Data and downloads

No release yet. Download links, file sizes, checksums and the DOI are coming with the first release. The URLs below show the pattern; they do not serve data yet.

Release files

Every release has one stem, OTC_{YYYYMMDD}, where the date is the UTC day the release was built. A second release on the same day gets a counter: OTC_{YYYYMMDD}.2.

Files in each release
File	Contents	Download
`OTC_{YYYYMMDD}.json`	Primary format. One document with every station, its constant sets, constituents, offsets, conventions, validation and provenance.	Coming with the first release
`OTC_{YYYYMMDD}.json.gz`	The same file, gzip-compressed.	Coming with the first release
`OTC_{YYYYMMDD}.jsonl`	One station document per line, for streaming and for clean per-station diffs between releases.	Coming with the first release
`OTC_{YYYYMMDD}.csv`	One long table for spreadsheets, with a `record_type` column (`station`, `constant_set`, `constituent`, `offset`, `convention`, `validation`, `tombstone`). Columns that do not apply to a row type are empty. RFC 4180 quoting, UTF-8, LF line endings.	Coming with the first release
`OTC_{YYYYMMDD}.parquet`	Optional. The CSV table, typed and compressed, for analysts. A release is complete without it.	Coming with the first release
`OTC_{YYYYMMDD}.sha256`	SHA-256 checksums of every file in the release, in `sha256sum` format.	Coming with the first release

Where downloads will come from

Files: https://data.opentideconstants.org/OTC_{YYYYMMDD}.{ext}, served from Cloudflare R2. Every release stays available, and a dated file is never overwritten.
Latest pointer: https://data.opentideconstants.org/OTC_latest.json is a small manifest, not a data file. It gives the current release date, its format_version, its DOI, and the URL, size and SHA-256 of each file. It is updated only after every file of a release is uploaded and verified. Read it first, then download the dated files.
Archive: the same public files are deposited at Zenodo with a DOI for each release (coming with the first release). See Citation.

Check a download with:

sha256sum -c OTC_{YYYYMMDD}.sha256 --ignore-missing

Format

The JSON document has five top-level fields: format_version, release, conventions, licences and stations.

station: A stable station_id in the OTC namespace (never changed, never reused), name, country, position, type (reference or subordinate), aliases in other systems (NOAA, GESLA, TICON, XTide, Kartverket, Slackwater), and the recommended_set_id. A removed station stays as a tombstone with its reason.
constant set: One per source record: source, record id and version, record span, datum, convention_id, licence_id, QC status and flags, the constituents, and provenance. A station can have several sets; one is recommended by a published rule.
constituent: OTC canonical name and the source's own name, Doodson number, speed (°/h), amplitude (m), Greenwich phase (°), uncertainties, and why it was kept. Dropped constituents are listed with the reason.
subordinate offsets: Reference station, time offsets at high and low water (minutes), height offsets, and whether they are ratios or additive, as NOAA gives them.
convention: Phase reference, V0 model, nodal handling, constituent-table version, and the result of the per-source check that confirmed it. See Conventions.
validation: Per station and window: time mean absolute error, 95th percentile, bias, height error, range error, missed and extra events, the reference source and its distance, and the same numbers for the previous release. See Validation.
licence: SPDX id (or a LicenseRef- id), provider, required citation, and the assembled attribution text. See Licence and attribution.

JSON Schema

otc-0.1.schema.json: the draft JSON Schema (draft 2020-12) for format_version 0.1. Each line of the .jsonl file validates against #/$defs/station.
example.json: a small example document with illustrative values (not data). It has one station with two constant sets and one tombstone.

The format version is separate from the release date. A new field is a minor change; a removed or renamed field, or a changed unit or convention, is a major change. The schema may still change before the first release.

Converters

Converters, in both directions, with a lossless round trip as a release gate: TICON CSV, XTide TCD (through libtcd), and NOAA harmonic-constant JSON. The reference predictor will publish conformance vectors (expected high and low water times and heights at named stations and dates) so that other predictors can check their results. Both are coming with the first release.