Frequently Asked Questions

When does the NCWF forecast show the most skill?

The NCWF forecasts are based on storm extrapolation. The forecasts are LIMITED to extrapolation of regions of existing storms. Therefore, areas of storm initiation, growth, and decay are NOT captured. Regions of short-lived isolated storms are often not well forecast. Forecast skill is best for long-lived linear multi-cellular storm systems. The quality of the forecast can be accessed in real-time by using the "performance" option. A complete set of documentation is available under "product description".

Why don't all storms have forecast? 

The NCWF algorithms only provide forecasts for storms that meet specific criteria. These criteria are primarily based on area coverage. An elliptical filter (designed by MIT/Lincoln Laboratories) is used to remove perishable scales. The concept is to 1) only forecast for storms that are likely to persist for an hour and 2) not to fill an image full of small polygons that don't add meaning to the product.
Occasionally a large storm will not contain a forecast. This may be due to insufficient storm history (an isolated storm has to be identified and tracked for 40 min before a forecast is shown) or a poor motion vector. The algorithm automatically checks to make sure that motion vector is not too fast and that the direction of motion is reasonable. Poor motion vectors are eliminated and a nearest neighbor motion vector is used. However, when the storm is isolated and a nearest neighbor motion is not available then the storm will not contain a forecast.