Re: Meeting at ESRIRuss Rew [address@hidden] -> direct acces s to netCDF and/or HDF5 data for their GIS tools

Russ Rew wrote:

> > To: address@hidden
> > From: "Haaring, P.A. (MIA-RIKZ)" <address@hidden>
> > Subject: Re: 20040712: Meeting at ESRI
> > Organization: National Institute for Coastal and Marine Management, The Hague, 
> > The Netherlands
> > Keywords: 200407121326.i6CDQNaW028420
> >    
>
> Hi Pieter,
>
>  
>
> > do you have any updates on your post in the netcdf-hdf mailing list
> > regarding "direct access for netCDF/HDF5 files in ESRI tools"?
> >
> > I would like to write a article on this subject for our internal GIS
> > newsletter and would like to know how if ESRI have made any formal
> > commitments about this.
> >    
>
> I did hear a bit more from someone else at the meeting about a later
> decision at ESRI to support access to netCDF data.  Here's a summary
> of the phone call I got on June 14:
>
>  This morning I got a call from David Maidment (a leading authority on
>  water resources and GIS, author of ArcHydro, and professor at UT
>  Austin) who had just been talking with Steve Kopp (a senior ESRI
>  developer) about results from a meeting we participated in at ESRI
>  last week, organized by Dr. Maidment.  At the meeting, I gave a
>  presentation on netCDF and had a separate hour-long discussion with
>  Scott Morehouse (ESRI's Director of Software Development and
>  "visionary on advancing the theory and practice of GIS" according to
>  the ESRI's "Modeling Our World" publication).  Jack Dangermond, ESRI's
>  founder and CEO, was also involved in some of the discussion.
>
>  According to Maidment, ESRI has since determined to add netCDF access
>  to their products, allowing them to read, view, and query netCDF data.
>  It's now on their immediate agenda, and may make it into an upcoming
>  release.
>
>  I had suggested that they handle georeferencing problems by specifying
>  that datasets must comply with particular netCDF conventions to be
>  accessible from ESRI products, such as the CF Conventions
>  <http://www.cgd.ucar.edu/cms/eaton/cf-metadata/CF-1.0.html> or the
>  somewhat wider set of conventions supported by John Caron's GeoGrid
>  class.
>
>  This will simplify integrating geoscience datasets with GIS systems
>  and is about the best outcome we could have hoped for, certainly
>  better than using a whole-file conversion approach, or using geoTIFFs
>  as a bridge format.  I'm fairly certain much of the desire to support
>  netCDF came from the Atmospheric Data Model Workshop in January 2004,
>  organized as part of the NCAR GIS initiative, so credit is also due to
>  Olga Wilhelmi, Terri Betancourt, and others who participated in that
>  workshop.
>
>  
>
> > I noticed also a move to GML encapsulation of netcdf in a recent
> > presentation from one of your colleges. John Caron in "NetCDF to
> > GIS: the Good, the Bad, and the Ugly" concluded "Possible solution:
> > extend GML to allow binary data, specify the georeferencing once in
> > GML. Recommend the simplicity of the netCDF data model"
> >
> > Is this the new direction unidata is taking? Or betting on two
> > horses?
> >    
>
> I forwarded your question to John Caron, in case he has comments.  I
> think he has been working on recommending netCDF as a form for binary
> data for some of the OpenGIS protocols, such as WCS.
>
> --Russ
>  

the gist of the AMS paper was that the actual data transfer is the weak 
link in the WCS spec; geoTIFF doesnt look like a long term answer. GML  
needs a way to deal with binary data to really work for scientific data.

I would say we will explore netCDF from a WCS server, but we havent 
decided to commit to it. we have beeen waiting for a demand for GIS 
services to develop in our community, and to see what form that will 
take. GML is a complex spec that would take a while for us to implement, 
but we have some colleagues who could help.

we are also waiting for ESRI to clarify exactly how they will proceed on 
their netcdf integration work; i would guess they are in the prototyping 
phase, and wont commit for a while. its seems unlikely to be WCS based.