Observable Properties Register Design

This page collects inputs and emerging design ideas for a register of observable properties.

Design goals

1. observable properties to have canonical URI identifiers that can be cited in Web context, including JSON-LD and RDF data forms.
2. Compatibility with SOSA and SWE standards
3. existing identifiers used in external systems to be preserved
4. meet informational requirements of key application domains - such as WMO and GEOSS systems
5. a register of registers with delegated governance to communities of practice on a per register basis

Proposed Solution

• a canonical data model as an RDFS class model
• a django plug-in to the OGC Definitions Server (as per DGGS registry), mapped to the RDF model
• batch import of key exemplars
• reverse engineer participating project observartion parameter into sub-registers

Candidate ontology

• TBD

Related work:

• http://def.seegrid.csiro.au/static/ontology/ogc/property/ (need to get a handle on scope here)

Definition Sources

Essential Climate Variables

• overview: https://public.wmo.int/en/programmes/global-climate-observing-system/essential-climate-variables
• example: https://www.ncdc.noaa.gov/gosic/gcos-essential-climate-variable-ecv-data-access-matrix/gcos-atmosphere-upper-air-ecv-earth-radiation-budget
• example SKOS implementation : http://vocab-test.ceda.ac.uk/ontology/gcos/gcos-content/gcos-ontology.ttl (does not express technical details)

OSCAR (WMO)

• overview: https://www.wmo-sat.info/oscar/variables
• example: https://www.wmo-sat.info/oscar/variables/view/161

others TBD

Notes

The OSCAR example has "requirements" per variable, with thresholds for different classifications of results, on a per-application basis. This seems quite powerful and should be achievable - its not too complex a data model.

Variables bound to nomimal values (e.g. species taxon) need a reference mechanism - this is perhaps best done using the RDF-QB model of a measurement component, allowing SKOS ConceptScheme in conjunction with rdfs:range

-- RobAtkinson - 10 Oct 2018