I spent years reading water before I ever wrote a line of code. So when I say this, it comes from both sides of me: the sea on a map and the sea in front of you are not the same object.
Myrtos Beach is one of the island's best-known beaches. White pebbles, steep cliffs, water that shifts from turquoise to deep blue in a few meters. On most summer days it looks like the safest place on Kefalonia. It is not always.
This is the story of the first marine layer in ARGOS, and of the honest limits of what a model can promise you about the sea.
The night it earned its keep
On the night the marine layer went live, the model logged an exposure score of 73.3 for Myrtos. Class: high. Window: 22:00 to 23:00 EEST. Waves from 316 degrees, about 2 m significant wave height.
Nothing about that sentence is measured at the shoreline. It is computed from Copernicus Marine model data, twice a day, for fourteen pilot beaches around the island. The nearest model cells sit 3.1 to 6.9 kilometers offshore, and the model speaks from there. What the beach does with that energy is a translation, not a measurement.
I want that sentence to sit in your chest for a second, because it is the whole piece.
The sea's shape
Within the ARGOS EMODnet window, the deepest cell is 3,841 meters. Our bathymetry comes from EMODnet at roughly 100 meters per pixel, with contours drawn down to 3,500 meters. Kefalonia's west coast is where that deep water leans closest to the shore.
Along Myrtos' open-water bearing of 291 degrees, the corrected EMODnet contours put 25 m of water about 2.1 km from the shoreline, 50 m at 3.0 km, and 100 m at 4.2 km. The 200 m contour lies about 10.0 km offshore. The beach's assigned CMEMS cell sits 3.36 km from the beach centroid, between the 50 and 100 m contours; at roughly 100 m per pixel, this is shelf-scale context, not swim-zone bathymetry.
That geometry matters because it is why a beach can look like a swimming pool and behave like something else. The energy does not need a storm to arrive. It needs a direction, a period, and a floor that shoals the right way.
The layer
Fourteen beaches. Copernicus Marine Mediterranean forecasts: waves and two-dimensional surface currents, refreshed twice daily. The exposure score is a direction-gated wave score at 70 percent plus a current score at 30 percent. The gate exists because a wave from the wrong direction is noise, and from the right direction it is the whole story.
The bathymetry you just read is context, not an input. Our exposure score does not use the EMODnet shelf layer directly; the CMEMS models carry their own model-resolution bathymetry. It knows the sea state. We supply the geography.
What we could not validate, so we are saying so
There is no independent wave buoy on Kefalonia. The published POSEIDON network has no Kefalonia station; its named Ionian mooring is Pylos. We checked. The answer is no.
So this layer ships unvalidated, first in the validation queue. The interim path is cross-checking against KefaloniaPress daily beach scores and the quality documentation that ships with the Copernicus products themselves. The first version of this story had a cleaner ending. This is the honest one.
What it is not
Not a rip-current warning. Not measured beach safety. Not a lifeguard-grade forecast. Not a safe-or-unsafe certification for any beach, ever.
It is an exposure screening: modelled offshore wave and current conditions, translated into a beach exposure assessment. Direction-gated, twice daily, bilingual, and public about its own uncertainty.
The ask
If you are a lifeguard, a boat operator, or someone who has watched Myrtos for thirty years: your observations are the validation this layer is missing. The map will only get as honest as the ground truth it is fed.
Built in Kefalonia. Watching over the places we call home.
Links:
https://argos-geo.org/map · https://github.com/argos-geo/kefalonia-digital-twin
Top comments (0)