← Back to the mapSARGASSUM PUNTA CANA

Our forecasts, verified

No promises — measurements. Every day, yesterday’s forecast is compared with what the satellite actually observed. Here is the method, and the numbers — including the ones we miss.

Our forecasts, verified

No promises — measurements. Every day, yesterday’s forecast is compared with what the satellite actually observed. Here is the method, and the numbers — including the ones we miss.

100% correct "clean water" forecasts · 21 checks · calm season · 100% across all regimes · rare alerts flagged low-confidence
The method

How a forecast is made

Copernicus / NOAA satellite, 4 passes a dayThe AFAI floating-algae index is sampled offshore of every beach, pixel by pixel — never a regional average.
A drift model, not a crystal ballBeaching persistence (5-day half-life), wind and offshore mats: the 7-day forecast is recomputed at every satellite pass.
An automatic backtest, every single dayEvery forecast is archived, then checked against the satellite observation on the target day. Fully automatic — nobody edits the numbers.
When data is missing, we show nothingNo invented figures: if a measurement does not exist, the section disappears rather than guessing.

100% of our "clean water" forecasts (calm season, 2026-07-27 → 2026-07-28) proved correct (over 21); the rare alerts stay flagged low-confidence until the data confirms them.

100%
"Clean" Forecasts
Correct at 100% (21 samples) · 100% all-season

Period: 2026-07-27 → 2026-07-28

Accuracy, measured

How good are our forecasts

100% of our "clean water" forecasts (calm season, 2026-07-27 → 2026-07-28) proved correct (over 21); the rare alerts stay flagged low-confidence until the data confirms them.

From July 27 to July 28, 2026, 21 forecasts were checked against satellite observations across 21 beaches.

Accuracy is measured on our longest-running network — 20 Caribbean beaches (Martinique & Guadeloupe), where the verification archive is deepest. The exact same model (pipeline v3) powers the Punta Cana forecasts.

100%
correct status 1 day ahead
100%
across all horizons

A forecast counts as a “hit” when the predicted status (clean, moderate, avoid) matches the status the satellite observed on the target day. Mean absolute error on the AFAI index: 0.024.

HorizonHit rateChecksDisplayed confidence
Day +1100%2160%

The model also publishes its own confidence: high at day +1, low at day +6. When it is low, read the forecast as a trend, not a certainty.

When a beach is already loaded, we keep it 'red' until data (a clear satellite drop or a ground report) confirms it cleared: beached sargassum doesn't leave without collection. A conservative choice — it can make us over-alert on a beach that was cleaned in the meantime.

What we miss

21 of 21 beaches score above 90%. The hardest ones to predict — published anyway:

  • Grande Anse d'Arlet (Martinique) — 100%
  • Anse Mitan (Martinique) — 100%
  • Anse Noire (Martinique) — 100%

Exposed beaches whose state hovers around a threshold (clean / moderate) remain the hardest to predict.

Freshness

The data you are looking at

Latest satellite measurement July 28 at 02:05 PM UTC ·
12 monitored sampling points · data re-checked 4 times a day · this page is regenerated with every update.
See the live map →
Get the morning verdict
The same satellite measurement, daily in your inbox. Free.
You're in! First email in 3 days.
We watch the sea for you.
SARGASSUM PUNTA CANA · Data: Copernicus Marine & NOAA ERDDAP · backtest recomputed automatically