CDO

I have a problem with unstructured grids and remapping.
After trying the example from the CDO User´s Guide (June 14; D.2. Example description for an unstructured grid), which didn´t work for me, I tried to find the problem.
(by the way, I think there the Example from the User´s guide has an error - the file has 30 cells and the picture has around 42)

This is what I have done so far.

FILES:
------
world.nc is the variable "fraction land" from ERA-Interim
plot_unstructured.R is an R-script to plot unstructured-grid data
GRID_rotlatlon_world_10degree.txt is a grid from -170 to 170 lon and -80 to 80 lat
GRID_rotlatlon_small_world_10degree.txt is a grid from -150 to 150 lon and -60 to 60 lat
GRID_unstructured_world_10degree_ONEvertex.txt is a grid from -170 to 170 lon and -80 to 80 lat
GRID_unstructured_world_10degree_FOURvertex.txt is a grid from -170 to 170 lon and -80 to 80 lat

CDO-USAGE:
----------

step A: remapped world.nc to a coarse 10degree grid:
(1) cdo -remapbil,GRID_rotlatlon_world_10degree.txt world.nc world_10degree.nc

step B: remap world_10degree.nc to an unstructured grid using nearest neighbor
(2) cdo -remapnn,GRID_unstructured_world_10degree_ONEvertex.txt world_10degree.nc world_10degree_nn.nc

step C: remap world_10degree_nn.nc back to the (rotated) latlon grid
(3) cdo -remapnn,GRID_rotlatlon_world_10degree.txt world_10degree_nn.nc world_10degree_nn_nn.nc

RESULTS:
--------
For command (1) and earlier (not documented) steps, I only used CDO 1.5.3
For the other steps I used different versions:

command (2):
CDO 1.6.4 prints error [cdo remapnn (Abort): unstructured grid without latitude coordinates!]
CDO 1.5.3 , 1.6.1 and 1.6.7 all seem to produce a correct netcdf-file (based on viewing the data with the R-script plot_unstructured.R)

command (3):
CDO 1.5.3 and 1.6.1 produce wierd data (see pictures)
CDO 1.6.7 produces a field that is constant=1

an overall picture is given by the uploaded file "remapping.png"

ANALYSIS:
---------
'cdo griddes' & 'cdo sinfo' give me the following informations on the world_10degree_nn.nc file
CDO 1.5.3 and CDO 1.6.1 show "nvertex = 0"
CDO 1.6.4 and CDO 1.6.7 show no information about nvertex

The results seem to be indepented of the usage of the unstructured grid file (1 vertex or 4 vertex)
The problems seem to be indepented of the usage of remapnn and remapdis

Remapping in the last step to a smaller domain (GRID_rotlatlon_small_world_10degree.txt) shows very differnt results for CDO 1.5.3 and 1.6.1
that don´t look like subsection

---

Should the unstructured grids work? or did I made error, when creating them?
I would be grateful if someone can send me an unstructured grid (or data on an unstructured grid) that work with cdo remap...