Web Coverage Tile Service SWG Roadmap
At the original ad hoc meetings in Washington, Frascati, Calgary and Tokyo, and at the initial formal meeting in Barcelona 2015, the following were repeatedly agreed:
The name stays as WCTileS, rather than Web Data Cube Service, to emphasise that it is a coverage service. It was agreed it was not really a "tiling" service.
We agreed to try an agile approach with delivery quickly and often:
- We need to agree how to record & promulgate meeting decisions & issues, in effect, a Quality Plan;
- We need to agree how to track & progress technical work what tools: JIRA, GitHub, MSWord, etc;
- Scope needs to be tight, including explicit statements of out of scope.
At least 3 approaches were originally identified:
- WCS2.0 as is, no need for WCTileS at all - rejected as does not address issue of massive scalability for mobile users;
- WMTS parallel compatibility - viewed as still too heavyweight, WMTS SImple Profile better, but consensus on WMTS approach maintained;
- Keep It Simple (KISS) with Google API - this is what the 'proof of concept' demonstrator used.
Some technical requirements were identified:
- Each tile should have a URI
- No perceived need for many tile sets.
- Requests and responses should be asynchronous
Expand and publicize Use Cases and likely Clients:
- Add a geological use case;
- Write critique of WMTS and WCS for envisaged Use Cases
- start with easy scalar variables such as orographic height, single level, single time
- then try vector or tensor variables, such as wind or current velocity
- The try multi-level, multi-time
List requirements, to conform to OGC Modular Spec policy.
Simple conceptual modelling to help with scoping, terminology, encoding requirements, etc.
Identify overlaps/common/interfaces to other Standards & stakeholders:
- SWGs: WMTS, WMTS Simple, DGGS, WCS, OWS Context, GeoPackage, etc; (GeoPackage and DGGS SWGs very interested, actively involved)
- DWGs: including Met Ocean, Aviation, ESS, Defence, etc. (Met Ocean DWG actively involved)
Which delivery formats to be chosen: CSV, JSON, KML, HDF, GMLJP2, etc ?
Prioritisation & roadmap to identify what delivered first & what later:
- Gridded data - already been demonstrated;
- Point clouds - similar, but more complcated as tiles can have different content 'sizes';
- Geometry - could be much more complicated, without clarity of definition of what is in a tile and what is in adjacent tile;
- Other data types.
-- Main.clittle - 26 May 2015