Strange result when using remapbic
Added by Francis Rayder about 10 years ago
I have an input file of Earth's topography, with all the oceans (and land areas below sea level) set to 0. I used a fldmin test on the input file to make sure of this - the lowest value is 0. Now, when I use a bicubic interpolation to a T42 grid,...
cdo remapbic,t42grid input.nc output.nc
...the output file has some cells with negative values. How is it possible that there are any cells with negative values when the lowest value in the input file is 0?
When I use bilinear interpolation, I do not get negative values. So I can use that method instead. Still, I would be interested in what is wrong with the remapbic command here?
Replies (4)
RE: Strange result when using remapbic - Added by Uwe Schulzweida about 10 years ago
Hi Francis,
We can't reproduce the negative values with remapbic.
Could your please upload your input file?
Cheers,
Uwe
RE: Strange result when using remapbic - Added by Francis Rayder about 10 years ago
Here you are. I have also added the output file with the negative values.
RE: Strange result when using remapbic - Added by Uwe Schulzweida about 10 years ago
Bicubic interpolation uses the gradients of the neighbor points. Large gradients could cause that the results is outside the range of the input array. That is very similar to a spline interpolation.
RE: Strange result when using remapbic - Added by Francis Rayder about 10 years ago
I understand. Thank you for your reply!
As I wrote, I can use bilinear interpolation instead and get results without negative values. So it is not a problem, really.