Weather Forecasting Technology

WeatherPredict Inc. is an exclusive commercial provider of weather forecasts derived from the patented Superensemble™ (SE) process. SE technology was originally developed at Florida State University, and it serves as the backbone of WPC’s operational forecasting system.

Originally developed as a tool to generate accurate hurricane track predictions, the SE Forecast process is also used to generate gridded and city specific weather forecasts for many atmospheric variables, including temperature, precipitation, wind speed, wind direction, moisture, and isobaric heights.

The SE technique is used to gather valuable predictive information from some of the best available weather model forecasts, and further, to combine that information to generate a forecast that has superior accuracy. SE forecasting is a computational, statistical process that surpasses the predictive power of individual dynamical models and simple ensemble averaging techniques. In fact, SE weather forecasts have been shown to achieve significant enhancements in skill when compared to the forecasts that serve as inputs to the process.

A SE forecast is completely objective. That is, the forecasting technique does not include human forecasts. Because of this, SE forecasts are ideal when weather forecasts must serve as inputs to other statistical models, e.g., crop yield forecasts, electricity load forecasts, and consumer demand forecasts. Also of importance, SE forecasts and their associated data sets form an objective basis for probabilistic forecasting. Probabilistic forecasts add value to decision making processes, particularly when examining risk associated with a weather hazard.

SE forecasting requires the management of a large database of weather information. To run SE forecasts, WPC subscribes to, generates, or otherwise acquires raw model outputs from some of the world’s foremost meteorological agencies. WP maintains the model forecasts in an ever-growing database of input model forecasts. Additionally, WP maintains a growing database of observed, ground truth meteorological information. In the schematic below, one sees that historical weather observations and historical weather forecasts must be managed in near real-time.

Superensemble™ Technology

The Superensemble™ technique is the fundamental component of our forecasting process


It is important to note that the Superensemble™ is not a simple averaging technique. In many cases, Superensemble™ forecasts correctly lie outside the range of input model forecasts. In Figure 1 below, one sees an example of the Superensemble™ Forecast for Hurricane Ivan on 09/09/2004 at 18:00 UTC. Note that the red WPTCF track forecast is correctly positioned outside of the envelope formed by input model track forecasts. The official hurricane track is displayed in black.

Superensemble Graph

Click image for larger view