Where tide data comes from

Every number on this site comes from a public scientific source, most of it from instruments you can walk up to. This page follows that data from a sensor on a pier to the chart on your screen, explains how a tide can be predicted years in advance, and says plainly what each source is good for and where it falls short.

NOAA Tides & Currents

The tide predictions and observed water levels on this site come from the Center for Operational Oceanographic Products and Services (CO-OPS), the part of NOAA’s National Ocean Service that runs the National Water Level Observation Network (NWLON). The network has more than 200 long-term stations around the United States and its territories, and some have records more than a century long. San Francisco’s gauge at the Presidio has run since 1854, the longest continuous tide record in the Western Hemisphere.

This site follows 17 NOAA stations on the California coast, from Crescent City to San Diego. Each has a seven-digit id (Port San Luis is 9412110), which appears on every panel so you can check any number against NOAA directly.

What a tide gauge is

A modern NOAA water-level station is a small instrument shelter on a pier or seawall with a sensor looking down at the water. Many NOAA stations now use a microwave radar sensor that times a signal reflected off the sea surface. Older installations use an acoustic sensor that sends sound pulses down a tube inside a protective well, which damps out individual waves. Many stations also carry a backup pressure sensor, plus air and water temperature, barometric pressure and wind sensors.

The sensor samples about once a second. Every six minutes the station averages a three-minute window of those samples into one water level, which is why NOAA data, and the observed line on this site, comes in 6-minute steps. Waves and swell mostly average out; what is left is the tide plus anything that changes slowly enough to survive a three-minute average.

The station relays its data through NOAA’s GOES weather satellites, with cellular or internet links at some sites, and the readings are normally online within minutes.

Preliminary vs verified

The observed water levels you see here are preliminary: they pass automated quality checks (gaps, spikes, sensor disagreement) and are published at once. Weeks to months later, analysts review the record, fix datum shifts and sensor faults, and publish it as verified data. For today’s conditions preliminary is what everyone uses. For anything legal, engineering or scientific, use verified data from NOAA directly.

Stations without a live gauge

Five of the stations on this site (Half Moon Bay, Santa Cruz, San Simeon, Ventura and Newport Beach) have no real-time water-level sensor. For those, the console shows predictions only and the Tide Intelligence page points you to the nearest station with a gauge.

How a tide can be predicted years ahead

Weather forecasts fade after a week or two. Tide predictions are published years in advance and are usually right to within a few inches and a few minutes. The difference is that the tide is driven almost entirely by astronomy: the positions of the moon and sun, which are known with great precision far into the future.

Harmonic constituents

The gravitational pull of the moon and sun on the oceans can be broken into a set of regular waves, each with a fixed period set by astronomy. These are the harmonic constituents. Each has a short name:

• M2, the principal lunar semidiurnal constituent: the moon’s twice-a-day pull, period 12.42 hours.
• S2, the principal solar semidiurnal constituent: the sun’s twice-a-day pull, exactly 12 hours.
• K1 and O1, the main once-a-day constituents, which come from the tilt of the moon’s and sun’s orbits relative to the equator.
• N2, which follows the moon’s changing distance (perigee and apogee) over its elliptical orbit.
• Dozens of smaller ones, down to seasonal constituents like SA (the solar annual cycle, mostly the water warming and expanding in late summer and fall).

The periods are the same everywhere on Earth. What differs from place to place is how big each one is (its amplitude) and when it peaks (its phase), because the shape of the ocean basin, the continental shelf and the coastline change how the ocean responds. Those two numbers per constituent cannot be calculated from first principles for a particular harbor; they are measured.

Harmonic analysis

To measure them, NOAA fits a year or more of gauge readings with the sum of all those waves and solves for the amplitude and phase of each. Once those are known, predicting the tide is just adding the waves up for any future time, with small corrections for the 18.6-year cycle of the moon’s orbit (the nodal cycle). NOAA’s standard predictions use 37 constituents.

This is old science. Lord Kelvin built the first mechanical tide-predicting machine in 1872, a cabinet of pulleys and gears, one per constituent. The U.S. Coast and Geodetic Survey’s machine, nicknamed "Old Brass Brains", summed 37 constituents and printed U.S. tide tables from 1912 until computers took over in the 1960s.

What harmonic predictions cannot know is the weather. Wind, air pressure, river runoff and ocean warming all move the water, and none of them repeat on an astronomical clock. That gap is the residual, and it gets its own page.

Subordinate stations

NOAA cannot install a long-term gauge everywhere. For many places it publishes predictions as subordinate stations: a short gauge record (sometimes just a few months) was compared with a nearby reference station, and the result is a set of fixed corrections, so many minutes earlier or later for each high and low, and a ratio or offset for the heights. Santa Cruz, San Simeon and Ventura on this site are subordinate stations. NOAA gives them a high/low table but no hourly curve, so this site interpolates a smooth curve between the turns and says so on the Sources line.

Subordinate predictions are a little less precise than full harmonic ones, especially at the in-between hours. For most beach and harbor planning the difference is small.

How this site uses NOAA data

For each station the site requests, straight from your browser:

• Hourly predictions and high/low predictions for the coming days, used for the tide curve, the tide table and the activity outlook.
• Six-minute observed water levels and matching predictions for the last 24 to 72 hours, used for the "Now" reading and the residual.
• Water temperature, where the station measures it.
• Datums from NOAA’s metadata service: MLLW, MSL, MHHW and the rest, used for the datum ladder and the gauge on the Now panel.

Everything is requested in UTC and shown in Pacific Time. Responses are cached in your browser for 10 minutes, and if you install the site as an app the last good response is kept for offline use. Nothing is altered: if NOAA’s value and this site’s ever disagree, NOAA is right.

Open-Meteo: swell, wind, sea temperature and weather

Tide gauges measure water level, not waves. Swell, wind, sea-surface temperature and the weather in the SKY tile come from Open-Meteo, an open-source service that serves the output of national weather and ocean models through a free API.

• The Marine API provides significant wave height, swell height, period and direction (split into components), and sea-surface temperature, from global and regional wave models.
• The Forecast API provides wind speed and direction in knots, gusts, air temperature, sky cover, rain chance, UV and sunrise and sunset, from national weather models.

These are model forecasts, not measurements. A wave model sees the ocean on a grid of cells several kilometers across, so nearshore spots tucked behind points and kelp beds often see smaller waves than the model says, and exposed reefs bigger. Treat the swell numbers as an offshore reading of what is coming, not as wave height on your sandbar.

NOAA wave buoys

The Nearby sensors panel and the Coast Map’s buoy layer use measurements from the National Data Buoy Center (NDBC), including buoys operated by Scripps Institution of Oceanography’s CDIP program. A wave buoy is a floating sphere or disc with motion sensors that measure how it rises, falls and tilts on the swell. From that motion it computes the wave height, period and direction many times an hour; NDBC publishes waves hourly and wind every 10 minutes.

The site shows the two nearest buoys to your station and compares each one with what the model expected at the same spot over the last six hours. When they disagree by a lot, trust the buoy for what is happening now. For Port San Luis, the site also reads the buoy’s full wave spectrum to spot long-period northwest swell arriving at the outer waters and estimate when it will reach the coast.

Accuracy at a glance

For navigation, use NOAA’s official products, nautical charts and the U.S. Coast Guard’s notices. This site is a planning tool.