@@ -18,7 +18,7 @@ This example will show how to create an animation based on ocean, vegetation, so
.. image:: combined-image.png
In this example we assume a very basic level of familiarity with Paraview. If you feel lost in between, consider looking at :ref:`create-image-sea-surface-speeds`.
In this example we assume a very basic level of familiarity with Paraview. If you feel lost in between, consider looking at :ref:`create-image-sea-surface-speeds`. This multi-variable visualization requires a substantial amount of :ref:`paraview-palmod-pre-processing`.
* For general advice on the NetCDF CF reader see :ref:`netcdf-cf-reader`
* All files need the same time axis. Different models will usually use different specifications, sending you straight to hell. One solution that works: resample all files to the same number of time steps, then use::
ncks -Av time MASTER_FILE FILE1
ncks -Av time MASTER_FILE FILE2
...
to get all on the same time axis. If that causes trouble (e.g. because of float/double conversions), look into ``cdo -settaxis``.
.. warning::
The time axis must not contain negative values (don't ask how long it took to figure this out)
* For extruding with topography it's best to have the topography in the same file as the data of interest. See :ref:`extrude-land-surface` for hints.
* MPI-OM 2d data needs::
cdo -setzaxis,surface IN OUT
to get rid of the layer-dimension that does not have any content.
* The reader cannot display distorted grids in the plane (e.g. most ocean models)
* The reader cannot handle a 2D field that has a depth axis (use ``cdo -setzaxis,surface INFILE OUTFILE`` or ``ncwa -a NAME_OF_DEPTH_DIMENSION INFILE OUTFILE`` to prepare files; affects a lot of MPI-OM output).
* Ocean models tend to give positive values for depth. That requires a couple switches in the reader, see :ref:`load-nemo`.
* For MPI-OM velocities, first use ``cdo -mrotuvb`` to get data to North/South East/West orientation, and interpolate u and v onto the cell centers.
* Nemo grids cause weird hick-ups.
* One file with many time steps tends to be a lot faster than many files with one timestep each.