Re: trying to load a 2D time series

Hi Eric-

Eric Davies wrote:

I'm trying to load a grid representing x, y, t from a data base.
I used to do this by creating a Gridded3DSet in which I'd duplicated the x y planes to form a second set (various display functions complained when there was only one z) since I had to handle x, y, z, t
grids as well.

This approach stopped working when the visad library started checking the grids for "correctness". (I suspect numerical precision since the visad test only used floats for its calculations).

The correctness code has not been released except as a test and is only
in place for the Gridded2DSets.  Did you update your visad.jar for the
IDV?  You indicate below that you are using the 1.2b2 IDV source and
that would not have an update visad.jar.

So I tried adding some extra logic to generate Gridded2DSets, ie, doing it the clean way. The coordinate system is a subclass of LambertConformalConic.

When the FieldImpl.toString() is printed, the text is:
FieldImpl
  Type: (Time -> ((x, y) -> Maximum_temperature))
  DomainSet:
    visad.Gridded1DDoubleSet: Dimension = 1 Length = 1
      Dimension 0: Length = 1 Range = 1.1256192E9 to 1.1256192E9
  Range value 0:
    FlatField
      Type: ((x, y) -> Maximum_temperature)
      DomainSet:
        visad.Gridded2DSet: Dimension = 2 Length = 78624
          Dimension 0: Length = 312 Range = -2716550.5 to -1136855.1
          Dimension 1: Length = 252 Range = 756974.0 to 2031905.0
      RangeSet[0]:
        FloatSet: Dimension = 1
                ....

Now, the grid doesn't show up on the visualization pane, instead there is a line displayed on the visualization pane that reads:
   domain must be only spatial
in a vector font.

IDV appears to be able to display a GRIB file that has x, y and t dimensions, so it must be possible.
What am I doing wrong?

I'm not clear on what you are doing in your grid construction.  Could
you send a code snippet of the old and new?  Can you just create a
Gridded3DSet with a 2D manifold instead of repeating the values?

I've tried recompiling the IDL source so I could step through the code to see what the GRIB code is producing, but the compiler is complaining about missing methods. I'm using idv 1.2b2.

Any way you could try this with the 1.2 code?

Any pointers appreciated.

Send me some code snippets of creating the grid and then
we can go from there.

Don
*************************************************************
Don Murray                               UCAR Unidata Program
dmurray@xxxxxxxxxxxxxxxx                        P.O. Box 3000
(303) 497-8628                              Boulder, CO 80307
http://www.unidata.ucar.edu/staff/donm
        "Time makes everyone interesting, even YOU!"
*************************************************************