The predictability of weather and climate forecasts is determined by the pr
ojection of uncertainties in both initial conditions and model formulation
onto flow-dependent instabilities of the chaotic climate attractor. Since i
t is essential to be able to estimate the impact of such uncertainties on f
orecast accuracy, no weather or climate prediction can be considered comple
te without a forecast of the associated flow-dependent predictability. The
problem of predicting uncertainty can be posed in terms of the Liouville eq
uation for the growth of initial uncertainty, or a farm of Fokker-Planck eq
uation if model uncertainties are also taken into account. However, in prac
tice, the problem is approached using ensembles of integrations of comprehe
nsive weather and climate prediction models, with explicit perturbations to
both initial conditions and model formulation; the resulting ensemble of f
orecasts can be interpreted as a probabilistic prediction.
Many of the difficulties in forecasting predictability arise from the large
dimensionality of the climate system, and special techniques to generate e
nsemble perturbations have been developed. Special emphasis is placed on th
e use of singular-vector methods to determine the linearly-unstable compone
nt of the initial probability density function. Methods to sample uncertain
ties in model formulation are also described. Practical ensemble prediction
systems for prediction on timescales of days (weather forecasts), seasons
(including predictions of El Nino) and decades (including climate change pr
ojections) are described, and examples of resulting,probabilistic forecast
products shown. Methods to evaluate the skill of these probabilistic foreca
sts are outlined. By using ensemble forecasts as input to a simple decision
-model analysis, it is shown that probability forecasts of weather and clim
ate have greater potential economic value than corresponding single determi
nistic forecasts with uncertain accuracy.