This page carries a cut of the global tide model around each of these places and the NOAA constants of the stations near them, so it needs no key. The hosted API answers any coordinate on Earth, and 2 951 ports by name.
Rising
Drag the ribbon, or focus it and use the arrow keys. The curve is the water level through this local day,
computed for this place from the tide model. The quay, the boat and the sky are drawn, not measured — the water is the answer.Drag the ribbon. The quay is drawn, not measured — the water is the answer.
This is the request behind what you just dragged.
Nothing on this page is mocked. The panel is the live response for the exact instant under the playhead — same fields, same rounding, same datum, same source.
datum=
source=
One place. One year. One image.
Every pixel is a real water level — 35 040 of them per place, computed in your browser as you watch. The wall is what a year of /v1/tides/series looks like at once.
Spring and neap, found — not assumed.
The tide lags the Moon by a day or two, and by how much depends on the place. So the spring tide is the day with the largest actual range, not the day of the new moon, and the answer says which day that was.
When is there enough water?
The question people actually ask. Name the level you need and the answer is a list of windows with their edges solved on the curve to the second — and a flag when a window was already open when the day began, instead of a fake crossing at midnight.
Where naive implementations go wrong.
Each of these is a real answer from the same code, with one assumption changed. The difference is the product.
A height is nothing without its zero.
Charts count from chart datum: MLLW in the United States, the lowest astronomical tide almost everywhere else. A tide height above mean sea level compares with neither. Same water, same instant, three different numbers — and the answer names which one it is.
The Moon's orbit wobbles. So does your amplitude.
The plane of the Moon's orbit precesses once in 18.6 years, and lunar tides swing by about four per cent with it. Code that freezes the node factor at 1.0 is wrong by that much — always, and silently.
Two sources. The answer says which.
A global grid cell is 14 km across; Puget Sound is narrower. In Seattle the grid is off by up to a metre — and says so with unresolved_here. Fed the constants NOAA publishes for its station, the same engine is within three centimetres of NOAA's own table. So it uses the station, and names it.
Seven endpoints. Any coast on Earth. Any date between 1900 and 2100.
Computed from the EOT20 global tide model and, in United States waters, from the harmonic constants NOAA publishes for its stations. Checked against NOAA's own predictions at eighteen stations: 3 to 25 centimetres on open coasts. Where the model cannot resolve a place, the answer says so.