Institute of Computer Science, Academy of Sciences of the Czech Republic
Pod vodárenskou vezí 2, 182 07 Prague 8, Czech Republic
Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic
Bocni II/1401, 141 31 Prague 4, Czech Republic
In this chapter we present a nonlinear enhancement of a linear method, the singular system analysis (SSA), which can identify potentially predictable or relatively regular processes, such as cycles and oscillations, in a background of colored noise. The first step in the distinction of a signal from noise is a linear transformation of the data provided by the SSA. In the second step, the dynamics of the SSA modes is quantified in a general, nonlinear way, so that dynamical modes are identified which are more regular, or better predictable than linearly filtered noise. A number of oscillatory modes are identified in data reflecting solar and geomagnetic activity and climate variability, some of them sharing common periods.