Wenli Yang wrote:
Hi John et al,
You Asked:
Can anyone who knows WCS 1.2 summarize 1) whats in the core, and
2) what is allowed in the extensions?
The core/extension discussions are documented in the following
page/document:
http://portal.opengeospatial.org/twiki/bin/view/WCSrwg/DivideSpecificationIntoBasePlusExtensions
http://portal.opengeospatial.org/files/?artifact_id=22874&version=1
Unfortunately one needs to be an OGC member to read them. I try to summerized
the key capabilities of core and extension bellow.
Core capabilities:
a) support 2D spatial/temporal domain, essentially the two (near) horizonatl
spatial dimensions, which are most commonly lat/lon coordinates but
can be any other projected coordinates;
b) support one 2D range field (i.e., one variable with 2 dimensions);
c) do not have to support subset of range field, i.e., do not have to
support subset of non-spatial/temporal axes of a variable, which may be
2D array; and
d) Support nearest neighbor interpolation.
As I have stated previously in the WCS SWG, I think limiting the core to
2D is a problem. There is no concept of a "basic" 2D coverage described
in the approriate ISO spec that then has more complicated variants. A
coverage is described as a fairly N-dimensionalish concept, and I think
the core WCS specification should address this. If a general mechanism
for axis and range subsetting is identified, then most of the complexity
of dealing with time and other more full-featured types falls only to
the communities that need to represent things that way in their data.
Implementation complexity scales with need. The dimensionality of your
coverages just need to be advertised as part of describeCoverage, and
client and server implementations that need to deal with more complex
data provide the necessary capability.
This way you have put your faith in the coverages model to be able to
represent the various types of data (which seems like a valid assumption
for a Web Coverage Service) rather than deciding a-priori that the
easiest thing for most communities to handle is a particular sub-type of
coverage when it is not uniformly easy or natural for every domain.
Most scientific domains have a use for representing time and altitude
and it is more difficult for them to leave them out.
I have yet to report back to the WCS SWG about a comparison in
complexity of the two approaches, but I feel that the WFS spec has a leg
up in maturity on the WCS spec at least partially due to the generality
by which they deal with their filtering capabilities. The need for
larger numbers of extensions go away when this is resolved.
