[geotk] WKT and TemporalCRS

[Martin Desruisseaux]

Hello Aaron

Le 16/11/10 18:36, Aaron Braeckel a écrit :
> just following up on a conversation with Martin today.  I am building a 4-D 
> temporal CRS, writing it out to WKT, and trying to read it back in.  I need to 
> go both ways for my client/server app.  It sounds like the TemporalCRS is not 
> standardized, and there is not currently any WKT reading support for it.  I 
> can think of two options:
>
> 1) Decide on a format and implement TemporalCRS support
> 2) I change courses and use another standardized format (GML or something 
> else) for CRS representations
>
> Any thoughts?
While GML support is definitively on the radar, extending the WKT specification 
to support TemporalCRS is by far much easier than providing a full GML support 
in current Geotoolkit.org implementation. What we need however is to agree on a 
"standard" WKT specification for temporal CRS.

The current WKT specification is available there:

http://www.geoapi.org/snapshot/javadoc/org/opengis/referencing/doc-files/WKT.html

This specification defines horizontal and vertical CRS, but no temporal CRS.

Proposed approach:

1) Agree on a WKT specification for TemporalCRS on this mailing list.
2) Publish the proposal on GeoAPI pending.
3) Propose to the OGC to make the updated WKT specification part of GeoAPI.

What we need to agree on:

1) A standard name for TemporalCRS in WKT. I propose "TIME_CS" for consistency 
with the existing "VERT_CS" (WKT uses "CS" for what ISO 19111 call "CRS". This 
disagreement exists for historical reason and I think we should follow WKT 
tradition for new WKT element even if it doesn't match ISO 19111 naming).

2) Standards axis direction for the time axis. I suggest "PAST" and "FUTURE".

3) A syntax for the temporal datum. I suggest something close to the existing 
"VERT_DATUM":

TIME_DATUM["<name>", <epoch>, {,<authority>}]

where <epoch> is the Julian day of the temporal axis origin (the day 0 in the 
TIME_CS).

Example of what we may get, for a TemporalCRS where day 0 in January 1st, 2000 
at miday, and units are days:

TIME_CS["My Temporal CRS",
     TIME_DATUM["My Temporal Datum", 2451545],
     UNIT["day"],
     AXIS["Time", FUTURE]]

Any though on that?

     Regards,

         Martin