Replacing Fill Values/Missing Values with NaNs
Added by Simon Driscoll 12 months ago
Hello,
I have files from the Danish Meteorological Institute - temperatures and sea ice fraction. (There are two products for each, a later one and an earlier one). Hence four example netCDF files representing the two eras/products and two variables.
I believe CDO can be used to replace "Fill Values/Missing Values" with NaNs. There are many ways and options it seems to do this.
(I would like to be absolutely sure I have the correct command, as later I will regrid these files - if I understand right, if I have a missing value that is numeric (e.g. -128) and I regrid data, this could be regridded to e.g. -120, where a later check for 'Fill or Missing Values' in my regridded data would not understand, miss it, and accidentally interpret it as a real physical value.)
Therefore I think it makes sense to fill any/all FillValues and Missing Values with NaNs before regridding (as then I can just ignore any/all NaN values in my newly regridded netCDF safely), and I'd like to be absolutely sure I get them all. Is it ok (not too cheeky) to ask for the correct commands that work/do this for these specific files using CDO? I have attached them here. I can then use a for loop in bash to iterate through all 4 variants of these files.
Thank you so much!
Simon
Replies (1)
RE: Replacing Fill Values/Missing Values with NaNs - Added by Karin Meier-Fleischer 11 months ago
Hi Simon,
there is no need to replace the missing values because the _FillValue attribute is set and so the missing values are used like NaNs.
If you have a look at the files with e.g. panoply you'll see the missing areas and when remapping with CDO the remap routines take care of them.