Stochastic models have contributed significantly to our understanding of
variability and predictability on seasonal-to-centennial timescales. We begin
with a brief discussion of stochastic models of atmospheric variability, and
then consider in some detail how coupling with the ocean introduces new kinds
of behaviour, such as the 'reduced thermal damping effect' and 'spatial
resonance'. The large difference in the damping rates for atmospheric and
oceanic anomalies also has significant implications for predictability. We
explore this through an analysis of non-modal growth in stochastic coupled
systems, using two simple examples: a damped inertial oscillator and a
mechanistic model of tropical Atlantic variability. We show that non-modal
growth can at times give rise to increased predictability, as compared to
modal growth. We conclude with a discussion of some of the outstanding issues
in stochastic modelling of climate variability.
Copyright. All Rights Reserved.