Open tide data, combined into one dataset.
OpenTideConstants (OTC) combines official agency constants, constants that we fit from gauge records, and tide-model constants into one consistent dataset: one format, declared conventions, provenance for every record, and the accuracy of every station against official predictions. It is rebuilt automatically from the latest upstream data and released under CC BY 4.0, with a DOI for each release.
Helgoland. In the proof of concept, TICON-4's mean time error here was 88–90 min. Our fit, with the time base corrected, was 6.7 min.
Status: in development. The first release is coming. This site describes the method and the format. Downloads, DOIs and release accuracy figures come with the first release. The only accuracy numbers on this site come from a proof of concept at 11 stations, and they are labelled as such.
Three kinds of source, one release
Each source has its own licence, format and conventions. OTC reads each one with its own adapter, checks it, and writes everything into one format. At one location, official constants rank before a gauge fit, and a gauge fit before model constants.
-
Official constants
Published by the agency responsible for the gauge. Ranked first.
-
Gauge records that OTC fits
Sea-level records, fitted after the time-base checks. Ranked second.
- GESLA
- PEGELONLINE Germany
- Rijkswaterstaat Netherlands
- Marine Institute Ireland
- DMI Denmark
- SMHI Sweden
- FMI Finland
- UHSLC Fast Delivery
-
Model constants
From global tide models at station points, always labelled as model constants. Ranked third.
One OTC release
OTC_{YYYYMMDD}
The first release is coming.
- One format. JSON with a published JSON Schema, plus JSON Lines and CSV.
- Declared conventions for every constant set.
- Provenance for every record.
- Accuracy for every station against official predictions.
- Every set kept. A station lists every constant set, with one recommended by a published rule.
- SHA-256 checksums for every file, and a DOI for each release.
Licence, attribution, access and coverage for each source are on Sources.
Why one consistent dataset matters Proof of concept
The German WSV gauge records in GESLA-4.0 are in German civil time but are labelled UTC. TICON-4 fitted them as they were, so its phases for those gauges are late. A dataset that combines sources must check every time base and convention before it uses them.
At Helgoland, the proof of concept scored TICON-4 at 88–90 min mean time error, and our fit of the same station, with the time base corrected, at 6.7 min. These are proof-of-concept numbers, not release results.
Helgoland
Schematic of one tide. Time of high water against the official prediction.
- Official prediction
- TICON-4, mean time error 88–90 min
- Our fit from GESLA, mean time error 6.7 min
Schematic: each curve is the official curve shifted by that set's mean time error in the proof of concept of 2026-10-07. The direction of our fit's shift is for illustration only. All 11 stations and the method are on Validation.
What is different
-
One dataset from many open sources
Official constants, constants fitted from gauge records and, later, model constants, in one format. Every station has a stable id, its aliases in other systems (NOAA, GESLA, TICON, XTide, Kartverket, Slackwater), and every constant set from every source, with one recommended by a published rule.
-
Accuracy for every station
Every release publishes, per station, how well the constants reproduce official predictions: time and height errors, missed and extra events, and the same numbers for the previous release.
-
Declared conventions
Every set of constants states its phase reference, time zone, astronomical-argument model, nodal corrections and constituent definitions. A converter never guesses.
-
Time bases checked
Every input record is checked for a wrong or undeclared clock before it is used. A record whose time base cannot be explained is not used.
-
Automatic and current
Weekly watchers detect upstream changes. A build that passes all gates is published with no person in the loop. A build that fails keeps the last good release live.
-
Open format
JSON first, with a published JSON Schema, plus JSON Lines and CSV. Every file is named
OTC_{YYYYMMDD}and comes with SHA-256 checksums. -
Open process
The build code, the rules, every automated decision and the issue tracker will be public. Anyone will be able to report a problem with a station.
Sources at a glance
| Source | What OTC takes | Licence | Rank |
|---|---|---|---|
| NOAA CO-OPS | Official constants and subordinate-station offsets | Public domain (US Government) | Official |
| Kartverket (Norway) | Official constants | CC BY 4.0 | Official |
| NOAA CO-OPS currents | Official tidal-current constants | Public domain (US Government) | Official (currents) |
| GESLA | Gauge records that OTC fits (derived constants only) | CC BY 4.0 on derived constants, per-provider attribution | Gauge fit |
| PEGELONLINE (Germany) | Gauge records that OTC fits | DL-DE->Zero-2.0 | Gauge fit |
| Rijkswaterstaat (Netherlands) | Gauge records that OTC fits | CC0 | Gauge fit |
| Marine Institute (Ireland) | Gauge records that OTC fits | CC BY 4.0 | Gauge fit |
| DMI (Denmark) | Gauge records that OTC fits | CC BY 4.0 | Gauge fit |
| SMHI (Sweden) | Gauge records that OTC fits | CC BY 4.0 SE | Gauge fit |
| FMI (Finland) | Gauge records that OTC fits | CC BY 4.0 | Gauge fit |
| UHSLC Fast Delivery | Gauge records that OTC fits | Free use and redistribution, with citation | Gauge fit |
| EOT20 | Model constants at station points | CC BY 4.0 | Model |
| FES2022 | Model constants at station points | AVISO licence (CC BY 4.0 based) | Model |
At one location, official constants rank before a gauge fit, and a gauge fit before model constants. Model constants are always labelled as such. Sources that were checked and not added are listed under Considered, not added. See Sources for licences, attribution, update cadence and coverage, Method for how each source is processed, and Why for the problems this project addresses.