NOTE: The galeon mailing list is no longer active. The list archives are made available for historical reasons.
Keiran- it is exciting for us WCSers to get a clearer pic on what is actually wanted when talking about irregular meshes. Can we somehow start collecting requirements in an ordered manner, say a Wiki page? What I am most interested in: - data structures: all information about what you expect (eg: general TINs vs warped grids) - operations: you mention subsetting, what about others? reprojection I guess? Note that WCPS (Web Coverage Processing Service), a WCS extension, allows processing of coverages, and extending this in parallel makes me ask deeper here. - formats: NetCDF, of course ;-) ...any other formats wanted by the communities? - "anything else" that contributes to the scenario - in the end, some use case scenarios IMHO serve best. We might start with a matrix for the communities (is metocean one or several in this respect?) and gradually massage it into one requirements list for presentation to the WCS group. @Ben: as the WCS Twiki is not open to all on this list, could we have some place with r/w access for all GALEON people on the GALEON Wiki? -Peter Keiran Millard wrote:
Dear Jon, I have been watching this discussing unfold over the last week, a relating back to use cases I am familiar with. Early discussions have asked "what is a WCS use case?" and also by implication "what is a WFS use case?" - indeed (IMHO) high level convergence between the variousOGC specifications is needed to address this.My concern comes from the perspective of the coastal community and the discussion on 'lowest common denominators' focussed on serving 2D regular grids. Datasets based on 2D regular grids are increasingly obsolete technology in the coastal community and so technology that does not support serving unstructured meshes is unlikely to be adopted. The typical coastal community has to deal with a range of coverage types; single point time series, profiles, grids and meshes. I would argue that the coastal community has requirement for simple 'bulk transport' of data across a range of "coverages" - points, grids and meshes. Any service that delivers this needs to support subsetting; but possibly not coordinate conversion. The other requirement is for extensive metadata about the measurements. At this present time we deliver a single point timeseries coverage through a WFS. It works, but certainly not scalable to other coverages. However, it needs to be considered that really what I am being served is a feature that describes the observation process {O&M / CSML territory), plus the result of the observation as a bonus. So in summary, agreement on what's in and out of scope for both WCS and WFS is needed. Keiran
galeon archives: