
Formulation of gaussian probability forecasts based on model extended-range integrations
Abstract
A sample of 40, 44-day winter forecasts is used to investigate the predictability of 850 hPa temperature over Europe. These forecasts exhibit a significant skill when averages of day 5 to day 14 and day 15 to day 44 are considered. This skill is, however, very close to that of the trivial climatology forecast. A probability forecast is performed, using a gaussian density with the deterministic forecast for the mean, and the climatological standard deviation (SD). The rank probability score (RPS) of such a forecast is better than, but again very close to, that of the probabilistic climatology forecast. The categorical forecast is also studied as a limit case when the SD is zero. The RPS is minimal when using the conditional probabilities of the verification analyses, but the results are not widely improved when robust estimates are used. The results could be widely improved if we used a suitable SD in our forecasts. However, the attempts to predict a priori the optimal SD lead to non-significant results. The best available probability forecast, in our local gaussian approach, uses, for the mean, the regression of the verification analyses by the model forecasts, and, for the SD, a scaled climatological SD.
© 1994 M. Déqué, J. F. Royer, R. Stroe, published by Stockholm University Press
This work is licensed under the Creative Commons Attribution 4.0 License.