Unfortunately, there is no single "golden rule" for deciding which weather model will be the most accurate. Forecast performance can vary depending on the location, weather situation, and forecast period. One model may perform particularly well for a period of time, while another may provide a better forecast on a different day.
For this reason, we recommend comparing multiple forecast models rather than relying on just one.
If the models are showing a similar forecast, you can generally have more confidence in the expected conditions. If the models differ significantly, this indicates greater uncertainty in the forecast and is a good reason to continue monitoring subsequent forecast updates.
High-resolution forecast models
PredictWind provides access to a range of high-resolution forecast models, including PWG, PWE, AROME, NAM, HRRR, UKMO and ICON. These models provide additional detail, particularly in coastal areas and locations where local terrain has a significant influence on the wind. For a High resolution coverage map please see https://forecast.predictwind.com/coverage-map/
PWG and PWE are PredictWind's proprietary high-resolution weather models. They are available at resolutions down to 1 km in selected areas and are particularly useful for identifying local variations in wind caused by coastlines, terrain and thermal effects.
We also provide several high-resolution models produced by leading national weather agencies:
AROME – produced by Météo-France, with forecasts at approximately 1.3 km resolution over France and surrounding parts of Western Europe. It is designed for detailed, short-range forecasting.
UKMO – produced by the UK Met Office. In addition to the global UKMO forecast, a 2 km high-resolution model is available across the UK and Ireland. This provides additional detail for local effects such as wind funnelling, thermals and wind shadows.
ICON – produced by the German Weather Service (DWD). ICON is available globally, with higher-resolution forecasts over Europe. The 7 km ICON model provides detailed European coverage, while the 2 km ICON-D2 model provides very high-resolution, short-range forecasts over parts of Central Europe.
NAM – produced by NOAA and provides high-resolution regional forecasting across North America. It is useful for resolving smaller-scale weather features and the effects of terrain.
HRRR – also produced by NOAA, this 3 km model provides very high-resolution, short-range forecasts across the United States and is updated frequently.
Because these high-resolution models have different geographical coverage and forecast durations, the models available to you will depend on your location and how far ahead you are looking.
High-resolution forecasts can be particularly valuable in coastal areas and around complex terrain, where hills, mountains, valleys and coastlines can have a significant effect on local wind conditions. However, higher resolution does not guarantee that a model will always be the most accurate. Forecast models use different data and modelling techniques, and their performance can vary depending on the weather situation. This is why we recommend comparing all the models available for your location rather than relying on a single forecast.
Compare the models
PredictWind gives you access to multiple global and regional forecast models, allowing you to compare different forecasts and identify where there is agreement or uncertainty.
You can view independent model validation and accuracy results here:
With only one or two models, it can be difficult to know how much confidence to place in a forecast. Having access to multiple models makes it much easier to identify a consensus between the forecasts or recognise when there is significant uncertainty.
For an explanation of the different weather models available in PredictWind, please see:
Check forecasts against observations
Another useful way to assess model performance is to compare the forecasts with live weather observations in your area. This can help you see which model is currently tracking the observed conditions most closely.
When you are on passage, we recommend updating your Weather Routing at least twice a day so that your route uses the latest available forecast data. Continue comparing the models as conditions develop rather than relying on a single forecast from earlier in your passage.
Your own observations are also an important way to assess which model is performing best. Compare the forecast conditions with what you are actually experiencing onboard, using your boat instrumentation to check true wind speed and direction, barometric pressure and other available weather data. You can also compare the models with live weather observations from nearby stations where available.
By regularly comparing the forecast models with the conditions you are actually experiencing, you can build a clearer picture of which models are tracking the current weather situation most accurately and use this information when interpreting the forecast ahead.