Remapbil with curvilinear grids centered on the North Pole
Replies (5)
RE: Remapbil with curvilinear grids centered on the North Pole - Added by Ralf Mueller over 5 years ago
hi Timothee!
what's the issue? your source file has missing values indicated with
iicefrac:_FillValue = 1.e+20f ; iicefrac:missing_value = 1.e+20f ;
So I expect missing values in the target field, too. your source grid looks really awkward
| source | remapped |
 |
 |
There seem to be an issue with the coordinate bounds:
It seems to me, that some area of the target grid is simply not covered by the source grid, so there have to be missing values at these locations.
hth
ralf
RE: Remapbil with curvilinear grids centered on the North Pole - Added by Karin Meier-Fleischer over 5 years ago
Hi Timothee,
Ralf is right the input file looks strange.
Source grid: 
Target grid: 
However, if you want to remap the source data to the target grid, in this case curvilinear to curvilinear,
you should use another remap operator:
cdo remapdis,target-grid.nc,50 source-grid.nc remapped_dis.nc
Remapped grid: 
-Karin
RE: Remapbil with curvilinear grids centered on the North Pole - Added by Timothee Bourgeois over 5 years ago
Hello Karin and Ralf,
Thank you for your replies and I apologize for my unclear explanations.
My issue might be clearer with the figure below. The remapping with remapbil works well everywhere except in the North-West Pacific (see the black circles).
Do you think that it might come from my awkward target grid and its coordinate bounds? If yes, do you have an idea on how to define the latter to solve this issue? I have never noticed any issue using other dataviz or processing tools.
Regarding to the remapdis operator proposed by Karin, it solves my Pacific issue ! But it produces unwanted gradients along the coasts (see below left) that remapbil does not. When I mask the land cells of my source data, it removes the coastal gradients produced by remapdis (see below right) but it seems that remapdis is less accurate for my remapping application (at least with the range parameter 50). I do not find in the CDO user guide an explanation on this range parameter, in order to understand how it works and how to adjust it. remapdis is way more accurate with a low range like 5 but do you know if I should put a further lower range or if it may produce artefact?
Thank you so much for your help!
Timothée