Re: [galeon] Fwd: CDM feature and point types docs

Case 1 is related to the datum of observation. Case 2 is the sensor geolocation. Both of these are pertinent. I'm not sure if Case 3 is (A)the reprojected observation (or actual geolocation of the observation if different from Case 2) or (B) the location of actual processing. If (A), it's pertinent. If (B) it's a curiosity.
"Procedure" can well contain information on Cases 1 and 2, and 3(A). Case 3(B) isn't pertinent, unless I'm really missing something.
One thing to consider is that within a subset of domains, interoperability is realistic. Over the general solution set, it becomes really problemmatical when real use cases intrude. 

gerry

Simon.Cox@xxxxxxxx wrote:
An observation typically has several locations associated with it – e.g. where the world was sampled, where the instrument was (different if it is a remote sensor, or a lab instrument), where the data was processed to generate the result that is reported (may be different again) – all of which may be of interest in particular use-cases.
The OGC O&M standard reconciles this by separating the **feature-of-interest** and the **procedure** each of which may have location (e.g. a point). 

If you elide this detail then your observation model will not generalize.
That may be OK within your community or domain, but will probably hamper interoperability with other domains. 

Simon