How Far Ahead Are Tide Forecasts Accurate?

Short answer: standard tide predictions are reliable within a defined window — for most academic and government predictors that's roughly 1970 to 2025, a 55-year span. Outside that window, predictions technically still run, but their accuracy drops because the underlying equations rest on three assumptions that quietly drift over decades: coastal topography, global sea level, and the astronomical constants themselves.

This isn't a bug in any one predictor. It's a fundamental limit on what tide tables can do.

Why there is a standard window at all

Most tide predictors — both government tables and academic tools — use the harmonic method: each station's tide curve is decomposed into a sum of sine waves whose frequencies come from Earth-Moon-Sun geometry. The amplitudes and phases for each wave come from years of water level observations at that specific station.

That derivation has a hard requirement: the station's physical surroundings must stay roughly the same. Sandbars shift, harbors get dredged, sea walls go in, coastlines erode. Any of these changes how water flows in and out at the station, which changes the tide curve — in ways the original harmonics can't predict.

When the maintainers of the academic WWW Tide/Current Predictor at the University of South Carolina documented why they capped their standard range at 1970–2025, they pointed to this assumption: the harmonic constituents are calibrated to observations from a specific window, and outside that window the constituents start drifting measurably from current reality.

Three independent sources of drift

When you push tide prediction past the standard window, three things start to fail in parallel:

1. Coastal topography changes. Sand moves. Channels silt up. Harbors get deepened. Any of these changes how water flows in and out at the station, which changes the tide curve. The harmonic method assumes the bathymetry is stable. Over 10–20 years, this assumption breaks for a lot of stations.

2. Global sea level rise. This is the slowest but most certain drift. Mean sea level has been rising globally at a few mm per year. Over a century, that's enough to shift the entire tide curve up. Government tide agencies periodically re-derive the harmonic constituents to absorb this, but predictions made in 1995 for the year 2050 won't reflect the 2050 sea level.

3. The mathematical assumptions themselves. This is the deepest issue. The harmonic method assumes that Earth-Moon-Sun orbital mechanics are stable on the timescales of prediction. They are — but only for a century or so. The longer you project, the more the astronomical constants drift.

The maintainer of the academic WWW Tide/Current Predictor put it bluntly in the predictor's own documentation, in a section explicitly titled Distant Date Issues (2005/2006 revision):

"The mathematical underpinnings for the predictions have some assumptions built into them that are local (in a time sense) to our century (or thereabout). We don't know how bad it is to violate those assumptions by stretching the maths to cover distant dates."

That quote is from a page that exists to warn users about the feature. It is one of the few places anywhere — academic, government, or otherwise — where the limits of tide prediction are stated this directly to non-specialists. Most public-facing explanations of tide prediction stop at "we use harmonic constituents" and don't go further.

What "local to our century" means in practice

The astronomical part of tide prediction — Earth-Moon-Sun geometry — is fine for any date. Those cycles are well understood and the long-period constituents are computed analytically, not measured.

Where things get shaky is the empirical part: the harmonic amplitudes and phases that came from real-world observations at a specific station. A station measured in 1960–1990 will have harmonics tuned to that 30-year window. The further out you go past 1990, the more you're asking those harmonics to extrapolate behaviors they weren't measured for.

Practical breakdown:

Why "distant date" mode forces UTC

If you've ever used an academic tide predictor and seen it ask you to manually convert from UTC to your local time, you bumped into this. The "distant date" feature in tools like XTide forces UTC output because:

You can still get local-equivalent times by adding your offset, but the predictor won't do it for you. That's a deliberate design choice to avoid giving you a wrong-looking local time.

"Boldface" vs "legacy" stations

Academic predictors like XTide classify stations into two groups:

If you try to run a distant-date prediction for a legacy station, you'll get an error. This is by design — the maintainers don't trust the older harmonics for long-range work.

This distinction is mostly invisible in commercial tide apps, which usually compute predictions only for current and near-future dates anyway. It's only relevant if you're trying to build your own tide prediction pipeline from raw data.

What this means for your tide planning

For trip planning, fishing, surfing, sailing — all the practical uses of a tide chart:

The honest answer: tide predictions are one of the most reliable kinds of weather-adjacent forecast, but only on the timescales they were designed for. Outside that, you're extrapolating.

For current station-by-station predictions, see the Maine locations directory or the Charleston, SC reference page. This article is the explainer underneath.

References

  1. Dean Pentcheff, WWW Tide/Current Predictor — Distant Date Issues, 2005/2006 revision. Recovered via Wayback Machine, 2006-06-23 snapshot: https://web.archive.org/web/20060623153740/http://tbone.biol.sc.edu/tide/distantdate.html
  2. Dean Pentcheff, WWW Tide Prediction FAQ, 1999. Recovered via Wayback Machine, 1999-11-28 snapshot: https://web.archive.org/web/19991128144929/http://tbone.biol.sc.edu/tide/faq.cgi
  3. David Flater, XTide (current project home): https://flaterco.com/xtide/