Island Lab · Pollica · Cilento Coast

The local digital twin.

Public data, corrected by local truth. A working model of one bay, honest about what it knows.

A walkthrough · use → ← or click to move

Why

The sea gets decided on numbers that are not about your site.

Why · the stakes

A third of the Mediterranean's meadows, gone in fifty years.

14–20 L

oxygen per m² each day from a healthy meadow

~700 t

carbon per hectare locked in the seabed matte

~1 cm

growth per year. Lost, it does not come back.

Posidonia oceanica is the lungs of the Mediterranean. Because it cannot be replanted on any human timescale, protection is the only cheap option, and protection means seeing stress before it becomes loss.

Why · the gap

Most decisions about the sea run on borrowed numbers.

A regional model describes a wide area. A survey was taken on another day, miles from the water that matters. A nearshore bay can sit several degrees off the regional average. The decision gets made on the gap between the two, and most of the time no one ever sees the gap.

Why · the tradeoff

Earth observation already solved this.

One sharp satellite sees fine detail, rarely. A swarm of coarse ones sees everything, daily. You use them together: a broad, cheap sweep finds where something is changing, and cues a scarce, precise instrument to look closer. Tip and cue. We run the same loop at the scale of a single bay.

How

A local digital twin, honest about itself.

How · reconciliation

Take the global model. Bend it toward what the water actually measured.

Not sensor fusion: data assimilation and downscaling. A coarse public model, continuously corrected by local observation, confident right at each instrument and honest about the guess elsewhere. The difference between the regional prediction and the local reality, the delta, is the product.

How · why local diverges

Everything that makes your cove yours happens below the model's grid.

The regional grid cell is kilometres wide. The shape of the seabed, tidal mixing, a river mouth, a day of sun and slack wind layering warm water over cold: the model averages it all away. Oceanographers call these sub-grid processes. For a meadow, they are the whole story.

How · the architecture

Two twins: the water, and the life in it.

The water twin

Models the conditions at the meadow: temperature, light, clarity, downscaled from public data to this exact place.

The living twin

A mechanistic model of the plant itself, turning those conditions into health, stress, and a forecast. The actual biology, not a correlation.

How · the living twin

Heat kills by mortality. Darkness kills by starvation.

Above about 28 °C shoots die even in clear, bright water: a fever. When runoff dims the water, photosynthesis stops and the meadow drains its reserve battery: a famine. The model tells the two apart, names the dominant stressor, and forecasts where it is heading.

How · honesty

Every value carries how much to trust it.

Some uncertainty is the ocean's own randomness. Some is just how stale the data is and how well the model has been tested. We surface both, as a trust score and a confidence band, and we tag every number by where it came from:

Local our sensor Open public data Derived our model Synthetic placeholder

How · adaptive sampling

The twin knows where it is blind.

The zones where the model is least sure and the meadow matters most are exactly where one real sensor buys the most truth. The free public-data twin is the sweep. Your sensors are the precise instrument. The twin decides where they earn their keep.

How · the horizons

From now, to eight days, to the season.

A nowcast of the present, a short-range forecast to about eight days, then subseasonal and seasonal outlooks. Climatology, the long-term normal, is not the forecast; it is the baseline the forecast is judged against. The anomaly, the departure from normal, is the story.

What

Infrastructure, handed to a community.

What · the path

Start on public data, free. Sharpen sensor by sensor.

TIER 0

Public baseline

A real twin on public data, at no cost. Honest about its limits.

TIER 1

Ground-truthed

One buoy at the meadow. The twin sharpens around it; numbers turn measured.

TIER 2

Operational

A fleet, routed to where information is worth most. Forecasts you can run on.

What · the working example

Pollica, today.

The water twin, the meadow map, the action layer, the live fleet tracker, and the Posidonia model underneath, all running on public data and honest about every number. Not a report. The start of a project a community owns.

See it live: the project dashboard →

The short version

The global model tells you what is probably happening across a region. The local twin tells you what is actually happening at your site, and how much to trust it.

Island Lab · Pollica