Wind speed
Temperature
Waves
Distance
Rider weight
Sizes every kite, wing, sail and board we suggest. No account needed.
Hides rows in the comparison table. The mean, agreement and score always use every model — filtering the view never changes the forecast. What are these?
A weather model carves the atmosphere into a 3D grid and steps the physics forward hour by hour. No single one is right — so we ask 14 of them the same question and show you all the answers side by side.
The width of one grid cell. A 25 km model averages your whole bay into one number and will miss a sea breeze; a 2 km model can see the headland. Fine grids cost compute, so they cover small areas and run only 2–3 days out.
How often the model re-runs with fresh observations. An hourly model reacts to this morning's reality; a 12-hourly one may still be showing yesterday evening's idea of today.
The row under the table. When the models cluster, the forecast is worth planning around; when they scatter, treat any single number as a guess. We measure it as the spread of the model winds, so one outlier cannot condemn an otherwise solid forecast.
Run for the whole planet, so every spot gets all of them.
NOAA, United States
America's workhorse. Long range and frequent updates, but a 13 km grid smooths out coastlines, so it tends to under-read thermal sea breezes.
ECMWF, Europe
Widely rated the most accurate global model beyond day two. Coarse grid, so trust it for the pattern rather than the exact strength at your beach.
ECMWF, Europe
ECMWF's machine-learning model: it learned from decades of past weather instead of solving the physics. Fast and surprisingly good — useful as a second opinion, not a lone source.
DWD, Germany
The German service's chain, and the strongest all-rounder over Europe because the nested grids get down to 2 km near the coast.
Environment Canada
Canada's global model. Independent of the American and European chains, which is exactly what makes it worth having in the average.
Met Office, United Kingdom
Met Office Unified Model. Strong on Atlantic systems, and its 2 km UK grid handles island and channel effects well.
High-resolution national models. They only cover their own patch, so they appear on a spot page only where they actually reach.
Météo-France
AROME is one of the sharpest models in Europe: at 1.5 km it actually resolves sea breezes and valley winds. Short range, so it matters most for tomorrow.
DWD, Germany
The high-resolution core of the German chain over the Alps, the North Sea and the Baltic. Short range, high detail.
KNMI, Netherlands
Dutch HARMONIE-AROME. Hourly re-runs make it the best short-notice call for the North Sea coast.
DMI, Denmark
The Danish HARMONIE run, covering the Baltic and North Sea up to Iceland — prime Scandinavian kite water.
MET Norway
The finest grid we carry. 1 km over Norway, Sweden, Denmark and Finland, where fjords and islands wreck coarser models.
Arpae, Italy
Italy's ICON-2I over the central Mediterranean — the local read on Tramontana and Mistral days.
JMA, Japan
Japan's chain. The global half is coarse; the 5 km MSM is what makes it worth asking around Japan and Korea.
Bureau of Meteorology, Australia
Australia's ACCESS-G. The local authority on the Southern Ocean systems that drive the Australian season.
Open the units menu in the header (the kn button) and expand Forecast models to hide the ones you do not care about. That only trims rows from the table — the mean, the agreement reading and the rideability score always use every model available for that spot.
meteoblue is a commercial service with its own paid API, and ALADIN is not published as open data by the services that run it. The closest equivalent here is Météo-France's AROME, built on the same modelling family and included above.
Model data via Open-Meteo, which publishes the full specifications for every chain listed here. Figures above are theirs.