Re: [galeon] [WCS-1.2.swg] March 25 Teleconference Minutes

• To: "Whiteside, Arliss E (US SSA)" <arliss.whiteside@xxxxxxxxxxxxxx>
• Subject: Re: [galeon] [WCS-1.2.swg] March 25 Teleconference Minutes
• From: Ben Domenico <Ben@xxxxxxxxxxxxxxxx>
• Date: Wed, 25 Mar 2009 18:47:55 -0600

Arliss,
That's an excellent summary of what was a very informative and useful
telecon for me.  It touched on nearly all the major issues that arose as a
result of GALEON 1 experiments.

One clarification I would add is noted under item 4 in your summary below.

I am going to take the liberty of forwarding a copy of this to the GALEON
list because many of them will be interested and are not on the WCS SWG
list.  I look forward to even more fruitful discussions of these topics at
the OGC TC meeting next week.

Many thanks.
-- Ben

On Wed, Mar 25, 2009 at 5:59 PM, Whiteside, Arliss E (US SSA) <
arliss.whiteside@xxxxxxxxxxxxxx> wrote:

> March 25 Teleconference Attendees: (not a quorum)
> Steven, Keens, PCI Geomatics
> Arliss Whiteside, BAE Systems
> Wenli Yang, George Mason University
> Max Martinez, ERDAS, Inc.
> Peter Baumann, Jacobs University Bremen
> Ben Domenico, National Center for Atmospheric Research (observer)
> John Herring, Oracle (observer)
>
> Summary:
>
> As chair, Steven moderated this teleconference.  We discussed only one
> planned topic: Generalized Domain: allowing non spatio-temporal axes in
> domain.
>
> 1) Arliss noted that the email traffic indicated that the proposed
> generalized domain, although useful for other purposes, may not be
> needed for:
>
>   a) Allowing use of a pressure domain dimension, if the WCS domain
> supported the ISO 19111-2 extension allowing (only one) parameterized
> axis.
>
>   b) Allow a second time domain dimension, if the second time axis is
> represented in the coverage range.  Ben said he needs to consider this
> possibility further.
>
> 2) Steven asked if the domain axes need to be independent.  Arliss
> stated that the coverage domain is now defined by a BoundingBox in a
> referenced CRS, and that the axes of all CRSs must be independent per
> ISO 19111.
>
> 3) Ben stated that a pressure domain dimension would be independent of
> horizontal position (e.g., latitude and longitude).  This pressure is an
> atmospheric pressure, where its relationship to elevation (above sea
> level) or height (above ellipsoid) may vary some with horizontal
> position.  More significantly, sample points in a pressure domain
> dimension are not evenly spaced.  Ben said they also frequently use
> non-even spacings of grid points in the horizontal axes (e.g., latitude
> and longitude).  They also sometimes use point clouds, not grids.
>
> 4) Non-even spacing of grid points in any domain dimension is a WCS
> problem, for both a parameterized axis (per ISO 19111-2) and the
> proposed generalized domain.  Arliss said that non-even spacing seems to
> require WCS to define how to specify a grid coverage domain in a CRS
> with one or more dimensions without evenly spaced points.  One possible
> way would be to define a DerivedCRS that is used as the baseCRS of the
> coverage GridCRS.  Such a DerivedCRS would use a coordinate Conversion
> from its baseCRS, and a coordinate Conversion can be based on a table of
> corresponding values.  By definition, a coordinate Conversion is
> error-free.


Ben's note:  there was specific mention of a possible coordinate conversion
being from a baseCRS that is in index space to a spatial or parametric
coordinate where the baseCRS is the index into at table that gives the
values of the coordinate points.  The particular example we consered was the
list of unevenly-spaced pressure levels  that serve as the vertical
coordinate points for many fields in the output of forecast models.  This
example is important for the met-ocean community in two ways.  First ISO the
19111-2 extension apparently allows us to work with parametric coordinates
(e.g., pressure); second the DerivedCRS described above provides a
 mechanism for handling uneven grid spacing.


>
>
> 5) Interpolation between non-even spaced grid points in a pressure
> domain dimension would require special interpolation methods.  However,
> interpolation between grid points is often not desired by clients of
> such data, and that can be handled by specifying "none" WCS
> interpolation.
>
> 6) Max suggested that this SWG delay a decision on Generalized Domain
> until we find a clear need.  Many persons present agreed that this
> decision could be delayed.
>
> Arliss Whiteside
>
>
>
> _______________________________________________
> WCS-1.2.swg mailing list
> WCS-1.2.swg@xxxxxxxxxxxxxxxxxxxxxxxx
> https://lists.opengeospatial.org/mailman/listinfo/wcs-1.2.swg
>
>