Ensemble forecasts allow statements about the reliability of operational prognoses. The accuracy of the forecasts depends not only on the forecast period, but on the stability of the weather situation as well. As the atmosphere is a chaotic system, small disturbances can cause huge reactions. In an instable weather situation for example a few tenths of a degree difference in the starting temperatures over the Atlantic can produce a completely different weather situation over Central Europe on the third prognostic day.

For the calculation of ensemble forecasts the operational model is simplified slightly (rougher temporal and spatial resolution) and run 50 times with slightly changed starting conditions. This results in 50 additional curves for every prognosed parameter. A high correlation of the ensemble forecasts implies stable atmosphic conditions and very reliable forecasts. If the results have a high spread, an instable weather situation and uncertain forecasts should be expected. For the 80%-band (confidence intervall) the 5 most extreme results on the upper and lower border of the span have been sorted out. The span-width of the left 40 runs informs about the interval of the possible weather development.

The graphics show the operational run (red), the confidence intervall (80%-band) (gray) and the average of 50 runs (white). Please notice that these data are based on the operational run of the ECMWF. They refer to a grid point of the models computing grid. In the categories weather forecasts Germany and weather forecasts Europe for the city forecasts, these pure model results have been processed with a statistic treatment. So they really represent the meteorological conditions in the respective cities. The ensemble forecasts should be used only for an estimation of the prognostic reliability. For actual forecasts please use the values from the category weather forecast - time series.