Skip to content

Meeting 2026 06 23

Jump to bottom
amilan17 edited this page Jun 25, 2026 · 9 revisions

Date/Time

Tuesday 23 June 2026 13h - 14h30 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 X
Ian MCDONALD EUMETSAT @McDonald-Ian
Steve OLSON USA @solson-nws
Ján OSUSKÝ HMEI @josusky X
David PODEUR France @davidpodeur
Julia SIELAND Germany @jsieland X
Yutaro TONOOKA Japan YutaroJMA
Yusuke NAKAGAKI Japan X
Anna MILAN WMO Secretariat @amilan17 X
Washington OTIENO WMO Secretariat X

Agenda

Notes (DRAFT)

  • Review of previous actions: the group worked through outstanding actions from the April meeting (KPI overview table in the WIS2 Guide, KPI link added to pywcmp, link-check deduplication) and confirmed they are complete. Anna opened an issue for a metadata-analysis dashboard (referenced as issue ~306 during the call).

  • Secretariat updates (Anna):

    • Fast Track 2026-1 has concluded, covering WCMP2, WTH, the WIS2 Guide, and WIS2 in general; all tagged releases are now on GitHub and the fast-track branches have been merged.
    • Anna is still working on deploying the updated codes and topics; the WIS2 Box already picks these up on restart.
    • More topics and KPI-related guide content are expected ahead of INFCOM-4 (targeting around 26 November); KPI material needs to be finalized and double-checked by end of June for translation/submission.
    • There is no other fast-track schedules for WIS in 2026; new amendments should go through INFCOM-4. However, the Secretariat wants to avoid blocking new WIS2 topics from being approved, and can put together an ad-hoc fast-track if needed. They are also looking at shortening the NFP review cycle (currently roughly 2 months minimum under the fast-track procedure) to one month.

  • Metadata analysis demonstrations:

    • Tom — pywcmp / pywiscat: demoed pywiscat, which runs metrics against a downloaded Global Discovery Catalogue (GDC) metadata archive — e.g. breaking down record counts by centre ID, by Earth system discipline, and by recommended/required field completeness. This is being superseded by a move to a daily GDC job that publishes metrics directly via the open metrics standard, rather than the current on-demand/real-time pywiscat run. Anna asked whether a single KPI (or a single record) could be checked in isolation rather than running the full suite; Tom agreed this is feasible and a good idea for a future pywcmp enhancement.
    • Guillaume — EUMETSAT Elastic/Kibana dashboards: presented how EUMETSAT crunches service log files into Elasticsearch/Kibana dashboards (daily stats, geolocation, trends over time) for operational monitoring; suggested the same approach could be applied to metadata/KPI analysis (e.g. dashboards highlighting which records or KPI areas are non-conformant), without needing heavy infrastructure. Clarified that GDCs will keep publishing metrics via the open metrics standard either way (it can feed Prometheus/Grafana or an ELK-style pipeline) — Tom noted the underlying frequency is shifting from real-time to a daily batch job. In response to Anna's question about scale, Guillaume noted Elastic/Kibana scales with available storage/infrastructure and that EUMETSAT's usage doesn't require significant scaling investment.
    • Lara — Met Norway / Arctic Data Centre URL checker: presented a prototype URL/link checker built against the Arctic Data Centre's metadata catalogue (12+ million records, mostly satellite data granules). It checks resolvability and HTTP status of links such as data access, WMS, user guide, and homepage URLs, logging broken/working status; results could feed Prometheus/Grafana or a simple report. Open design questions noted: how to handle expected/legitimate downtime (e.g. brief outages vs. links that fail consistently over time), URL relevance/importance weighting (data access vs. landing page), and scaling/performance given the large number of records and the application server load. Not yet in production.

  • Discussion following the demos: the group agreed there's value in moving toward dashboards (Grafana- or Elastic-style) for GDC metadata/KPI content, complementary to the existing technical open-metrics monitoring. This is expected to be a longer-term initiative and a good discussion point to raise at INFCOM around the KPI dashboard concept.

  • KPI status / where KPI guidance lives:

    • Anna asked how much real-world testing/validation has been done of the KPIs against actual records, given the goal of finalizing this for the WIS2 Guide by end of June. Tom confirmed Canada's GDC runs full KPI checks after each ingestion and reports back.
    • Lara raised that there's a gap between the WCMP2 specification and the KPI documentation — e.g. a KPI like "title in sentence case" isn't traceable back to anything stated in the spec itself, making it hard for someone reading the spec alone to understand what's expected.
    • Julia agreed, noting the original KPI thinking was rooted in WIS 1 and may need revisiting; she gave an example of around eight WIS 1-era checks that could likely be consolidated into a single, better WIS2-appropriate KPI.
    • Anna raised the option of embedding KPI rules directly into the relevant clauses of the WCMP2 standard rather than keeping them as a separate document. Tom noted two structural options (interspersing them into the spec vs. an informative annex) but felt KPIs don't map cleanly onto the spec's existing structure of requirements/recommendations/permissions, since a KPI isn't itself a specification clause.
    • Ján noted the WIS2 Guide doesn't currently spell out the KPIs explicitly (no KPIs existed yet when that guide text was written), and suggested that over time, as metadata analysis shows a KPI is consistently met across most/all centres, it could graduate into becoming an actual recommendation in the specification — i.e. KPIs as a feeder mechanism for future spec requirements, not a parallel structure.
    • Precedent check: Anna confirmed the WCMP1 KPIs were added to the WIS2 Guide (not the Manual) as a full copy-paste annex. The group agreed to follow the same pattern for WCMP2: an annex in the Guide.
    • Lara separately raised a scoring/quality question: do KPIs currently just check for the presence of a field (e.g. title, description), or do they assess actual quality/consistency of the content? Tom and Lara agreed presence-only checks should be removed where they duplicate existing WCMP2 conformance checks (since basic required-element conformance is already enforced upstream in pywcmp before any KPI scoring happens) — KPIs should focus on quality/consistency criteria, not redundant presence checks.
    • Anna will check with Hassan and Enrico to confirm officially where KPI documentation should live and report back.
    • Tom proposed adding a column to the KPI table linking each KPI back to the specific element/clause in the WCMP2 specification it relates to, to close the traceability gap Lara raised; Anna agreed this is worth doing suggested as inline references near the relevant spec sections.

  • Review timeline: Anna noted the review window for this KPI/Guide work is short — finalized and double-checked by end of June (one week from the meeting date). Julia volunteered to help with review and will fold it into the "title is present" KPI removal work she's already doing; she will open a PR by 30 June. Tom will also do a pass / pull request to remove the "title is present" KPI check and add the specification-link column.

  • Lara's draft proposal — KPI for concepts/topic categories: with only ~5 minutes remaining, Lara briefly previewed a draft idea (following on from the last meeting) for a new KPI assessing the richness of "concepts" attached to a metadata record beyond the mandatory topic category — e.g. rewarding records that add concepts from additional, well-chosen schemes/vocabularies (such as Earth discipline), and considering whether the concept's URL is resolvable. Open questions noted: whether scoring should be binary (yes/no) or graduated per scheme/URL, and how to handle the mandatory vs. optional concept fields (title, description, URL). Anna and Tom agreed this needs more time and should not be squeezed into the current short-timeline review — to be picked up at a later meeting given its importance.

  • Next meeting scheduling: discussed timing — Anna is on leave for the last two weeks of July; the group considered early August but leaned toward late August instead to avoid a quiet first week back. Settled on Monday 24 August, 12:00 UTC, for 1.5 hours (earlier start time than this meeting).

Actions

  • Tom: Make sure KPIs are in the WIS2 Guide branch for INFCOM DONE / in motion https://github.com/wmo-im/et-w2it/issues/44#issuecomment-4788237698
  • Tom: add individual KPI analysis option to pywiscat
  • Tom: get team access to the Global Monitors to review their approach
  • Anna: check with Hassan and Enrico on where the KPIs are going - Annex of the WIS2 Guide
  • Tom: add link to the element in the specification in the KPI descriptions and remove "title is present" from KPI check DONE https://github.com/World-Meteorological-Organization/pywcmp/pull/161
  • Julia: once over for consistency and open PR as needed by June 30
  • Tom: open PR to remove the "title is present" KPI check (duplicate of existing WCMP2 conformance check) and add a specification-link column to the KPI table
  • Lara: continue drafting the proposed KPI for metadata "concepts" / topic-category richness (resolvable URLs, scheme diversity) — to be discussed in full at a future meeting, not rushed into the current KPI review
  • All: review the WCMP2 KPI annex content ahead of the end-of-June deadline for WIS2 Guide submission

Next meeting

2026-08-24 12 UTC

TT-WISMD WMO Official website

Clone this wiki locally