Premium
Predictability experiments with persistence forecasts in a red‐noise atmosphere
Author(s) -
Fraedrich Klaus,
ZiehmannSchlumbohm Christine
Publication year - 1994
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.49712051608
Subject(s) - predictability , autocorrelation , ensemble forecasting , noise (video) , atmosphere (unit) , meteorology , environmental science , statistics , climatology , econometrics , mathematics , computer science , geology , geography , artificial intelligence , image (mathematics)
Individual and lagged ensemble forecasts of persistence in a red‐noise atmosphere are analysed to obtain information on predictability experiments performed in an imperfect model/ensemble environment. By examining the lead‐time‐dependent error budgets of individual and ensemble forecasts, various measures of predictability are analytically determined: the initial and saturation error, the error growth rates, the limit of predictability, error and squared error distributions depending on initial conditions, and the systematic and non‐systematic error, etc. Furthermore, weather‐regime dependent predictability can be studied directly by using different autocorrelation time‐scales of the red‐noise atmosphere. Finally, ensembles of lagged forecasts are constructed to analyse the relation between the forecast errors of the ensemble mean and the dispersion within the ensemble. Despite the simplicity of this external predictability experiment, the error budgets show features that may be qualitatively compared with those of numerical weather‐prediction and climate‐model systems.