#	Model: Conceptual, Logical, XML	Issue	Date	Submitter
1	Logical, XML	Using Constituent as Feature-of-Interest	11-08-2015	B. Brodaric
2	Logical, XML	Linkage between GWML2 and TimeSeriesML	11-08-2015	B. Simons
3	XML	Change documentation to state that we used GML 3.3	12-02-2015	E. Boisvert

1. Using Constituent as Feature-of-Interest

Discussion:

OM_Observation featureOfInterest can be the samplingFeature (GW_Well) or the sampledFeature (GW_Aquifer). For some observed properties (e.g. pH) the former doesn’t make sense, so need to be able to distinguish the two. The featureOfInterest for the observations was mapped to the GW_Well (i.e. the samplingFeature), but it should be the ultimate featureOfInterest (e.g. the GW_FluidBody.

The Observations & Measurements specification proposes two patterns for featureOfInterest:

(1) the feature of interest can either refer to the real world object that is being observed, or

(2) to a sampling feature (SF_SamplingFeature) that acts as a proxy for the real world feature (which is then accessible through the sampling feature’s sampledFeature property).

Because the property featureOfInterest has cardinaliy 1..1, a data provider must choose one of these approaches. Problems occur when a sampling feature is chosen: while many observations can refer to the same domain feature, it is impossible to navigate back to the sampling feature. Solutions have been proposed using OM_Process to document the relationship between the observation and the sampling feature, but they are cumbersome and require much encoding. Consequently, the JSON (OGC 15-100) and RDF encodings for O&M include an extra samplingRegime property that is absent from the GML encoding. In GWML2, there is no restriction on featureOfInterest—it can refer to a real world feature or a sampling feature.

Solution:

Put in change request to OGC to modify O&M conceptual model and OMXML to have ultimate featureOfInterest (0..n) and samplingStrategy (0..n) for intermediate sampling features.

2. Linkage between GWML2 and TimeSeriesML

Discussion:

For GW_Flow, gwFlowVolumeRate is OM_Measurement: that constrains the result to be a gml:Measurement , so a single value with a uom but these may be ranges.

Additionally, the use of OM_Measurement precludes using TimeSeries as a result (uses a specialisation of OM_Observation). The GWML2IE did not investigate whether there were any properties that have data type OM_Measurement that could be a time series.

Solution:

Investigate whether there are any properties that require data type to be OM_Observation.

-- BruceSimons - 08 Oct 2015

-- BoyanBrodaric - 12 Aug 2015

3. GML 3.3 encoding

The encoding rule we used (we might want to check if it's all true) is actually GML 3.3 - specially the CodeType -> gml:ReferenceType. I always thought it was a INSPIRE thing, it's not. We must change the documentation and state that we used GML 3.3 (GML 3.3 is not a new schema, it's actually an extension of 3.2 + encoding patterns)

-- EricBoisvert - 02 Nov 2015

4. GeoSciML 4

• Pro
  • All optional properties - consistent with GWML2.0
  • On OGC standardization process
• Con
  • schema will change
  • instance will change
  • service will change