Despite the intense interest in local weather and forecasting, there has never been an objective way to determine the best and most reliable forecast from the large number of available forecasts from the National Weather Service, local TV and radio stations, the Weather Channel, etc. Besides these professional forecasting sources, it's possible that meteorologists who do not forecast for a living (professors, researchers, etc.) or motivated amateurs may be undiscovered sources of valuable local forecasts.

Our forecasting contest provides a forum so that forecasts from many different sources could be systematically compared over a period of months or years. This may reveal the best human forecasters, and/or it may reveal that automated statistical forecasts such as the various NWS MOS products may be difficult to beat on a consistent basis for some locations.

It has been found that a forecast formed by some simple combination of several independent forecasts often outperforms any member of the set. We will test various methods of combining forecasts to generate the best possible local forecast for each location.