Findings on O&M 1.0.0

I really like the Observations and Measurements model. Parts 1 and 2 offer a solid generic structure that captures the aspects of sampling, but it also has plenty built-in flexibility to accommodate the specific needs of practical applications. While the model and especially the documentation are far from easy to understand, once you have found your way it offers a powerful semantics without being overly designed.
Having said that I also feel this version 1.0.0 was almost ready for release, but not entirely so.
The biggest problem is that the UML models of part 1 and part 2 use a different version of GML than the Xsd’s do. It seems that this has introduced several issues.

Below are the remarks/findings I encountered so far
[list]
[*]The observedProperty of an Observation is modeled as a PropertyType in het UML model, but it is of type Phenomenon in the observation.xsd.[/*:m]
[*]There is a Process class in the UML model of part 1 and also in part 2. Unfortunately these have a different type in the xsd’s. In the observation.xsd it is a ProcessPropertyType, in specimen.xsd it is a ReferenceType. These can and should be the same.[/*:m]
[*]The samplingTime attribute is modeled differently in part 1 and part 2. It is of type TM_Object in the Observation class (swe:TimeObjectPropertyType in observation.xsd), and it is of type TM_GeometricPrimitive in the Specimen class (gml:TimePrimitivePropertyType in specimen.xsd). These can and should be the same.[/*:m]
[*]SamplingFeature is modeled as a specialization of FeatureType. The downside of this is that a feature that already exists in its own inheritance tree can not also be a SamplingFeature.[/*:m]
[*]Although SurveyProcedure is described as “Informative” in the documentation of part 2, it is used as a referenced type from SamplingFeature and thus it is in the standard.[/*:m][/list:u]

All in all I can recommend looking into this model to everyone who is active in this field.

[/list]

creed's picture

Findings on O&M 1.0.0

And this back from the editor of the OGC O&M document.

thanks.
I was aware of all of these, but good to see that someone is reading closely.

On the issue of "a feature that already exists in its own inheritance tree can not also be a SamplingFeature" I would point out that
(a) any feature type from any domain schema can be used as the feature-of-interest of an observation
(b) the sampling feature taxonomy is provided as a convenience to users, and there is no compulsion. These types are found in broad use already, so this is just a formalization of current best practice, though with harmonized terminology, and with some features added to make it consistent with the O&M model.

---
creed's picture

Findings on O&M 1.0.0

From the editor of the OGC O&M standard:

- thanks.
I was aware of all of these, but good to see that someone is reading
closely.

On the issue of "a feature that already exists in its own inheritance
tree can not also be a SamplingFeature" I would point out that
(a) any feature type from any domain schema can be used as the
feature-of-interest of an observation
(b) the sampling feature taxonomy is provided as a convenience to users,
and there is no compulsion. These types are found in broad use already,
so this is just a formalization of current best practice, though with
harmonized terminology, and with some features added to make it
consistent with the O&M model.

---
creed's picture

Findings on O&M 1.0.0

Apologies for the delay in responding.

Thank you for your comments regarding O&M. I will pass this information onto the O&M Editor. FYI, O&M is now a new joint work item activity with ISO TC 211. Any errors or inconsistencies will be corrected as part of that activity.

Regards

---