpolated CMIP6 ORCA grid to ECCO2 without major issues. It is true that there are some cells in the north where the interpolation fail a little. Espcifically in the siberian shelf where there is an irregular discontinuity (see red anomalies in Figure). In order to correct that area I wrote a python script to locate those naughty points. It is not difficult to find them because they are spikes (the go from 10 to 3000). You can run a simple script which compare the adjacent values, and then interpolate them.
About your specific problem, orca grid does not have discontinuity over 180W but, cdo handles really well the discontinuty in the greenwich meridian (see Figure).
Kind regards,
Estanislao