Additional semantics from OMS

[KathiSchleidt]

Under TSML, we're working on a CovJSON encoding. To avoid the old approach of pairing an O&M Observation with the Timeseries Coverage for additional semantics on the value being conveyed, we're proposing integrating these OMS concepts within the coverage (conceptually we use rangeType from 19123, on implementation we use an extension to the CovJSON parameters.

This approach entails adding the following attributes from the oms namespace:

• "oms:observingProcedure"
• "oms:observer"
• "oms:host"
• "oms:featureOfInterest"

Note on the duplication of the locational information between coverage domain and oms:featureOfInterest: in the coverage domain, we can only provide a point location. The target of the oms:featureOfInterest can then be a more fully fleshed out feature, including more detailed geo info (e.g. polygons).

[jonblower]

This looks sensible and valuable to me - and I think it's possible with the existing spec, assuming that the oms prefix can be registered. So is this issue essentially asking for that prefix to be registered?

Regarding the location information, you can use a PolygonSeries coverage type if you want to express that the coverage domain is a polygon. But I do appreciate that you might want to describe the feature of interest in more detail.

[chris-little]

@jonblower @KathiSchleidt We usefully discussed this on 2025-04-09. We think this would be compatible with similar changes to API-EDR, and we could probably support the attributes via a profile specified using API-EDR Part 3. We could also create a version API-EDR Part 1: Core V1.3.

I agree that registering the oms prefix is a good idea.

[chris-little]

@KathiSchleidt Where should the oms prefix resolve to please?

[KathiSchleidt]

Based on today's discussion, seems all we really need is an oms context file.

I've created a timeseries example that includes this context.

Only thing I haven't managed is how to do this including a namespace, any advice on how to do this (or should we just accept that namespaces are a relict from bygone XML days?)