Preprints, 16th AMS Conference on Weather Analysis and Forecasting Phoenix, Arizona, 11-16 January 1998, J24-J27 J2.4 VALUE OF WEATHER FORECASTS FOR ELECTRIC UTILITY LOAD FORECASTING

Weather forecasts can be evaluated in a variety of ways. Murphy (1993) defined three kinds of forecast “goodness”: consistency, quality, and value. Consistency refers to the relationship between the forecasts and the “true beliefs” of the forecaster, quality refers to the relationship between the forecasts and weather events, and value refers to the relationship between the forecasts and the benefits or losses accrued by users. Most forecast evaluation studies have focused on the quality question, in large part because it is, in some sense, the most tractable and requires only the collection of information about forecast and observed weather. The potential for great economic value of weather information has been highlighted recently (Katz and Murphy 1997) The complexity of the relationship between quality and value of forecasts has been illustrated by Roebber and Bosart (1996). Two primary results can be drawn from their studies. First, high (low) quality does not necessarily correspond to high (low) value for different forecast users. Secondly, benefits are often concentrated in a small number of forecast situations. Thus, even though weather information may not be significant on a large number of days, it may be important enough on a small number to have large economic impact. Brooks et al. (1997) have evaluated the quality of forecasts presented by a variety of media sources and the National Weather Service (NWS) for Oklahoma City for a period of one year, using a distributionsoriented approach (Murphy and Winkler 1987). The forecasts include maximum and minimum temperatures out through five days lead time and precipitation probability forecasts from two of the stations out to seven days. In summary, the different forecast sources all have strengths and weaknesses and selecting an