Impact of the OGC CIS 1.1?
A new, extended and improved version 1.1 of GMLCOV (now called Coverage Implementation Schema, CIS) is about to be released by the OGC. This new version will enable grids that have mixed regular and non-regular axis, and coverage partioning (a single coverage aggregated from multiple parts) among other things. I think the CIS 1.1 document has not been published by the OGC yet, but as far as I know it was proposed by the WCS SWG to the OGC Technical Committee be released during the OGC Nottingham meeting in September 2015.
It seems likely to me that we would need to add recommendations or TG requirements considering the coverage model and encoding in the WCS TG, so I think we should consider referring to the new CIS 1.1 rather than the old GMLCOV 1.0. I'm sure Peter Baumann can give more details on the release schedule of CIS 1.1 and the contents compared to GMLCOV 1.0, as well as the relation of the WCS 2.0 and CIS 1.1.
I know that at least in the meteorology use cases the mixed grids (like regular spatial + irregular height / time axis) would be very welcome possibility to use in their coverage-based INSPIRE datasets.
#1 Updated by Peter Baumann almost 5 years ago
I have added the current document + schema as https://ies-svn.jrc.ec.europa.eu/attachments/download/1229/cis_1-1.zip on the main MIWP-7b page. This is identical to what has been forwarded to ISO TC211 as well.
OGC Status: WCS.SWG approved it at the Sep OGC TC meeting, since then waiting for being processed by OAB. Once that is done, it will be rolled out for public RFC (request for comment). After that, TC vote can start.
Any questions, don't hesitate to contact me.
#4 Updated by Peter Baumann almost 5 years ago
A good question, Ilkka! You are addressing a formal weakness in OGC's procedures: they mention revision numbers of the documents in standards, which freezes dependencies in an undue way. I have complained about that at the last TC meeting.
In practice, however, things work the expected way. GML 3.3 extends GML 3.2.1, and despite other standards expressly reference GML 3.2.1 the common understanding is that they "work" with GML 3.3 as well. Same here: WCS relies on coverages, as defined in GMLCOV/CIS (to not create undue confusion we refrained from renaming the existing GMLCOV 1.0 into CIS 1.0). As v1.1 is a backwards compatible extension, WCS 2 can support this version as well, without any change. Hence, INSPIRE can continue working with the same WCS specification as before.
#5 Updated by Ilkka Rinne almost 5 years ago
Ok, so a compliant WCS 2.0 server can deliver it's coverage descriptions (DescribeCoverage) as well as the entire identified Coverages (GetCoverage) using either GMLCOV 1.0 or CIS 1.1 XML Schemas.
However for interoperability it might be a problem if we cannot agree on the coverage schema to be used (recommended?) in the INSPIRE context. Even if the CIS 1.1 is backwards compatible with GMLCOV 1.0, the schema namespace is different, and the some data providers may use data structures provided only in CIS 1.1 (like cis:GeneralGridCoverage). If some clients using this service only understands GMLGOV 1.0 based coverages, this would be problem.
#6 Updated by Peter Baumann almost 5 years ago
This versioning is a normal process, it can (and will) happen to any spec at some time. Yes, 1.1 has (necessarily) a different namespace. but I guess in practice this should work as each coverage contains its proper reference URL.
Re the problem that CIS knows more coverage types than a client may like: Already in GMLCOV this is the case, at least when it comes to the number of dimensions. OI, for example, will want to constrain to 2D Rectified Grid Coverages. Same with CIS: INSPIRE might constrain what a server is allowed to deliver when acting as an INSPIRE DOwnload Service. Further, already in the Capabilities a client learns about the type of a coverage and can choose wisely.
Just to note, getting away from Rectified vs Referenceable was initiated from our own suffering in getting Referenceable straight in EarthServer. The per-axis discrimination is suppossed to be much simpler (and, BTW, we have also done slight cosmetic changes that make, eg, coordinate parsing easier).
Hope that helps - but I am still newbie to INSPIRE so let me know if I'm missing something.
#8 Updated by Peter Baumann almost 5 years ago
Just coming back on this: I'd suggest to base on CIS 1.1:
- it covers all previously existing GMLCOV structures (just a new namespace - we had a discussion on this issue in OGC, and as a result OGC GML experts were satisfied with a solution where GML contents is identical up to the namespace)
- it will most likely be adopted by OGC + ISO, so becoming the Gold Standard
- and because of ISO adoption it will be stable for quite some time ;-)
just my 2 cents,