Every six hours and thirteen minutes, give or take the moon's mood, the sea at Penhallow Point turns around. It has done so through two world wars, nine lighthouse keepers, and the entire history of the telephone. Since 1861 this observatory has written the turning down1 — first in pencil, by lamplight, from a float in a stilling well; now sixty times a second, by radar, from a sensor the size of a kettle.
1 Penhallow Point, its observatory, and every reading in this essay are a fiction built to demonstrate a craft. The physics is real; the measurements are invented. See A Note on the Data.A tide gauge is a patient instrument. It does one small thing — where is the water now? — and repeats it until the repetition becomes an archive, and the archive becomes an argument. One reading tells you whether to move the deck chairs. A century and a quarter of readings tells you something harder to look at directly.
This is our ninth annual essay. It makes three observations, in rising order of consequence: the sea is higher than it was; the tide itself remains a piece of celestial clockwork we can take apart; and the two facts together are quietly rewriting the meaning of an ordinary high tide. The charts below draw themselves as you scroll, because a line you watch being drawn is a line you read more carefully.
Part I
The Century Line
Averaged over a year, the sea is a very steady correspondent. Waves cancel, storms cancel, even the tide cancels — what remains is the level itself, and the level keeps a diary. Below is that diary for our composite of eight stations, expressed as millimetres above the 1900 mean.2 Scroll, and read it the way it was written: slowly, then all at once.
2 A “mean sea level” is always relative to a datum — an agreed zero. Ours is the 1900 annual mean at Penhallow Point. Change the datum and every number shifts; the shape of the curve does not.Fig. 1 · illustrative dataAnnual mean sea level · mm above the 1900 datum · synthetic series, method below
Plotted one year at a time, the early record barely trembles. The line wanders — a cool decade here, a run of stormy winters there — but it wanders around a slope, and the slope is the story.
Fit a line through 1900–1990 and the sea rises about 1.4 mm a year — a coin's thickness annually. Slow enough that a harbourmaster could spend a whole career and honestly notice nothing.
From 1993, satellite altimeters began checking gauges like ours from orbit, and the two records agree. That agreement matters: from here on, the line is corroborated from above.
Zoom the axis to the satellite era and the same data reads differently: about 3.4 mm a year now, and the residuals hint the curve is still bending. The diary has changed its handwriting.
Two-and-a-half times faster is a phrase that slides past the ear, so hold it against something solid. At the older rate, the sea needed most of a lifetime to climb a hand's width. At the newer one, it does so in under half of one. Nothing about the water looks different from the sea wall. Everything about the bookkeeping is.
Before that bookkeeping can alarm anyone, though, it has to be separated from a much louder signal — the tide itself, which moves not millimetres a year but metres a day. To subtract the tide, you must first understand it. Happily, the tide is the most predictable large thing on Earth.
Part II
A Tide Is a Sum of Sines
The tide looks like weather but behaves like astronomy. Because the moon and sun move in known orbits, their pull on the ocean can be written as a stack of pure cosine waves — the harmonic constituents — each with a fixed period set by celestial mechanics and an amplitude measured at each port.3 Victorian engineers summed them with brass gears and wire; we sum them with a for-loop. The prediction is the same, and it is startlingly good.
3 The names are a catalogue system, not poetry: M2 is the Moon's twice-daily (semidiurnal) wave, S2 the Sun's; K1 and O1 turn once a day. Hundreds more exist; four carry most of the water at Penhallow.Fig. 2 · illustrative dataPredicted tide at Penhallow Point · metres about mean level · synthetic constants
Forty-eight hours at Penhallow Point. Two highs and two lows a day — unequal twins, one tide always a little taller than its sibling. This curve looks organic. It is anything but.
Strip everything else away and here is M2, the principal lunar wave: a pure cosine with a period of 12 h 25 m, carrying most of the water. The tide's odd timetable — later by about fifty minutes each day — is simply this wave's signature.
Add S2, the sun's wave, at 12 hours exactly. It is a third the size of M2 and slightly out of step — and that small disagreement in period is a slow metronome we will meet again in a moment.
The once-a-day waves K1 and O1 are small, but they explain the unequal twins: they lift one high tide and lean on the other. Sum all four curves and the observed tide reappears, almost exactly.
Stretch the window and M2 and S2 drift in and out of phase, breathing a fourteen-day rhythm into the envelope: springs when moon and sun pull together, neaps when they argue.4 Ask any harbour: the calendar of the sea is written up there.
4 Springs lag the new or full moon by a day or two — mariners call the delay the age of the tide. Nothing to do with the season of spring; the water “springs up.”Part III
When High Tide Comes Ashore
Now put the two halves of this essay together. The tide is a fixed clockwork riding on a rising floor. Raise the floor by twenty centimetres and no single tide looks different — but the tallest tides, the springs that used to kiss the top of the quay, begin to step over it. Engineers call the result nuisance flooding: seawater in the car park on a calm, sunny afternoon, no storm anywhere in sight.5
5 Our working definition: any day the predicted-plus-observed level exceeds 0.55 m above mean higher high water — roughly the level of the old harbour road. A storm can do it; increasingly, arithmetic alone does.Counted per year, those quiet exceedances are the clearest translation of millimetres into consequences that we know how to draw.
Fig. 3 · illustrative dataDays per year above the minor-flood threshold · synthetic counts, method below
Each bar is one year; its height, the number of days the sea stepped over the old harbour road. For the first three decades the record is mostly silence — whole years without a single exceedance.
Average each decade and the staircase appears. Under one day a year through the 1960s; around three by the 1990s; into double digits now. No storm did this. The floor came up to meet the clockwork.
Sum the most recent decade and the count reaches 0 flood days 2015–2024, at one fictional harbour
Part IV
Eight Stations, One Direction
The sea is one; the land disagrees. Some coasts are still rebounding from the last ice age, others are sinking under drained marshes and groundwater withdrawal — so each station adds its own vertical motion to the global line.6 Here is the relative record at all eight of our stations since 1950, with each trend measured over the satellite era. The slopes differ. The direction does not.
6 Relative sea level = ocean rise plus land motion. A subsiding quay experiences twice the global rate; a rebounding one, almost none. Adaptation budgets care about the relative number.A Note on the Data
No measured data appears anywhere on this page. Penhallow Point is a fictional observatory, and every series above is synthetic — generated in your browser, on load, by the page's own JavaScript from a seeded random-number generator (mulberry32, seed 1861). The shapes are tuned to resemble the published literature on sea-level rise and tidal harmonics, so the essay can demonstrate honest data storytelling without borrowing anyone's measurements.
The century line (Fig. 1) is a quadratic trend with autocorrelated noise:
h(t) = −0.06·t + 0.0146·t² + n(t) mm, n(t) = 0.7·n(t−1) + ε, ε ~ U(−5, 5), t = years since 1900
The tide (Fig. 2) is a sum of four cosine constituents with real astronomical periods and invented local amplitudes and phases:
| Constituent | Meaning | Period (h) | Amplitude (m) |
|---|---|---|---|
| M2 | Principal lunar, semidiurnal | 12.4206 | 1.42 |
| S2 | Principal solar, semidiurnal | 12.0000 | 0.46 |
| K1 | Luni-solar, diurnal | 23.9345 | 0.15 |
| O1 | Principal lunar, diurnal | 25.8193 | 0.11 |
Flood-day counts (Fig. 3) draw each year from a noisy exponential, count(y) ≈ 0.55·e(y−1950)/21, floored at zero — a shape consistent with how exceedance counts respond to a linear rise in the mean. The eight station series (Fig. 4) are the century line rescaled by an invented vertical-land-motion term per station.
Charts are inline SVG, drawn and animated by hand-rolled vanilla JavaScript — no charting library. The full source is readable in this page. For how it was designed and critiqued, see the build guide.
Coda
The sea has no opinion about any of this. It keeps the moon's timetable, rides the rising floor, and files its report at our stilling well sixty times a second, the way it has since the observatory was young. The line will be a little higher next year. We will draw it again, and we will label it honestly.
— The observers, Penhallow Point