Meeting 2026 04 15 - wmo-im/tt-wismd GitHub Wiki

Date/Time

Wednesday 15 April 2026 13h - 14h UTC

Location

MS Teams

Attendees

Team

Name Affiliation/Country GitHub handle Attendance
Tom KRALIDIS (Chair) Canada @tomkralidis X
Guillaume AUBERT EUMETSAT @gaubert X
Takumu EGAWA (observing member) Japan
Lara FERRIGHI Norway @ferrighi X
Sylvain GRELLET France @sgrellet
Xinqiang HAN China @Amienshxq
Vera KORBULAKOVA Russian Federation @VeraKorbulakova
Ian MCDONALD EUMETSAT @McDonald-Ian
Yusuke NAKAGAKI Japan TBA
Steve OLSON USA @solson-nws
Ján OSUSKY HMEI @josusky X
David PODEUR France @davidpodeur
Julia SIELAND Germany @jsieland

|Anna MILAN|WMO Secretariat|@amilan17|X

Agenda

Notes

  • Tom opened the meeting with a welcome to Yusuke NAKAGAKI as a new task team member, taking over from Yutaro TONOOKA.
  • The team reviewed previous actions.
  • Secretariat updates (Anna): The fast-track amendments for FT2026-1 was circulated to national focal points. Comments are due by 8 June. Responses received so far have been positive (acknowledgements, no objections).
  • The team discussed the following:
    • Open PR on editorial changes to the KPIs: Anna ran Grammarly over the document and made editorial/spelling/grammar changes without altering meaning. Two specific points arose: 1) Commercial footnote link: A footnote citing "88% of online consumers" pointed to a .com site. The group agreed this was inappropriate for a WMO/UN manual. Tom removed the footnote during the meeting. 2) Table of contents in Word/PDF outputs: Anna noted the auto-generated Word document does not produce a nice table of contents from the Markdown source, unlike the online version. No action required but noted for awareness.
    • Daily KPI link checking: Tom proposed that running link validation daily is acceptable; the concern about "annoying" upstream servers by checking links every day is not significant given that the GDC runs checks locally on its file system replica. Issue 2298 is closed. The nightly re-run approach is retained. Lara raised a useful nuance: static metadata elements (e.g. title, description) don't change unless the record is updated, so daily re-running may be redundant for those. Tom acknowledged this, but agreed the cost of implementing differential logic would exceed the cost of just re-running the full set locally. Lara noted that it would be valuable that KPI link checks should have deduplication implemented so that identical links are not tested more than once in a KPI check/run
    • Metadata dashboards: Tom opened a discussion on whether the task team should define more detailed metadata quality dashboards beyond the existing high-level KPIs visible in the Global Monitor. Guillaume expressed strong interest; he described MET France's use of Elasticsearch/Kibana for monitoring service statistics, noting that ops teams had extended initial dashboards in unexpected and useful ways. Lara described MET Norway's internal panel-based notebook tool for checking URL integrity and metadata quality (not yet in production). Tom referenced NOAA work that sliced and diced WIS records in useful ways. Anna suggested a clickable UI to browse keywords and associated records as a valuable addition.
    • KPI for Themes and Concepts / Keywords: Anna raised the absence of a KPI for keywords/themes. Free-text keywords are inherently non-normalised; measuring them as a KPI has limited value. Tom noted they are organic tags — useful for market analysis but not governing behaviour. Lara argued that a KPI should reward use of controlled vocabularies via the themes element (which carries a URI), since those entries are interoperable and mappable across vocabularies. Anna suggested a simple KPI: "has at least one additional theme/concept beyond the mandatory topic hierarchy discipline. Tom proposed a scoring approach analogous to the existing links KPI — extra points per additional theme/concept found beyond the required minimum, turning it into a percentage score. The full keyword tag cloud and frequency analysis would live in the dashboard exercise rather than in a formal KPI.
    • Monitoring Topic Hierarchy usage: Guillaume asked whether there is any monitoring of which topics are being subscribed to and how many messages are published per topic. He noted the hierarchy has grown very large and usage data would help the team decide where to trim or expand it. Tom agreed this is valuable and described the Global Replay Service (Canada running it, currently truncating to the last 72 hours for notification messages; monitoring messages retained for approximately one week). The Global Replay API: Takes all WIS2 notification messages and inserts them into an OGC API Features endpoint, adding the topic as a property. Also supports MQTT replay: the requester specifies a date range, subscriber ID, and topic; the service replays past messages via MQTT using the same interface as normal subscription. Tom shared the link in the meeting chat (publicly available). Guillaume will share the link with his team and will explore potential use cases. Anna also raised whether we could analyse appropriateness of topic use (e.g., are publishers using topics correctly, or should some data be published to a different/new topic?). Tom acknowledged this as valuable but difficult to operationalise. Anna and Guillaume suggested cross-referencing keywords in discovery metadata against the topic being used as a proxy check. Noted as something to explore further once the W2IT issue is open.
    • Public Forecast / Text-Based Forecast Topic: Tom presented the pull request he opened adding a topic for text-based (human-readable/narrative) weather forecasts. Current status: the PR exists under the predictions subtree; the main open question is the name of the leaf node. The options discussed were "human-generated forecast" — Tom's initial proposal but Ján had concerns; "plain text" / "textual" — Ján's suggestion; "text-based forecast" — NOAA's usage; "text forecast" — but "forecast" already exists as a parent. Anna questioned whether a further sub-categorisation is needed at all, noting the existing temporal qualifications (short-range, medium-range, etc.) already exist under predictions. Tom clarified that these forecasts do come from prediction systems — either a human forecaster writing them or an automated text generator applied to model output — so they fit under the existing temporal nodes. The question is whether a leaf node at the same level as deterministic/probabilistic is needed for textual representation. No consensus reached in meeting. Agreement to bottom out the name with Ján via the GitHub issue. Tom will note in the issue that there was limited discussion in the task team on the name specifically, and that "plain text" may be a lesser evil but further input is welcome.
    • MetOp-SGA1 (Pre-Operational Satellite Identifier): Anna raised a request (from Simon) to add a topic for MetOp-SGA1, the pre-operational designation for a not-yet-operational satellite. The operational name (MetOp-D) already exists in the hierarchy. Anna's concern: is this a pattern that will recur for every new satellite (all of which have different pre-operational names vs. operational names)? She noted that the requester was aware of the experimental topic but the request was described as "more political than anything else. Guillaume confirmed this pattern will recur for EUMETSAT satellites — pre-operational and operational names regularly differ. He supports the request but acknowledged its political dimension. Tom noted that users who know the pre-operational name can filter it on their client side. No decision reached. Tom will leave the issue open; the team will await the requester's response on whether the experimental topic is sufficient or the addition is needed. Anna will follow up in the issue and with the requester about the experimental topic as an alternative.

Tom thanked everyone for their interest and input, noting that the session covered a lot of WCMP2 ground that was valuable for the catalogue work.

Actions

  • Steve add provenance link to one of the examples and open a PR
  • Anna open an issue to add an overview table to the KPI section - DONE https://github.com/wmo-im/wcmp2/issues/300
  • Tom to update pywcmp to add KPI link check deduplication - DONE https://github.com/World-Meteorological-Organization/pywcmp/pull/156
  • Anna open a GitHub issue in WCMP2 on considering dashboards for deeper metadata analysis.
  • Guillaume to give a live demo of EUMETSAT Elasticsearch/Kibana-based dashboard at the next meeting.
  • Lara to show MET Norway's panel-based notebook tool at the next meeting.
  • Tom to demo what pywcmp and pywiscat can currently do from the command line for metadata analysis at the next meeting.
  • Lara to lead drafting of a KPI for themes and concepts (open a GitHub issue in WCMP2 and draft the text).
  • Tom to open an issue in W2IT (WIS2 Implementation & Transition) on further analysis of topic usage via the Global Replay service — how much use different topics are getting used. DONE https://github.com/wmo-im/et-w2it/issues/57

Next meeting

2026-06-23