Re: NetCDF advice requested

Hi Tommy,

Sorry I've taken so long to deal with your question.  It sort of fell
through the cracks, and I just noticed today that it hadn't been
answered yet.

For representing georeferenced gridded data in netCDF, we have some
examples that we worked out a couple of years ago.  These use a dummy
dimension "nav" and the convention that every variable that uses the
"nav" dimension is specifically for georeferencing.  If there is only
one georeferencing in a netCDF file, the nav dimension is of size 1, but
it could be larger for multiple sets of variables, where each set uses a
different value of nav to describe its georeferencing. 

Some examples of this convention are in the directory

  ftp://ftp.unidata.ucar.edu/pub/netcdf/cdl/

for example ruc.cdl representing the RUC/MAPS model on a Lambert
Conformal grid.

I wrote a paper about another approach to georeferencing:

 Rew, R. K., 1998: <a
 href="http://www.unidata.ucar.edu/staff/russ/MapGeoGrid/ams98.html";>
 Georeferencing with Java: An Example of Executable Metadata.</a>
 <em>Proceedings, 14th International Conference on Interactive
 Information and Processing Systems for Meteorology, Oceanography, and
 Hydrology,</em> American Meteorological Society, January, Phoenix,
 Arizona.  The slides used in the presentation, and the source code for
 the prototype, are also <a
 href="http://www.unidata.ucar.edu/staff/russ/MapGeoGrid/index.html";>available
 online.</a><p>

Of course there are other georeferncing conventions worked out by other
groups, e.g. the COARDS conventions.  For more of these, see the netCDF
Conventions page at

  http://www.unidata.ucar.edu/packages/netcdf/conventions.html

--Russ

>>>>> "Tommy" == Tommy Jasmin <address@hidden>
>>>>> wrote the following on Mon, 02 Nov 1998 15:33:09 +22304808

> guys -
> as you may know, I recently wrote a couple of NetCDF related applications
> here at SSEC.  one, a Java-based McIDAS area to NetCDF converter that
> works (well) only on GVAR data.  the other, a more general ADDE server
> in C that works on any data type, and utilizes the McIDAS library to do
> calibration on the fly.

> one big concern and ongoing source of debate is that there is no standard
> for representing geographic image data in NetCDF format that I know of.
> Tom Whittaker would like me to take a 1st shot at proposing something.
> I was just wondering if you could give me any general advice, references
> for information, etc.

> a specific question I have is whether there are examples of CDL where
> certain variables or attributes are present conditionally based on
> the data source.  the reason I ask this is because I don't want to 
> assume the data source is a McIDAS area file, and if there are different
> possible input formats, then there may be some variables or attributes
> that are only relevant to a certain format.
> -------------------------------------------------------------------------
>           *    |    *    /  +
>     +        --*--     /          Tommy Jasmin
>        |   +   |   +  /  +   *    Interactive Systems Development Group
>      --*--          /             University of Wisconsin
>        |     *     /     |        Space Science and Engineering Center
>     +            /     --*--      address@hidden
>           *     /        |        http://www.ssec.wisc.edu/~tommyj
>                *                    
> -------------------------------------------------------------------------