Hi John,
I agree with Jonathan and Karl that the current conventions are sufficient
to describe the WRF grids.
I don't see that the proposed stagger attribute adds any new functionality.
> The stagger attribute associates 2 coordinate axes, indicating that they
> represent the same coordinates, but are offset from each other. A
Why not use standard names to make these associations? For example, the x
and x_stag axes could each have the standard name projection_x_coordinate.
Then determining the stagger is a simple matter of inspection of the
coordinate values.
> and 6 new 4D vertical coordinates (possibly only 3 if they dont need to
> keep P and PB seperate) (in principle you might need as many as 16 4D
> variables).
You mentioned in an earlier post (27 Oct 2003) that the pressures are
defined on the unstaggered grid. In that case the pressures on the
staggered grids are obtained by interpolation and it wouldn't be necessary
to include them in the file. Also, as you point out it's probably only
necessary to include the total pressure, and not the base and perturbation
parts.
Your previous use of formula_terms to indicate that the base and
perturbation parts should be added to obtain the total pressure was not
consistent with the intent of that attribute. It is designed to provide a
formula to calculate a dimensional quantity like height or pressure from
the dimensionless coordinate values and other data like surface height or
surface pressure.
Brian
On Tue, Jan 27, 2004 at 04:47:18PM -0700, John Caron wrote:
> Jonathan Gregory wrote:
>
> >Dear John
> >
> >If I have understood this correctly, a CF-compliant file would have to
> >include
> >the variables latForU and lonForU, because it is mandatory to supply 2D
> >lat and
> >lon variables if lat and lon are not axes. These variables should be
> >listed in
> >the coordinates attribute of U, so the data of U can be located in latitude
> >and longitude. Isn't it OK to do it in this straightforward way?
> >
> >There would then be no relation shown in the file between the two grids. My
> >point is that this relation (the staggering) can be deduced by inspecting
> >the
> >coordinates. Given that this is the case, I don't see the need to
> >introduce any
> >new convention to indicate it explicitly, because this would be redundant.
> >
> >Best wishes
> >
> >Jonathan
> >
> >
> >
> Yes, if you are willing to explicitly include latForU and lonForU, you
> wouldnt need the staggering.
>
> To be explicit, for WRF, you will need 4 extra 3D variables (in
> principle you might need as many 8 3D variables).
>
> float lat2(t, y, x_stag);
> float lon2(t, y, x_stag);
>
> float lat3(t, y_stag, x);
> float lon3(t, y_stag, x);
>
> and 6 new 4D vertical coordinates (possibly only 3 if they dont need to
> keep P and PB seperate) (in principle you might need as many as 16 4D
> variables).
>
> float P2(t, z, y, x_stag);
> float PB2(t, z, y, x_stag);
>
> float P3(t, z, y_stag, x);
> float PB3(t, z, y_stag, x);
>
> float P4(t, z_stag, y, x);
> float PB4(t, z_stag, y, x);
>
> It seems to me like a lot of overhead. But the real problem I think is
> that it obscures the relationship of the coordinate systems, making it
> more difficult to combine the staggered U,V,W fields into a vector wind.
> Its quite a simple and common relationship, deducing it is non trivial
> and likely wont be done automatically.
>
> Im willing to tell the WRF people they need to do that, but if they dont
> want to i guess they will have to create their own Convention.
>
>
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Wed Jan 28 2004 - 09:38:40 GMT
