[Gdal-dev] Re: GDAL netcdf convention CF support

[Frank Warmerdam]

On 8/21/05, Gerry Creager N5JXS <gerry.creager at tamu.edu> wrote:
> Frank, et al,
> 
> I think we need to be thinking either about making GDAL more interested
> in 3D/4D datasets, as that's coming whether we're ready or not... in
> fact, for me, it's already here.. and then see if we need to consider
> forking or augmenting Mapserver to accommodate it.
> 
> I see a lot of conceptual problems for Mapserver, and, for that matter,
> a LOT of GIS packages in incorporating 3D/4D data, and presenting it
> properly.  This has been an issue for GIS practitioners since just about
>   forever: They think in 2 dimensions and can't, as a group, appreciate
> the need for either vertical, or time.  They represent time series data
> as a series of static images.
> 
> I see, in Mapserver, a potential interface.  It's open-source and thus
> maleable for applications like this.  It won't be simple or
> straightforward, and in the end, Mapserver might serve solely as an
> example framework for a new 3D/4D display/dissemination tool, but it
> *is* a starting point.
> 
> I'm kinda thinking about a melding of Mapserver and IDV, in th elong run.

Gerry,

The idea of extending mapserver and GDAL to better support 3D/4D
datasets has been brought up a few time, and I have generally been
quite ... conservative ... in response.  I am concerned about substantial
complication and turmoil in the interfaces of both.

My personal opinion is that in the shorter term it would make more
sense to have a distinct multi-dimensional WCS server implemented
that might derive some code from MapServer, but that would otherwise
be quite distinct.  It would ignore all the vector and "portral" related support
in MapServer in favor of focusing on serving n-dimensional raster data 
via WCS.  

There could be an n-dimensional equivelent to GDAL as well, that might
even include a GDAL plugin for 2D datasets, but that would have it's
own interfaces that are more organized to n-dimensional data.  Perhaps it
could even be closely related to one of the existing n-dimensional file
format libraries like netcdf or HDF (my understanding is that the next
general netcdf is based on HDF5). 

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent