THE SHORT VERSION
A surf check should feel effortless.
The work underneath it is anything but.
PureSurfers brings ocean models, swell components, coastal observations, tide predictions and a growing library of places into one readable atlas. The aim is practical: help you decide where to look, when to go, and which questions still need a look at the ocean.
We build on extraordinary public science. Our contribution is the engineering between those sources and a useful surf check: reconciling times and units, preserving distinct quantities, managing shared data, exposing uncertainty, and turning a coast’s worth of information into something you can actually use.
MORE THAN ONE VIEW OF THE OCEAN
Three model families.
Room for disagreement.
The atlas compares NOAA GFS-Wave, ECMWF WAM and Météo-France wave guidance. Their agreement is useful; their disagreement is useful too. We retain the differences so you can see when the picture is less settled.
Those model families normally reach us through a shared Open-Meteo gateway. Different model origins do not mean three independent delivery systems. A direct NOAA route provides an alternative for GFS where supported; it does not create a fourth model.
The range across models is a range of outputs, not a statistical confidence interval. We also distinguish a model’s sampling cell from the beach pin. Several nearby breaks can share regional input without pretending there is a separate sensor at every takeoff.
Source: Open-Meteo marine model documentation ↗READ THE INGREDIENTS
The ocean doesn’t arrive
as one swell.
A useful long-period swell can be hiding underneath a larger, less useful sea. “Swell by swell” separates the height, period and incoming direction of the components a source actually supplies. You can inspect the ingredients instead of relying only on a combined number.
Component availability differs by model. An absent partition stays absent. And because “primary” and “secondary” can swap as their relative strength changes, the newer swell-matching logic compares direction and period rather than blindly following a rank label.
A resolved Trestles issue illustrates the work: one captured feed reported zeros through its entire partition forecast while another reported swell. We added a visible “all-zero feed unconfirmed” state and kept the available alternative readable. Real calm remains zero; uncertainty gets its own label.
CLOSER TO THE COAST
A second layer of evidence.
A different kind of number.
CDIP / Scripps adds buoy-driven nearshore wave-model context. Nearby NOAA buoy observations add an independent record of what an instrument measured at its own location. We display source times, quality information and same-source comparisons with roughly a day earlier.
These layers answer different questions. A nearshore model output is not a sensor on the reef. A buoy offshore is not a measurement of a breaking wave at your beach. We keep those identities visible instead of blending everything into an impressive-looking but ambiguous total.
Model-cell Hs
Significant wave height in a model cell. Useful regional evidence; not a promised face height.
Breaking-wave scenario
An experimental transformation under stated assumptions. Not a camera observation or a maximum set.
Observed surf
What someone actually saw, where and when. Our local logs separate regular faces from the largest set seen.
EXPERIMENTAL / FOUR RINCON PROFILES
The bottom changes
the question.
At Rincon, the newer experimental calculation follows an extracted bottom profile from a CDIP sample to a published spot coordinate. Tide predictions and bottom elevations share the same vertical reference, NAVD88, so the modeled water depth changes coherently with the tide.
Wave period and direction feed a one-dimensional shoaling and refraction scenario. If a wave reaches the model’s depth limit, the calculation retains that energy loss farther along the profile. Larger-wave inputs are transformed separately. There is no universal “low tide means 20% smaller” switch.
The less visible work matters just as much: inspecting survey footprints, checking missing terrain, tracing the actual input source, and invalidating a profile when its published pin moves.
A one-meter grid isn’t one-meter knowledge.
Our inspected Rincon grid includes historical interpolation and roughly 10 m source terrain. We disclose that. The output is an experimental wave height at the pin, not a calibrated breaking-face prediction. Four Rincon entries have profiles; the rest of the coast does not inherit a made-up correction.
Two-dimensional reef focusing, headland wrapping, moving sand and field calibration remain work ahead. A famous break does not get an automatic numerical boost because its name sounds right.
Source: NOAA / USGS Southern California CoNED metadata ↗Source: NOAA tide products and vertical datums ↗THE ENGINEERING YOU SHOULDN’T HAVE TO NOTICE
Less waste.
More traceability.
Shared caches let many surf checks reuse the same upstream work. Requests for the same data are combined where supported, concurrency is bounded, selected spots take priority, and provider cooldowns are respected. The new bottom profiles are processed ahead of time; visitors do not download giant seabed rasters.
Missing data stays missing. Stale sources are labeled. A current nearshore reading cannot silently fill tomorrow’s forecast. Extra decimal places preserve detail in the output; they do not establish extra measurement accuracy.
We test numerical behavior, source parsing, fallback states, mobile layouts and interactions. Those checks help establish that the software behaves as intended. Forecast accuracy requires a separate test against the ocean, across sessions and swell events.
WHAT EARNS TRUST
Show the work.
Keep checking the ocean.
Our ambition is a better surf check, supported by evidence you can inspect. We will keep adding capability—and keep saying what each new capability can actually tell you.