Better, More Accurate Wind Forecasts

How Spire Weather ranked #1 in wind forecast accuracy

PredictWind success story

PredictWind provides the maritime sporting and leisure community with high-quality ocean forecasts. The company offers its community two highly-detailed proprietary weather prediction models using the ECMWF and GFS global weather forecast models. Seeking to provide even more accurate forecasts, PredictWind tested Spire’s model in an effort to better predict wind speed and direction. The Spire-based model ranked #1 for overall accuracy.

The challenge

PredictWind offers an algorithm-based tool built on weather data to equip the maritime sports and leisure sailor with accurate and detailed maps, weather routing solutions, departure planning guides, forecast alerts, and live wind observations. As one of the top-ranked global weather model apps for sailors, PredictWind is always looking to provide the best wind predictions to their customers.

The solution

PredictWind created an objective, in-market study to assess the accuracy of several different weather forecast models. Over the course of four months, spanning February to June 2020, the PredictWind team compared actual wind observation data from across global locations to assorted weather models. These models included data from ECMWF, UKMO, GFS, and Spire.

The goal was to capture enough data points over time, including periods with extreme weather events and weather volatility, to gauge how well models could adapt to weather fluctuations and provide better, more accurate wind forecasts.

The results

PredictWind’s comparison ranked Spire as #1 for wind speed and wind direction prediction accuracy. The vertical resolution provided by Spire’s radio occultation data, combined with the 10,000+ weather data points collected per day via its proprietary satellite constellation, offered the most comprehensive, unbiased global data set resulting in improved forecasting.

Because Spire’s weather data spans from the ocean surface up to 120 km in the atmosphere, weather events are observed as they form at very high altitudes. This unique insight means not just short-term but medium-term forecasts become more accurate. This further level of granularity helped PredictWind create better longer-term models for their sailors.

Let’s take a look at the tested metrics:

Wind speed

This chart plots the wind speed mean absolute error in knots for Spire, ECMWF, UKMO, and GFS. Lower values are better. Spire’s model consistently has a lower mean absolute error (in knots) compared to the other test models over the course of a 7-day period.

Wind direction

This chart compares the mean absolute error, this time in degrees, for wind direction. Again, lower values are better. Spire’s weather forecast model ranks #1 for wind direction prediction accuracy for each day tested.