We estimated the Lyapunov exponents γ for the most energetic
PCs Ak with k ∈ [1, 30]. Because of the significant variability
in the short timescale fluctuations, we had previously reduced
the noise in each PC by applying a simple 13-point moving average,
thus filtering out fluctuations below 1 yr. The obtained
Lyapunov exponents γ are reported in Table 1 up to k = 18,
i.e., to the k-value where the exponential decay of the eigenvalue
λk spectrum (see Fig. 2) breaks. The first evident result of this
analysis is that we can clearly identify two main families of PCs
(and EOFs): the first, consisting of the first two PCs, characterized
by the same Lyapunov exponentγ = [0.16±0.04], and the
second consisting of all other PCs (k ∈ [3, 18]) with an average
Lyapunov exponent γ = [0.32 ± 0.07].