Hi,
The topology, layout and inventory service APIs are the best place to get resource and relationship data once their routes are enabled as described in the docs. The Observer's are ingest only.
On your modelling goal - I'd highly recommend treating them as a group instead if there's a likelihood of a given location having many resources, i.e. a supernode. I would use a token template to do this that uses a property on the resources and if it doesn't exist natively, use a file enrichment rule or file observer-and-merge to provide it.
thanks,
Matt