Product Coordination Group

PCG#8

April 2026

The two sessions of the meeting have been blended into a single report wherever possible for clarity and completeness. The presentation was given live in the first session and shown as a recording in the second. The text of this summary was edited with the help of AI.

Introduction

The OBIS Products Coordination Group met in two sessions. The agenda covered updates on the Products Catalogue, the status of the obistools R package, and a presentation from OBIS-AU and IMOS on an aggregated Australian seabird data product.

Products Catalogue

Live and sandbox instances. The Products Catalogue at products.obis.org was soft launched the previous month. Two instances are now available: the live production version, and a development sandbox (https://dev.products.obis.org:8443/) for exploring changes before they are launched. A video tour is coming soon, and user documentation is being updated alongside development. Stephen Formel explained that, although the idea of a static version of the catalogue had been explored, it became clear that CKAN offered many benefits, and the catalogue continues on CKAN. Issues are welcome on the GitHub repository (https://github.com/iobis/obis-products-catalog/issues).

Governance. The group is looking for two or three volunteers to brainstorm the governance of the catalogue, including gatekeeping and quality requirements, and alignment with IOC/UNESCO policies. The Secretariat will then implement what is agreed.

Should reports and presentations count as products? GitHub issue #15 proposes excluding reports and presentations from the catalogue.

→ Eduardo Klein found it interesting to include them, but noted that the governance could be complicated, and suggested leaving them out at first and revisiting the question later.

→ Katherine Tattersall saw real value in making slide decks accessible, especially for Nodes, but suggested they might be better hosted somewhere other than the catalogue.

The obistools R package

A question raised on the SMBD Slack asked whether obistools had been abandoned: it could not be installed on R 4.5.1, had not been updated since 2024, and its most recent version dated from 2019. Stephen Formel had discussed this with Pieter Provoost. The package is not abandoned, but it has no formal governance and needs work, including a review of its metadata and requirements so that it installs with current versions of R, and a thorough review of current and desired functionality. The PCG and DCG are appropriate places to find people for this.

→ Several members volunteered. Sachit Rajbhandari would like to contribute. Eduardo Klein saw a good opportunity to add functions for working with the OBIS cloud-optimised products, for example using DuckDB, and offered to explore links with LLM tools. Tom Webb expressed interest in helping scope potential functionality, and Jon Pye was also interested, particularly given his role in pyobistools. Dan Lear noted that the DCG would discuss the package later that week.

Australian seabird data product

Presented by Sachit Rajbhandari (OBIS-AU) and Eduardo Klein Salas (IMOS)

The presentation described a joint OBIS-AU/AODN project, an IOC ADU and IOC NODC collaboration, to aggregate Australian seabird data, together with the general IMOS strategy for producing and managing cloud-optimised datasets.

Purpose. The product was requested by the Australian Government department of environment. OBIS-AU and IMOS AODN share the goal of an end-to-end pipeline from raw data to a decision-ready product. Seabirds are key indicators of marine ecosystem health, and the data supports national environmental reporting and marine spatial planning, among other uses.

Quality control (OBIS-AU). The product draws on 41 datasets spanning 1943 to 2025, with 123,080 records. A data audit addressed typos and abbreviations, invalid timestamps and missing scientificNameAuthorship, bringing consistency across the datasets.

Aggregation (IMOS-AODN). The 41 datasets were harmonised and aggregated into a single Parquet file with rich metadata and automatic refresh. Added fields include survey type, an H3 spatial index and tags for Australia’s Marine Regions. The file is highly compressed, at around 5 MB.

Access. The product is discoverable in the AODN data portal (https://portal-beta.aodn.org.au/details/ec2c0ef9-3645-4ded-b617-c8297f6eb250?tab=summary). The orchestration code is on GitHub in a private repository, available on request. Demonstration notebooks in Python and R are publicly available (https://github.com/aodn/imos-user-code-library/tree/master/NESP) and allow very efficient data exploration, and training videos have been posted on the IMOS YouTube channel (https://www.youtube.com/@IMOS_AUS/playlists).

→ Laurent Chmiel called it an amazing resource and asked how the stakeholder is using it. Eduardo Klein explained that there had been multiple meetings with the department to establish the level of detail it needed and to adapt the product accordingly. The product was not developed for the general public, but the training videos and tutorials are public and allow others to use it.

→ Stephen Formel asked whether a feedback stream could be created to improve the quality of submitted data based on the findings of the quality control step. Sachit Rajbhandari explained that much of the data was legacy and in-house data, so this was a good opportunity to improve previously published data, and the QC process now in place will help audit new data as it is generated and published. Eduardo Klein added that seabirds were a comparatively easy case because all the datasets already existed in Darwin Core. Upcoming seagrass work will be harder, because the data comes in many schemas that will take more effort to standardize.

→ Katherine Tattersall noted that the seabird dataset is one of several aggregated products being produced in Australia, with other projects under way that also involve publishing data to OBIS as Darwin Core. She asked how the seabird Parquet file aligns with the OBIS Parquet files. Eduardo Klein replied that there is no standard for Parquet datasets, but the two should be compatible; the seabird product is a single file, whereas OBIS datasets are split across multiple files.

→ Silas Principe asked how long it took from the department’s first approach to this result. Eduardo Klein described it as a one-year project. Negotiations took the first two months, and the pilot covered six datasets. Seagrass proved the most complicated, while seabirds were easier. Product development took two to three months, with around 80% of that time spent on data harmonisation.

→ Jon Pye asked how the original data providers felt about the creation of an aggregated product. Katherine Tattersall acknowledged that aggregation is always a challenge, but the team had been rigorous about keeping provenance, so that the original research groups can always be traced and attributed. There had been no pushback so far, though the product is recent.

→ Stephen Formel asked how this work could be generalized and applied at the level of OBIS Nodes. Katherine Tattersall said that once data is in the Darwin Core Archive standard, producing aggregated datasets of this kind should be straightforward. Delivering a regional view would depend on the resources OBIS can devote to it, since the needs of each region have to be identified first.

→ Stephen Formel asked a further question about what this would mean for a Node without a large team. Katherine Tattersall noted that the code is open and provides a starting point, and that OBIS could perhaps offer support or training.

Action items