Forums » Support »
Bilinear interpolation, or not?
Added by F Lapointe almost 5 years ago
Hi,
I am currently transforming a dataset from spherical projection (lat/lon on a sphere) into stereographic projection (X/Y) using attached CDO_proj on my data test2.nc. I use ' cdo remapbil,CDO_proj test2.nc test_r.nc'.
The results are really strange as I only have a partial answer shown in test_r.nc, as it was only taking partially the data without interpolation as the remainder data are 'NaN'. I thought that 'remapbil' was interpolating my current dataset into the defined projection?
What happened?!
Thank you for your help.
Best,
François
I have forgotten to mention that I used:
ncatted -a coordinates,delta_SL,c,c,"X Y" test2.nc
prior to remap the data...
The geographical lat/lon coordinates contains confusing values.
Y is in the range of -93 to 2.35e+18 and the structure doesn't looks like latitude coordinates on a curvilinear grid.
The longitude values in X looks like latitudes.
Hi,
thanks for the output Uwe. I have found out where the data were corrupted--see new attached file. The 'Y' are longitude (-93 to 15) and 'X' are latitude (55 to 85). However, the problem persists as I have got sparsely output shown in 'test_r.nc'--see first e-mail. It seems that 'remapbil' is not interpolating data for other lat/lon on which they do not previously have data from tes2.nc. Do you know what could create such sparsely output?
Thanks...
François
Hi François,
Here are some plots of the original file, the polar_stereographic interpolation and the interpolation back to the original grid.
The result looks correct for me.
Original input file:
cdo remapbil,CDO_proj test2.nc pstereo.nc
cdo remapbil,test2.nc pstereo.nc regular.nc
Cheers,
Uwe
Thank you very much....
François