Re: [wcsplus] WCS 1.0+

Hi all,

Since I started the conversation off the list, thought I would move it to the list with a summary of the conversation so far. [And a reminder for those not yet on the list, the URL to subscribe is http://www.unidata.ucar.edu/support/mailinglist/mailing-list-form.html#subscribe.]

Ethan

Ethan wrote:
Here's the main part of the list for WCS 1.0+:

   * Allow multiple fields in a coverage range
   * Add asynchronous response handling
   * Use GML with CR that includes CV_ReferencableGrid (see OGC 07-112)
   * Move towards 1.1 style handling of binary response encodings
(postpone work till later, let CF-netCDF profile work run separately)

A few implementation details I would suggest for this effort:

   * Be more explicit about HTTP response codes
   * Use GET KVP only

Needed enhancements for the current TDS WCS 1.0 implementation:

   * Handle rotated poles
   * Handle British National Grid (Transverse Mercator)?
   * Improve handling of vertical levels and time range (currently
     limited to choosing one level or time point?)

And planned/hoped for implementations:

   * Servers: RAL and Unidata
   * Clients: GI-GO and ITT Vis (IDL) [Ben will contact ITT Vis]

Two additions to the server/client list:

[Dominic] RAL has a python client/library (currently aimed at 1.1 but likely 
for 1.0+ as well)

[Stefano] Also have Java server/library targeted at 1.1 but will experiment 
with 1.0+ as well)


Dominic asked about how to proceed in terms of a document:
How do people want to proceed from here to flesh this out?
Word/Google Docs/Wiki? (I think we talked in terms of not making it public till we have something more substantial to make public).

Jeremy wrote:
All, from Met Office perspective, I think Ethan captured the salient
points. I've copied Bruce Wright in on all the details (he'll be back
from vacation tomorrow) so that we can contribute effectively to this
endeavor. I've not forgotten that I promised to circulate my views on
the alignment of ISO19123 / Scientific Data Types / CF ... I'm aiming to
circulate that this week.

Stefano had some additions:
As to the document, in my opinion, there are some points to be made explicit for the sake of interoperability -after having reached a general consensus, of course. For example:

_Abstract level_:
a) The WCS 1.0+ coverage model implements the ISO-19123 specification

_Content level_:
a) The WCS 1.0+ coverage domain is 1-4D (x,y,z,t); optional extra dimensions are considered as co-domain fields; b) The WCS 1.0+ coverage domain is characterized by a fixed shape (i.e. x,y,z,t); ??
c) The "z" dimension may be either upper or lower oriented;
d) The "z" dimension may different from sea-level altitude; ??
e) The CRS metadata consists of a unique reference code defined in the EPSG namespace; ??

Encoding level:
a) The WCS 1.0+ DescribeCoverage metadata are encoded using GML version ??? b) The WCS 1.0+ GetCovereage answer may be encoded using at least one of the following well-known formats: CF-netCDF, GeoTIFF, HDF, ... ??? As we agreed, the objective is to try to keep the WCS 1.0 simplicity and effectiveness, enhancing it with few extra features in order to make it work better for Earth system science applications.

Thus, in my opinion, the main objective is to decide which is the best trade-off between generality/powerfulness and simplicity/feasibility. Indeed experts on specific domain systems have this knowledge and my guarantee the real implementation of the spec.


--
Ethan R. Davis                                Telephone: (303) 497-8155
Software Engineer                             Fax:       (303) 497-8690
UCAR Unidata Program Center                   E-mail:    edavis@xxxxxxxx
P.O. Box 3000
Boulder, CO  80307-3000                       http://www.unidata.ucar.edu/
---------------------------------------------------------------------------