Professional Forecasts, Available for Everyone

PredictWind’s goal is to deliver the most accurate forecast data available. Having access to the world’s top forecast models enables users to be assured of the best possible forecast. With the top-ranked global models, along with the high-resolution PredictWind modeling and the best regional models, PredictWind offers unparalleled forecast accuracy.

Global Models


Regional Models


Outstanding Accuracy

PredictWind Forecasts are generated by proven technology with the highest resolution model on the web. The forecast models are exclusive to PredictWind!

Read about who uses PredictWind to get an insight from the experts.

PredictWind Tools

PredictWind is not just a source of the most accurate forecast data, but the best tools to help you make good decisions based on the forecast.

See All PredictWind Tools

Global Models


The ECMWF (European Centre for Medium-Range Weather Forecasts) model is highly regarded by Meteorologists and top Navigators around the world. The ECMWF HRES model consistently rates as the top global weather model from a national weather service with the highest rating scores. In March 2016 ECMWF increased the resolution of their model to a record breaking 9km resolution, which is currently the highest resolution global model available. ECMWF data has a very high acquisition cost, and this is why the data is not widely used by many weather websites, and has been traditionally used only by top yacht racing teams and meteorologists.


Spire uses the community MPAS model: Model for Prediction Across Scales. Spire uses a unique technique of measuring the earth’s atmosphere with 3 times more radio occultation data than any other commercial entity. This gives an advantage in forecast accuracy for remote locations. The PredictWind validation report proves that the Spire model is #1 for wind speed and direction accuracy using data from offshore weather buoys. It is #2 behind the ECMWF for land-based weather stations. The Spire Model output is 12km resolution globally. Learn more about Spire Modelling here.


The UKMO model, otherwise known as the “Unified Model” by the UK Meteorological Office has a long reputation as a market leader in forecast modeling. The validation study shows the UKMO has very similar accuracy to the ECMWF model offshore, and slightly behind the ECMWF & Spire models for the land-based weather stations. The UKMO model is used by many Met Offices around the world, and some of the biggest weather companies in the world. UKMO has a 15km resolution model output.


The GFS (Global Forecast System) Model from NCEP is the most widely used data source as it is free of cost, and whilst it is a good model it has a lower accuracy rating score. The model is run at 13km resolution, but only output at 27km resolution globally, which is not enough resolution to predict local thermal sea breeze and geographic effects.


PWG is PredictWind weather model that uses the NCEP global initial conditions for the model run. PWE is the PredictWind weather model that uses the ECMWF global initial conditions for the model run. Both the PWG and PWE are only available from PredictWind.

Regional Models

PWG High Resolution

The PWG high resolution modeling is highly accurate in the short term and provides detailed forecast data such and land and thermal effects, which uses NCEP initial conditions. 1km & 8km resolution in popular coastal areas.

PWE High Resolution

The PWE high resolution modeling is highly accurate in the short term and provides detailed forecast data such and land and thermal effects, which uses ECMWF initial conditions. 1km & 8km resolution in popular coastal areas.


NAM is one of NOAA’s major weather models, which in this case covers most of North America. NAM is a mesoscale model, which means that the numerical analysis is able to model land, and other features, at a higher resolution than in a global model, leading to improved forecast accuracy.


The HRRR is a NOAA real-time 3-km resolution, hourly updated, cloud-resolving, convection-allowing atmospheric model, initialized by 3km grids with 3km radar assimilation. Radar data is assimilated in the HRRR every 15 min over a 1-h period adding further detail to that provided by the hourly data assimilation from the 13km radar-enhanced Rapid Refresh.


Arome is a small scale numerical prediction model, operational at Meteo-France since December 2008. It was designed to improve short range forecasts of severe events such as intense Mediterranean precipitations (Cévenole events), severe storms, fog, urban heat during heat waves. This model is highly regarded by top racing navigators and beats the ECMWF forecast.

PredictWind Wave Models

wave-model-icon@2xPWG / PWE

The PredictWind wind forecasts (PWG/PWE) are used to drive the WaveWatch III wave model run by PredictWind to produce an accurate 50km resolution PWG & PWE wave forecast.


WAM model from ECMWF that is run at an incredible 14 km resolution globally, and currently rated the best wave model from any national weather centre.

Offshore App Models

For the Offshore App we also supply the Arome and Hirlam models for Europe, as these models are well known and highly regarded. This data is only available in the Offshore App.

Get the Offshore App

Other Models

Our goal is to provide the very best data, but not confuse customers with every weather model that is available. Many weather sites include the following models, as a long list of models looks impressive.

For example the CMC (GEM), and Hirlam, NavGeM models are good models, but do not have the accuracy of the 4 models we provide - ECMWF, GFS, PWG & PWE models. The GEFS or the Ensemble GFS model is only interesting if you are looking at forecasts beyond 10 days.

Detailed Forecasts for your Local Area

Most marine forecasts give you a generalised wind forecast in the morning or the afternoon. PredictWind shows wind weather updates every hour, with a wind speed and direction map for your entire local area. The detail you can see in the marine forecast can help you plan your trip or possibly help you win your next yacht race. Check our wind forecast coverage of your local area.

Forecast Confidence

By offering the leading global models, best regional models, and the high resolution PredictWind models, you can have a higher degree of confidence in the forecasts provided by PredictWind.

Our Process

Condition Files Entered in Model

The initial conditions files are entered into the model. The files can be comparable to a 'photographic' snapshot that contains the current state of the Earth's atmosphere. PredictWind uses two sources for these files (from the USA and Europe) so you have a comparative result to give you confidence in the forecast.

50km Forecasts Generated

Imagine dividing the atmosphere into 50km x 50km square grids over your local area. These initial conditions files step forward in time to create a 5 day forecast. For each 50km x 50km square region the model uses a complex equation to calculate the meteorological fields such as wind, temperature, humidity, pressure, rain and their interaction.

8km Forecasts Generated

This process is then repeated for the 8km model run, though it is 'nudged' by the 50km model run.

1km Forecasts Generated

Finally the 1km resolution model is completed, which is 'nudged' by the 8km model run. Significant computing resources are required to run the model in this configuration for all the PredictWind locations around the world.

Maps & Graphs Generated

Wind forecast maps and graphs and text are generated for the 20,000 worldwide forecast locations.

Repeat Every 12hrs

This process is repeated automatically every 12hrs for both initial sources (i.e. 4 forecasts every day). There is no historical data involved, so each 12hr forecast is independent from the previous.

Get started with accurate forecasts now

Model Physics

All weather models use different physics, set up parameters, and many complex equations to calculate meteorological data such as wind speed, direction, temperature, pressure, rain and cloud.

The model used is exclusive to PredictWind. It has been developed over 25years by a team of research scientists, and has some unique features including a stretched grid. This stretched grid is the framework for the 50/8/1km resolution modelling. This enables the model to run without any boundaries, and is thus less constrained than other models that usually require significant damping. This sophisticated approach often reflects considerable variability in wind speed and direction, and is an advantage over other models.