sellonlatbox with curvlinear grid

Hi all,

I'm trying to select a region using sellonlatbox from a curvlinear grid:

Specifically, in the Indian Ocean from
40°E - 120°E, and 10°S - 10°N

While

cdo sellonlatbox,40,120,-10,10 test.nc test_selected.nc

works fine for most models I process, for some models this function ends up selecting the whole area from -180 to +180°E from 10°S to 10°N.
I'm not sure why that is the case.

I attached two sample files: working_file.nc the area is correctly selected, not_working_file.nc the area is not correctly selected and it's not clear to me why that is the case. I would be grateful for advice on how I can avoid this problem. For other regions, like in the eastern Pacific or the Atlantic, cdo sellonlatbox selects the correct grid for both files.

Download all files

working_file.nc (1.38 MB) working_file.nc
not_working_file.nc (8.22 MB) not_working_file.nc

Replies (3)

RE: sellonlatbox with curvlinear grid - Added by Ralf Mueller over 5 years ago

hi Thorsten!

thx for this report - both files look good regarding the coordinate definitions and I can reproduce this behavior with the current CDO release. Interestingly there is a break point when prcessing the not_working file 73 n degE:

fail

% cdo -sinfov -sellonlatbox,40,73,-10,10 not_working_file.nc
cdo(1) sellonlatbox: Process started
Warning (find_time_vars): Time variable >time< not found!
   File format : NetCDF4 zip
    -1 : Institut Source   T Steptype Levels Num    Points Num Dtype : Parameter name
     1 : unknown  CNRM-ESM2-1 c instant       1   1     17689   1  F32z : areacello     
   Grid coordinates :
     1 : curvilinear              : points=17689 (361x49)
                              lon : -179.5 to 179.5 degrees_east  circular
                              lat : -9.849413 to 9.849413 degrees_north
                        available : cellbounds
   Vertical coordinates :
     1 : surface                  : levels=1
cdo(1) sellonlatbox: Processed 1 variable over 1 timestep.
cdo    sinfon: Processed 1 variable [0.03s 69MB].

correct

% cdo -sinfov -sellonlatbox,40,72,-10,10 not_working_file.nc
cdo(1) sellonlatbox: Process started
Warning (find_time_vars): Time variable >time< not found!
   File format : NetCDF4 zip
    -1 : Institut Source   T Steptype Levels Num    Points Num Dtype : Parameter name
     1 : unknown  CNRM-ESM2-1 c instant       1   1      1568   1  F32z : areacello     
   Grid coordinates :
     1 : curvilinear              : points=1568 (32x49)
                              lon : 40.5 to 71.5 by 1 degrees_east
                              lat : -9.849413 to 9.849413 degrees_north
                        available : cellbounds
   Vertical coordinates :
     1 : surface                  : levels=1
cdo(1) sellonlatbox: Processed 1 variable over 1 timestep.
cdo    sinfon: Processed 1 variable [0.02s 68MB].

Will have to dig a little deeper for sure.

73 degE is also a special case for the lower latitude give in the command: -sellonlatbox,74,120,-10,10 works and -sellonlatbox,73,120,-10,10 fails

cheers
ralf

RE: sellonlatbox with curvlinear grid - Added by Uwe Schulzweida over 5 years ago

For the application of sellonlatbox on curvilinear grids the coordinates must be correct.
Otherwise, it cannot be determined whether the longitudes are cyclic or not.
The first and last column are duplicated. And the first and last row contain wrong values for the corners of the cell.
These rows and columns must be removed before sellonatbox can be applied. Here is an example:

cdo sellonlatbox,40,120,-10,10 -selindexbox,2,361,2,293 not_working_file.nc result

RE: sellonlatbox with curvlinear grid - Added by Thorsten Muzs over 5 years ago

Dear Ralf and Uwe,

Thank you very much for your help, now it is working!

(1-3/3)

Project

General

Profile

CDO