Open Ocean Data: Workshop Summary and Build Plan

Edge City India, Hub Goa · New City Workshop, Thursday 15 October 2026 · Ocean Climate Fund session

Summary

Ocean data is siloed, paywalled and largely unanalysed, and public collection has thinned. Ocean Climate Fund (OCF) wants to build a globally public, open-source ocean data resource that no one owns. The plan:

• A shared map layer on Gaia
• Data layers on top, including marine protected area (MPA) and BBNJ jurisdiction data
• Sensors to fill gaps
• An easy way for anyone to submit data and APIs
• An Ocean Oracle that pulls it all together and answers questions, through Zeke or any AI, as an API

The starting point already exists: about 25 ocean data APIs and an open GitHub repo at oceanclimatefund.com/data. What's missing is developers and a decision on what to build first. The room agreed on the map layer and a community data call.

Six build workstreams

1. Mapping layer on Gaia (the substrate)

• One consistent base map that every dataset geolocates onto. Ocean data is one of perhaps a hundred layers, connected but unrelated
• Built on the Gaia mapping APIs already published in the Hub Culture developer portal, delivered as an open-source Maps section on oceanclimatefund.com
• First build: base map component plus a pluggable layer format (GeoJSON or vector tiles) that any API or dataset can target

2. Open ocean dataset (mash up the 25 APIs)

• No one has merged all 25 API sets into one resource. The question is what appears when they're layered over each other
• First build: a normalisation pipeline mapping each API to a common schema (location, time, variable, unit, source, licence), stored openly and shown as layers on the map
• Position against existing open aggregators: Copernicus Marine, NOAA ERDDAP, EMODnet, OBIS, Global Fishing Watch and HUB Ocean's Ocean Data Platform. Ingest them where licences allow; the differentiator is the single map-based mash-up and AI access

3. MPA and BBNJ data

• Export the jurisdictional and data capability to set up MPAs faster and more cheaply, building on the Azores work. MPAs protect the area and become fishery engines for the edges within a few years
• Cross-collate BBNJ (high-seas) and MPA data as layers, connected to OCF's advisory role in the BBNJ process
• First build: MPA boundary layer (e.g. Protected Planet/WDPA) plus high-seas jurisdiction polygons on the Gaia map; then an "MPA kit" of templates and data tools

4. Sensors

• OCF is working on funding sensors. Hub Culture has had talks in China on robotic drop-in sensors, and an Indian sensor maker was mentioned in the room
• Data path: sensor, then satellite (e.g. Starlink), then dataset, then verification, then map layer. Add microplastics to the measured variables
• First build: a sensor data spec and ingestion endpoint. Study the Argo float network (satellite uplink, open data) before designing from scratch

5. Data and API submission ("call for data")

• 50+ OCF community member organisations, representing 200+ ocean organisations, hold data but mostly aren't data organisations
• An ingestion point where anyone can submit a dataset or register an API. Accepted APIs get listed on /data and mapped as layers. "So easy to do"
• First build: a submission form and registry (name, owner, licence, endpoint, coverage, update frequency), a review queue, then auto-listing
• Possible Ven or Ven Oxygen incentives for verified contributions

6. Ocean Oracle

• A front-end aggregator over all ocean data, plus an "ocean dashboard" and "voice of the ocean". Any AI can draw from it
• Train or ground Zeke on the open dataset so Zeke becomes an ocean expert, with the open data layer staying the public source Zeke cites
• First build: an Oracle API, or an MCP/API plugin, that queries the normalised dataset by place, time and variable. Zeke is the first client; other AIs can plug in. A possible SingularityNET model tie-in was raised

How they fit together

Sources (25 APIs, community submissions, sensors, MPA and BBNJ data) flow into ingestion and normalisation, then the open archive on GitHub. From there, the Gaia map layer and the Ocean Oracle API serve the OCF Maps section, the dashboard, Zeke and other AIs.

Context from the session

• Why now: NOAA cuts reduced public ocean data collection just as a strong El Niño arrives, and commercial brokers charge heavily for data that should be open
• Local reality: sporadic outside research that brands waters "polluted" meets local resistance. Continuous, locally owned data would be more credible
• Land and nature economics: Ven Oxygen staking at vo2.vc (polygon, proof of stewardship, AI oxygen estimate, payout). Landholders need cash now, not only tokens. Stan's answer: stack dozens of digital assets per parcel for recurring protection income
• OCF portfolio: The Reef Line reef tile (Miami); GigaBlue phytoplankton seeding (likely first blue carbon investment)
• Philosophy: regenerative consumption, not degrowth. Make the regenerative option the convenient one through incentives and design, as with London cycling
• Venues: Bermuda Sprint oceans track; BBNJ COP

How builders plug in

• India WhatsApp chat, then the Hub Goa hub, then projects, then code via the GitHub repo
• Hub Culture's Indian developers coordinate on Slack
• Model: open-source starter tasks first, then a flat fee per milestone (not hourly). Good work leads to the core dev team

People and leads

• On-the-water participant / floating concrete jetty builder: reef-safe moorings
• Participant group invited as an OCF community partner, to publish their APIs on /data
• Developer participant and Siraj: open-source starter tasks
• Haley (SingularityNET): beyond-LLM reasoning models; possible Zeke and ocean-model link
• Hydrogen-focused participant: seawater hydrogen, shipping transition
• Canadian participant; participant asking about the content side
• Indian "blue ocean" fund (~₹1,500 crore): identify the vehicle and access route

Fix before publishing

The Goa hub wiki is public, so these session claims need correcting or sourcing first.

Correct the figure:

• Phytoplankton oxygen: about 50% of Earth's oxygen (range 50–80%), not a flat 80%
• Ocean pH: about 8.2 before industrialisation, about 8.05 now. Not 9.6 → 8.4, and no established 8.4 die-off threshold
• Global EV share: about a quarter of new car sales, not 60% (China is near 60%)
• Ecological footprint: about 1.7 Earths a year globally, not 2.4
• BBNJ timing: in force since January 2026, with the first COP due within a year
• Azores: the largest MPA network in the North Atlantic, not the world's largest MPA
• El Niño: "five times bigger than ever" isn't a standard measure; cite the current NOAA or WMO outlook

Remove or reframe:

• Oxygen collapse: drop the 40-year carbon lag, the 200-year uninhabitability claim and the O2-to-ozone mechanism. Use documented ocean deoxygenation (about 2% of dissolved O2 lost since 1960) and acidification instead
• Ven Oxygen value: drop "almost certainly grow" in public material

Needs a source:

• NOAA's 80% cut
• OCF as "first private advisor" to the BBNJ
• The SingularityNET founder's name and AGI conference dates

Probably keep off a public wiki:

• The upcoming GigaBlue investment (unless announced)
• Developer pay terms
• The Greenland and Canada water-security reading (political)

Stan's added context

Next steps (suggested order)

• ☐ Post the call for data and stand up the submission form and registry (workstream 5)
• ☐ Spec the Gaia base map and layer format, and assign as the first open-source starter task (workstream 1)
• ☐ Normalise the first 5 of the 25 APIs onto the map as a proof of concept (workstream 2)
• ☐ Add the MPA and high-seas boundary layers (workstream 3)
• ☐ Draft the Oracle API and Zeke plugin interface (workstream 6)
• ☐ Write a sensor data spec; review Argo; follow up on the India sensor maker and China talks (workstream 4)
• ☐ Contact HUB Ocean about partnering vs. overlap
• ☐ Clear the "Fix before publishing" list, then post to the Goa hub wiki