Project

General

Profile

reducegrid - limit fields to a horizontal mask

Added by Ralf Mueller about 8 years ago

In the next CDO (1.7.2) release there will be an operator present, which can reduce data fields to non-zero locations of a horizontal mask. This can be used to
  • reduce data files
  • create new grids out of existing ones

Reducing to subgrid - based on topography

Lets start with an example on a regular grid: Suppose there is data on a horizontal grid and only the land points are relevant. Hence 2/3 of the grid points are superfluous.
For better visibility I choose a very coarse grid: r18x9. The input must be on the same grid - I use the internal topography for this

cdo -f nc -topo,r18x9 topo_r18x9.nc

Same for the mask
cdo -f nc -gtc,0 -topo,r18x9 mask_r18x9.nc
Now lets reduce the topography to the non-zero values, i.e. the land locations of the input file
cdo -reducegrid,mask_r18x9.nc topo_r18x9.nc reduced_topo_r18x9.nc
This look like this:

Looks strange, but the south pole is visible, whereas the north pole is pure water.

Reducing to subgrid - based on coordinates

The next input grid is a more realistic: A global 160km ICON grid. The mask is created wrt. to coordinates with the expr operator

cdo -f nc -expr,'mask=sin(clon(const)); mask=(abs(mask) > 0.5)' -remapnn,iconModelOutput.nc -const,0,r1400x720 imask.nc

Now I reduce a temperature field to that grid with
cdo -C -v -reducegrid,imask.nc iconGlobalTemp.nc maskedIconTemperature.nc
In paraview this looks like this

Reducing to subgrid - based on temperature

The last example is based on the GME grid, a hexagonal grid used for weather prediction at DWD. Again I use temperture for data

cdo -temp,ni96 temp_gme.grb

But the mask is based on that data: I want to cut out all locations with zero or negative temperature

cdo -gtc,0 temp_gme.grb mask_gme.grb

The reduction is called just like in the other examples. But the output format has to set to NetCDF, because general unstructured grids cannot be stored in GRIB1 or GRIB2 format.

cdo -f nc reducegrid,mask_gme.grb temp_gme.grb masked_temp_gme.nc

The resulting fields looks like this

Data reduction

For pure data reduction, user can decide to avoid coordinate bounds or coordinates at all:

  • coordinates are completely ignored in the output, if nocoords is given
    cdo -v reducegrid,<mask>,nocoords <ifile> <ofile>
  • if only the bounds should be suppressed, but coordinates are needed, use nobounds instead
    cdo -v reducegrid,<mask>,nobounds <ifile> <ofile>

NOTE, that this limits the usability of the output files: Without coordinates or bounds operations that involve data location are no longer possible, e.g. spacial averaging, geographical selection, ... If needed, the grid can be added later with the setgrid operator.

If you have any other ideas on how to use this operator, please share!

cheers
ralf