DEFINITIONS AND TERMS. 

        The ensemble mean provides the first order advantage of ensemble forecasting, i.e., a simple consensus of a large number of forecasts. Even greater benefits accrue through the use of the ensembles to assess confidence, or uncertainty, in the forecast. The spread provides the most direct measure of the likely skill, and therefore the confidence, of the forecast.  

        The spread is the standard deviation among the ensemble members, in units of meters. It varies both regionally and seasonally, much like the actual standard deviation of the height field. So, we expect high values in high latitudes and in the 'centers of action', low values in low latitudes, and moderate values in other locations. A standardized version, i.e. the spread as described above, divided by the climatological standard deviation of the height field appropriate to the date, is under development for the 6-10 day and week 2 average forecast periods. According to Whitaker and Loughe, Mon. Wea. Rev. Dec 1998, pp. 3292-3302, spread is useful as a predictor of skill when the spread is either very low (skill is likely to be high) or very high (skill is likely to be low).  

        Note that the spaghetti diagrams are daily out to day 7. Then spaghetti maps for the five day mean covering GFS forecast days 6-10 and the seven day mean over GFS forecast days 8-14 are provided.