|
|
This page is reserved for the verification of the here published longrange temperature forecasts. The basis may be the standard
deviation of every
month or those of the months with comparable dispersion. January and February have nearly the same values,
formerly (1961-90) at Berlin approximately 3 K,
nowadays 2.5 K. The forecast errors will be standardised with these values,
squared and subtracted from "1 ". So everybody will see how successful the different
progs had been. The range of these values
may reach from -35 = totally wrong forecast (6 times standard deviation) up to +1 = correct forecast, 0=average or no
improvement.
This indicates how good the RV = "reduction of variance" is realized by the forecast.
|