How many NOAA tide-prediction stations were tested?

3,499 non-Great Lakes entries from NOAA's tide-prediction directory were tested for the 2026 snapshot.

How many listed stations had no tide predictions?

5 directory entries returned no high/low prediction series in the initial sweep and in all 4 spaced confirmation probes.

Does an unresolved request count as a missing station?

No. All 48 unresolved requests stayed unclassified and published. Only an explicit no-predictions result repeated in every confirmation can enter the missing set.

Does this audit measure observed water levels?

No. It measures astronomical tide-prediction availability. Observed water levels, meteorological sensors, currents, and station operating status are separate NOAA products.

Can the station list change?

Yes. NOAA can add, remove, or reclassify stations and revise available prediction years. The generation date belongs with every result, and a new snapshot should be generated before using the data as a current directory.

Perigee data brief · NOAA predictions

NOAA tide prediction station coverage audit, 2026

Perigee tested 3,499 non-Great Lakes entries in NOAA's tide-prediction directory. Only 5 returned no high/low series in the sweep and all 4 spaced confirmation probes. The 48 requests that could not be resolved were kept out of the missing set.

Analysis by Ryan Cardin · Snapshot September 14, 2026 · Schema v1

Directory stations tested

3,499

Non-Great Lakes tide-prediction entries

Confirmed listed gaps

Unanimous after 4 spaced rechecks

Unresolved, retained

Never interpreted as missing data

Legacy route gaps

Preserved with one-hop station redirects

Download CSV Download versioned JSON Reproduce the audit

Listed stations with no confirmed prediction series

These entries appeared in NOAA's tide-prediction directory but did not return a three-day high/low series in the sweep or any confirmation. This says nothing about nearby stations or other NOAA products.

NOAA tide-prediction directory entries with no confirmed prediction series
NOAA station	ID	State or territory	Confirmations
MALAKAL HARBOR	1841275	Not provided	4
Carolina Forest, ICWW	8660754	SC	4
Eugene Island	8764311	LA	4
Big Salt Lake	9450623	AK	4
Fish Bay, St. John	9751388	PR	4

What the result does—and does not—mean

NOAA's directory is a capability index, not a guarantee that every station will return every date range indefinitely. A listed station can be decommissioned or temporarily inconsistent with the prediction API. Conversely, an observation-only or meteorological station can remain useful even when it has no harmonic tide series.

This audit therefore makes one narrow claim: whether a specific NOAA ID returned astronomical high/low predictions under the published request and confirmation procedure. It does not score station accuracy, measure API uptime, infer flooding, or treat predicted water level as an observed gauge reading.

View 25 confirmed legacy annual-route gaps

This separate cohort contains previously observed annual-table URLs whose station IDs are absent from the current tide-prediction directory. Each route redirects once to the still-useful canonical station page.

Legacy annual tide-table routes confirmed without predictions
Station	ID	Other NOAA products	Canonical destination
Robbins Reef, NJ	8530973	met	Station page
South Craney Island, VA	8638595	met	Station page
Sparkman Channel Entrance, FL	8726671	met	Station page
Fort Morgan, AL	8734383	met	Station page
Pinto Island Visibility, AL	8737005	met	Station page
Carrollton, LA	8761955	waterlevels, met	Station page
West Bank 1, Bayou Gauche, LA	8762482	waterlevels, met	Station page
Harbor Island Visibility, TX	8775223	met	Station page
UTMSI Visibility, TX	8775236	met	Station page
Texas State Aquarium Visibility, TX	8775302	met	Station page
S. Bird Island, TX	8776139	waterlevels, met	Station page
Baffin Bay, TX	8776604	waterlevels, met	Station page
Port Mansfield, TX	8778490	waterlevels, met	Station page
Angels Gate, CA	9410647	met	Station page
Los Angeles Pier 400, CA	9410666	met	Station page
Pier 17 Visibility, San Francisco Bay, CA	9414296	met	Station page
San Francisco Pier 1, CA	9414311	met	Station page
Oakland Berth 38 Visibility, CA	9414797	met	Station page
Point Potrero Richmond, CA	9414847	met	Station page
Davis Point, CA	9415141	met	Station page
Yaquina Bay, Highway Bridge, OR	9435372	met	Station page
Isabel Segunda, Vieques Island, PR	9752621	waterlevels	Station page
Salinas, Bahia de Jobos, PR	9755968	waterlevels	Station page
Arecibo, PR	9757811	waterlevels	Station page
Aguadilla, Crashboat Beach, PR	9759413	waterlevels	Station page

Methodology, provenance, and reproduction

Fetch NOAA's official tide-prediction station directory and remove Great Lakes entries, whose water-level behavior is not a harmonic ocean-tide series.
Request three days of interval=hilo predictions for each station using MLLW, English units, and local standard/local daylight time. Retry transport and ambiguous responses five times.
Treat every timeout, malformed reply, and ambiguous NOAA response as unresolved. If more than 5% of the directory stays unresolved, or more than 2% appears missing, abort without replacing the checked-in snapshot.
Recheck each candidate 4 times, waiting 45 seconds between probes. One valid series vetoes the missing classification; only unanimous explicit no-predictions replies are retained.
Confirm that a fully excluded directory entry does not separately publish observed water levels. Observation-bearing legacy stations keep their canonical station pages and only unavailable annual routes redirect.

Run npm run seo:station-prediction-coverage from the versioned generator source. The command queries the official NOAA CO-OPS Data API, applies the fail-closed gates above, and writes the JSON snapshot used by this page. NOAA's tide prediction help explains the underlying harmonic and subordinate-station products.

The CSV contains all 31 confirmed exception records with probe settings, source URLs, snapshot timestamp, and schema version. The JSON preserves the aggregate unresolved count and route-policy evidence. Neither download is a live directory; cite the snapshot date with the data.

Questions about the station coverage audit

How many NOAA tide-prediction stations were tested?: 3,499 non-Great Lakes entries from NOAA's tide-prediction directory were tested for the 2026 snapshot.
How many listed stations had no tide predictions?: 5 directory entries returned no high/low prediction series in the initial sweep and in all 4 spaced confirmation probes.
Does an unresolved request count as a missing station?: No. All 48 unresolved requests stayed unclassified and published. Only an explicit no-predictions result repeated in every confirmation can enter the missing set.
Does this audit measure observed water levels?: No. It measures astronomical tide-prediction availability. Observed water levels, meteorological sensors, currents, and station operating status are separate NOAA products.
Can the station list change?: Yes. NOAA can add, remove, or reclassify stations and revise available prediction years. The generation date belongs with every result, and a new snapshot should be generated before using the data as a current directory.

Use the right station and data product

Start with the NOAA tide-prediction methodology guide to keep predictions, observations, station types, datums, and freshness separate. Developers can query the live directory through the Perigee REST API and MCP server.