NOAA/NDBC latest observations
Primary real observation source when a station reports wave height, period and direction.
United States, Pacific islands, selected international partner stations
View sourceData sources
SwellOracle separates real observations from marine models. That distinction matters: a buoy measures the ocean with an instrument; a model estimates conditions where nearby measurements do not exist.
Australia: 14; Spain: 11; Ireland: 5; New Zealand: 1; Japan: 23; Canada: 19; United States: 7. Total: 80 stations included in the real-observation catalog.
Integrated sources include AODN/IMOS, Puertos del Estado, Marine Institute Ireland, Environment Canterbury, NOWPHAS/MLIT, ECCC/DFO via NDBC and NOAA/NDBC.
Stations are now connected to their regional hubs, with new national hubs created for Japan, New Zealand, Canada and the United States.
A station is treated as operational only when it provides a recent reading and useful surf variables such as wave height, period or direction.
Chile, Brazil, Mexico and South Africa remain outside this count because their current sources do not yet provide a stable integrated endpoint.
Open-Meteo Marine lets SwellOracle cover coasts without nearby buoys using named model points. Those points provide total wave, primary swell, secondary swell, wind wave, period and direction.
These readings should not be presented as real buoys. In the app they are labelled as marine model data so users know they are forecast-derived estimates.
The next priority is Copernicus Marine In-Situ for Mediterranean, European and global platform observations. After that, AODN/IMOS is the best candidate for Australia.
Regional national connectors can follow one by one: DHN in Peru, NOWPHAS/JMA in Japan, INCOIS in India, KHOA/KMA in Korea, Taiwan CWA, PNBOIA in Brazil, SHOA in Chile, NIWA/MetOcean in New Zealand and South African services.
Availability depends on the source, station, and published variable. A page may exist even when its latest observation is old or when there is not yet enough history to draw a 24-hour series.
SwellOracle shows the source type, update time, and available variables so you can distinguish a recent measurement from a historical reference or a model estimate.
Before entering the water, compare height, period, and direction with your beach exposure and check official marine warnings. SwellOracle data supports a decision; it does not guarantee conditions or replace maritime authorities.
Editorial guides, methodology, the editorial policy and the data-corrections page are published as original content with authorship, review dates and links to relevant sources.
Station pages remain available in the app. The sitemap prioritizes an editorial selection of real references with recent readings, so a technical catalog does not compete with useful session-planning guides.
Read the applied guides for Orange County, central Chile and deciding whether to trust a surf reading before you paddle out.
The Orange County Killer Surf Links partnership pilot now connects seven breaks to their NOAA/NDBC buoy pairs and direct Surfability Score™ pages.
This registry separates sources already active from candidate networks for adding real buoys in new areas.
Primary real observation source when a station reports wave height, period and direction.
United States, Pacific islands, selected international partner stations
View sourceGlobal model fallback used for named model points where real wave buoys are sparse; Chilean coast model points are refreshed as active stations.
Official Peruvian wave forecast maps and coastal products. Current public pages expose images, not reusable point observations, so SwellOracle keeps them separate from real buoy data.
Backend can use this when COPERNICUS_MARINE_POINT_URL is configured.
global, Mediterranean, Europe, southern hemisphere
Real in-situ wave observations are queried for configured regions, including Denmark and nearby North Sea platforms when Copernicus exposes useful wave variables.
Australian wave buoy observations are loaded from AODN THREDDS/OPeNDAP for the seeded East Australia, West Australia and Tasmania stations.
Active production connector through the public Portus real-time sea-state feed. Uses explicit station mappings and keeps coastal and deep-water observations identified separately.
Irish Marine Data Buoy Observation Network. The connector reads the latest 48 hours and keeps the newest sample per buoy.
Canterbury Wave Buoy measurements from Environment Canterbury / Earth Sciences NZ.
Public real-time XML map feed from Japan's MLIT NOWPHAS network.
Canadian Pacific buoy observations are processed by ECCC and distributed through the public NDBC latest-observations feed.
Verified public CENDHOC wave-data endpoint for Chilean observations. Integration remains gated on stable response/schema checks and confirmed reuse terms before presenting measurements as live buoy data.
Candidate for Peruvian coastal observations if a public, reusable feed with wave height, period and direction can be confirmed.
Candidate for Brazil and South Atlantic coverage through official Brazilian marine observations.
Candidate for India and Indian Ocean coverage if public wave feeds can be consumed reliably.
Use one connector per official provider once API format, license and reliability are confirmed.
Korea, Taiwan, South Africa, Argentina, Uruguay
These points are not real buoys: they are named model references for coastlines without nearby instrument coverage.
Northland west coast model, Northland east coast model, Raglan coast model
Para coast model, Ceara coast model, Rio Grande do Norte coast model
Arica coast model, Iquique coast model, Antofagasta coast model
North Peru coast model, Bayovar coast model, Salaverry coast model
New South Wales coast model, Queensland coast model, Victoria coast model
Galicia coast model, Canary Islands model, Balearic Islands model
Ligurian Sea model, West Sardinia model, Sicily coast model
Crete coast model, Aegean Sea model, Ionian Greece model
Bali coast model, Mentawai coast model, South Java model
When a good nearby buoy exists, use it as real confirmation. When it does not, a clearly labelled model point provides useful coverage without confusing measurement and estimation.