Skip to content

Meeting 2026 02 04

Jump to bottom
Tom Kralidis edited this page Feb 18, 2026 · 18 revisions

Date/Time

Wednesday 04 February 2026 15h - 16h UTC

Location

MS Teams

Attendees

Name Affiliation/Country GitHub handle Attendance
Tom KRALIDIS (Chair) Canada/ECCC @tomkralidis X
Lance BRAASCH Scripps @lancebraasch X
Gao CHEN NASA @gaochen-larc X
Markus FIEBIG NILU @markusfiebig X
Jörg KLAUSEN MeteoSwiss @joergklausen X
Thomas LATTER WMO / OceanOps ---
Thinesh SORNALINGAM Canada/ECCC @thineshsornalingam
Fabrice THEBAULT MFI @fabricethebault
Alex VERMEULEN ICOS --- X
Anna MILAN WMO Secretariat @amilan17 X
Dave BERRY WMO Secretariat @david-i-berry X

Agenda (draft)

Notes

Summary

This meeting focused on the development of WMDR2 (WIGOS Metadata Record version 2), with presentations from Tom Kralidis and Jörg Klausen showing different approaches to the new metadata representation. The team discussed the balance between creating a Minimum Viable Product (MVP) for Oscar Next Gen and ensuring a future-proof standard. Key topics included data variable organisation, instrument handling, time resolution requirements, and the relationship between WMDR2 and WIS2 metadata records.

Key Discussion Points

1. Data Organization and Variables

  • Gao Chen suggested starting from data variables rather than stations, as this provides a powerful network perspective with spatial distribution snapshots
  • Concern raised that the current scheme buries variable information deep in the hierarchy (station → instrument → data product), making it difficult to navigate
  • Tom Kralidis clarified that datasets should be described using the WMO WIS2 metadata profile, while WMDR2 focuses on facilities

2. Technical Requirements (Alex Vermeulen)

  • Time resolution information is missing — needed for both the observation itself and for data aggregation/reporting
  • Context information needs timestamps, as context will change over time
  • Vertical distance should support multiple values (e.g. for profile measurements)
  • All IDs should preferably use UUIDs or PIDs for global uniqueness
  • Person identifiers should use ORCID
  • Organisation identifiers should use ROR
  • Units need to handle unitless measurements (e.g. PPM, PPB for concentrations/mole fractions)

Note: Alex agreed to post detailed comments directly to the GitHub issue tracker.

3. Jörg Klausen's Approach

Methodology:

  • Extracted all elements from the WIGOS metadata standard with requirements and code tables (available in the results/ section of the repo)
  • Used Oscar exports as source examples to maintain diversity and support future migration
  • Transformed XML files to plain JSON, removing GML overhead
  • Created field-level mappings between WMDR elements and OGC Records (including cardinality, history flags, code lists, link relations)

Key differences from Tom's approach:

  • Used a feature collection rather than a single feature
  • Used geospatial location history rather than temporal geometry
  • Included elements beyond mandatory requirements to support full migration
  • Omitted instrument/equipment block — instrument data is unknown for the vast majority of existing records

Work location: Pushed to a model-mapping branch in the dev repository.

4. Instrument Handling Challenge

  • Instrument information is unknown for the vast majority of WMDR records
  • The facility → instrument → deployment model is unlikely to work in practice
  • The link must be made to the observed quantity (a mandatory element) rather than the instrument (not mandatory), since that is what the standard actually specifies

5. MVP vs. Future-Proof Standard

Tom Kralidis Jörg Klausen
Position MVP first Future-proof first
Rationale Spec must be proven by production software; tools like QGIS won't open complex representations Building a bare minimum will cause problems down the road
Key point MVP must not block future expansion These are related but distinct tasks — Oscar MVP ≠ WMDR2 standard
Common ground Both agree work should be code-driven, not hand-written Both agree XML is a dying species

6. XML to JSON Migration Strategy

  • XML will not be exported by the new system
  • XML input will be accepted during a transition period but will age without active management
  • An XML → WMDR2 converter is needed to keep existing information manageable
  • Tom confirmed it will be straightforward to auto-generate a minimum WMDR2 record from a WMDR1 XML
  • Some edge cases in the mapping won't be known until implementation begins

7. WMDR2 and WIS2 Metadata Relationship

  • Markus Fiebig raised concern about managing the same information twice in different formats
  • Scope mismatch: a WIS2 data product record and a WMDR facility record do not have the same scope — this is a long-standing unresolved problem
  • Current best approach: bidirectional links establishing a many-to-many relationship between WMDR and WIS2 records
  • Markus preferred workflow: one WMDR facility record links to all relevant WIS2 datasets, with links continuously updated via automated harvesting
  • Tom noted this is possible depending on data center workflow, but not automatic

Repository Structure

Repository Purpose Governance
WMDR2 Formal document, formal examples, formal schema Pull requests must be linked to an agreed issue; requires ≥1 reviewer approval
WMDR dev Sandbox for development, experimentation, migration scripts Lighter-weight — no strict governance required at this stage

Timeline Considerations

⚠️ INFCOM-4 deadline: November 2026. Missing this deadline could push adoption back 2+ years.

  • The team has approximately 3 months to complete the work
  • David Inglis Berry stressed the need to balance completeness against the INFCOM-4 timeline
  • Anna Milan advised: provide feedback with solutions, not just questions — avoid debates that don't move things forward

Key Decisions Needed

  • How to balance MVP requirements with a future-proof design
  • Feature vs. feature collection structure
  • Temporal geometry vs. explicit geospatial location history
  • How to handle missing instrument information
  • Standard to align with: OGC Records vs. STAC vs. other
  • Deployment semantics — what does a "deployment" actually mean in this context?
  • Governance model for pull requests in the dev repository

Actions

  • Alex: post detailed technical comments to GitHub issue tracker (can also send via email or Teams)
  • Markus: open GitHub issue for WIS2/WMDR2 integration requirement
  • Tom/Jörg: prepare example showing a WMDR2 record alongside an associated WIS2 record
  • Anna: ensure all team members have access to WMDR dev repository
  • Anna: rename master branch in wmdr2-devt repository
  • Tom: update WMDR2 repo README to reference the dev/sandbox repository
  • all: continue discussion on Issue #3 in WMDR2 GitHub repository
  • all: review Issue #6 (known issues with existing WMDR) (ignore Oscar-specific items; focus on standard/representation issues)
  • all: provide feedback with proposed solutions, not just questions

Next Meeting

Clone this wiki locally