To explore the possibility to set up a closed form description of
climatic variables and see how the fine scale variability of the
atmosphere enters in their description, to carry out systematic
predictability analysis of meteorological and climatic fields,
and to develop models of intermediate complexity between
low-order and GCM ones in order to study the relationship between
low-frequency atmospheric variability and the response of climate
to external forcings.
The specificity of the approach lies in the extensive use of the
tools of nonlinear dynamical systems, fractals and chaos
The main thesis of the project is that error growth and limited
predictability are intrinsic properties of the atmosphere and
climate. They arise from the instability of the underlying
dynamics and the multifractal character of geophysical fields,
and are intimately connected to the sensitivity to initial
conditions characterizing chaotic dynamical systems.
The predictability analysis will be based on the evaluation of
the Lyapunov exponents of the system, and on the direct
monitoring of the time development of small initial errors.
Special algorithms are available for accomplishing this, once the
model equations describing the system under consideration are
available. There exist also special methods allowing the study
of predictability properties directly from time series data,
independent of any modeling.
In order to identify the mechanisms behind limited predictability
and the laws governing the time evolution of small errors simple
mathematical models of chaos and low-order atmospheric models
will be used. Finally, the output of large numerical models of
the weather forecasting or GCM type will be studied in the light
of the knowledge acquired from the analysis of simple models.
This will contribute to the understanding of the relative roles
of the processes developing on various scales and to the
identification of those phenomena for which the prediction skills
can be improved.
A two-layer quasi-geostrophic model of the atmosphere, both in
the beta-plane approximation and in spherical geometry will be
developed in order to understand the feedback between low
frequency variability and the interannual variability. Special
emphasis will be placed to devise the simplest workable version,
with the minimum of free parameters and dynamical variables. The
statistical properties will be studied using cluster analysis for
the identification of the weather regimes.
Funding SchemeCSC - Cost-sharing contracts