Re: non-uniform grids pressure vs. time

Hi Jim,

The sfX/YCellBounds resources are not as general as one might hope yet.
They really only work for geodesic-type
grids. Your pressure vs. time grid is actually logically rectangular
and for now is best handled using the
regular ScalarField. If you set sfXArray and sfYArray using 2D arrays,
and use one more element than the
data array in X and/or Y, they become bounding coordinates. For
rectangular gridded data, this is the
way that works best currently.

I have attached a gsun version of the plot I created as a pressure vs
longitude example.
  -dave

On Dec 13, 2004, at 5:26 PM, James Boyle wrote:

> Using the geodesic.py example as a template I tried to write code to
> contour an irregular grid in time/pressure.
>
> The vertical coordinate is the CCM3 model grid in pressure space.
> Since the model is in normalized sigma coordinates, the boundaries of
> the
> vertical pressure cells vary with surface pressure which changes at
> each time. The times are uniformly spaced.
>
> from the geodesic.py code it would appear that all I need to do is to
> define is:
> sfXCellBounds
> sfYCellBounds
> sfXArray
> sfYArray
>
> to be able to produce some kind of plot. The sfXCellBounds are the
> corners of the time cells, sfYCellBounds are the corners of the
> pressure cells.
>
> This all seems very straightforward but when a run contour I get a
> great may errors of the form:
> .
> .
>
> unmatched node 1 3 59.986111 988.241821
> unmatched node 2 0 59.986111 988.241821
> unmatched node 2 3 59.986111 1805.201050
> unmatched node 3 0 59.986111 1805.201050
>
> .
> .
> What does this mean? It would seem that I am not defining the cell
> boundaries correctly - But there does not seem to be much room for
> variation.
>
> Is there any other documentation on how the cells need to be
> speciifed? I have looked over the MeshScalarField resources but have
> not found anything obvious that I am doing wrong.
>
> Presently, I am at a loss of what to try next.
>
> I recall that Dennis Shea posted some code to do this in NCL - but I
> have lost it. I think it used GSUN.
>
> Thanks for any help.
> --Jim
>
> _______________________________________________
> pyngl-talk mailing list
> pyngl-talk_at_ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/pyngl-talk

_______________________________________________
pyngl-talk mailing list
pyngl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/pyngl-talk