Non-uniform decay of predictability and return of skill in stochastic oscillatory models (Q2381549)

The authors examine the dynamical mechanisms that lead to the loss of predictability in low-dimensional stochastic models that exhibit three main types of oscillatory behaviour: damped, self-sustained, and heteroclinic. They show that the information expressed as a relative entropy (Kullback-Leibler divergence) between the invariant (climatological) distribution of the system and the probability density corresponding to the ensemble of realisations at finite time \(t\) decays non-uniformly as \(t\to\infty\). Long intervals during which the forecast provided by the ensemble does not loose any of its power are typical in all the three cases. Moreover, the information that the forecast provides about the individual variables in the model may increase, despite the fact that information about the entire system always decreases. The authors analyse the fully solvable case of the stochastic linear oscillator \(\dot x_1=\alpha x_1+\beta x_2\), \(\dot x_2=\gamma x_1+\delta x_2+\varepsilon \dot W\), \(\dot W\), being the Gaussian white noise. Further, the more general cases of stochastically perturbed non-linear oscillator and Duffing equations are studied numerically.