Climate forcing file: differences in shape between 2. and 0.5 degree files

[yansun8]

The routine reading in climate forcing cannot deal with .5 degree forcing files. This is because the forcing files for half-degree are with a dimension of "tstep land". The number for “land” is 67420 but not 360720.

The reading routine is Tools/readvar.py ; file with .5 forcing is e.g. :/home/orchideeshare/igcmg/IGCM/SRF/METEO/CRUJRA/v2/halfdeg/

Potential solution: adjust the reading depending on the resolution, and re-shape in case of .5 degree to 360*720 at the beginning of 'readvar.py'.

[vbast]

The .5 degree forcing files are normally written in "compressed" form. Instead of the three dimensions (time, lat, lon) the variables have two dimensions (time, land) and the data are stored only for the land points, whereas all ocean points are just dropped. Thus, the size of the forcing file in compressed form is 4 times less.

You need to add few lines to uncompress the variable data. In readvar.py after reading the variable data in line 56 you can add the following:

  da=f[varname_clim[index]][:]
  if "land" in f[varname_clim[index]].dimensions:
    land=f["land"][:]-1
    ntime=len(da)
    nlat=len(f.dimensions["y"])
    nlon=len(f.dimensions["x"])
    uncomp=np.ma.masked_all((ntime,nlat*nlon))
    uncomp[:,land]=da
    da=uncomp.reshape((ntime,nlat,nlon))

[dsgoll123]

Perfect, thanks Vlad - crash is avoided. I do some additional test before comitting the change set