Hi all,
This is a early "heads-up" on some outputs of a data modelling workshop I recently participated in that provided an opportunity to explore some of the relationships between "Simple" and "General" Features and Coverages, and the underlying phenomena being modelled.
A communique is being prepared with the position reached, but I see there is various discussion and effort being expended in areas of relavnce now, and I wanted to put forward some of the ideas to see if they prompted any discussion.
1) A coverage is a Feature describing the state ("range") of a spatially distributed phenomenon In many (if not all!) is in fact some sort of special case of an Observation Feature where the procedure is a remote sensor, a chain of models or a sampling regime.
2) a WPS may act on Feature Types
So the (usually contentious!) statetment #1 becomes a "no-brainer" when seen in the context of #2
This also allows us to coherently model a domain irrespective of whether we collect or package the data as features or coverages.
Other corollaries include:
* A WCS is really a convenience API for the general WFS, where the FeatureTypes served are coverages
* A WFS can equally serve a coverage feature
* A coverage feature can link the range to a streaming protocol to serve values - e.g. JPIP
A more profound outcome is that FeatureTypes are polymorphic - and the "representation views" of a "conceptual model" can be described in terms of the "operations" that can be done - Simon Cox has dubbed these "processing affordances". This is not that profound in some ways, since ISO19110 clearly states this, but in practice we have lacked mechanisms, and most domain models have been driven by inconsistent approaches - geometry-centric, semantic, processing-oriented aspects governing the typology. We can now resolve all these we beleive in a single methodology, that allows you to have typical "shapefile" representational views without crippling the underlying conceptual model of the domain (for example losing all the metadata for an observation).
If this leaves you cold, fine, you probably have minimal actual interoperability requirements, but hopefully this will allow a few of you to find some coherence in a range of activities which are being inexorably drawn together, and prime you to read and respond to the forthcoming findings.
Cheers
Rob Atkinson