Resonant phenomena in extended chaotic systems subject to external noise: the Lorenz96 model case

We investigate the effects of a time-correlated noise on an extended chaotic system. The chosen model is the Lorenz'96, a kind of “toy” model used for climate studies. Through the analysis of the system’s time evolution and its time and space correlations, we have obtained numerical evidence for two stochastic resonance-like behavior. Such behavior is seen when both, the usual and a generalized signal-to-noise ratio function are depicted as a function of the external noise intensity or the system size. The underlying mechanism seems to be associated to a “noise-induced chaos reduction”. The possible relevance of these and other findings for an “optimal” climate prediction are discussed.

💡 Research Summary

The paper investigates how temporally correlated external noise influences the dynamics of an extended chaotic system, using the Lorenz‑96 model as a prototypical “toy” for atmospheric and climate studies. The authors augment the standard Lorenz‑96 equations with an additive Gaussian noise term that possesses an exponential autocorrelation function characterized by a correlation time τc and an intensity D. Numerical simulations are performed with a fourth‑order Runge‑Kutta scheme, employing sufficiently long integration times to ensure statistical convergence.

Key observables include the temporal autocorrelation function, spatial correlation function, and the power spectral density (PSD) obtained via Fast Fourier Transform. Two signal‑to‑noise ratio (SNR) measures are defined: a conventional SNR that compares the PSD peak at a dominant frequency ω0 to the surrounding background, and a generalized SNRG that evaluates the ratio of the integrated peak area to the integrated background over the entire spectrum.

The study reveals two distinct stochastic‑resonance‑like phenomena. First, when the noise intensity D is varied while keeping the system size N and correlation time τc fixed, both SNR and SNRG exhibit a pronounced maximum at an intermediate D (≈0.15 for the parameters used). This peak coincides with a reduction in the largest Lyapunov exponent, indicating that the noise partially suppresses the intrinsic chaotic dynamics, allowing the underlying quasi‑periodic mode to become more coherent. Second, when the system size N (i.e., the number of lattice sites) is varied at a constant optimal noise level, the SNR again shows a maximum around N≈30. Small systems are overly dominated by noise, whereas large systems retain strong internal chaos that masks the resonant response; an intermediate size balances these effects, producing a “system‑size resonance.”

The influence of the noise correlation time is also examined. Short correlation times (approaching white noise) enhance the resonance peaks, while longer τc values smooth the response, suggesting that rapid, uncorrelated fluctuations are more effective at inducing the chaos‑reduction mechanism.

The authors interpret these findings through the concept of “noise‑induced chaos reduction.” By adding moderate, temporally correlated perturbations, the phase‑space trajectory is nudged toward lower‑dimensional attractors, diminishing high‑frequency chaotic components and amplifying low‑frequency, physically relevant structures. This mechanism explains why both the conventional and generalized SNRs improve at optimal noise levels and system sizes.

From a climate‑prediction perspective, the results imply that the presence of external variability (e.g., solar fluctuations, volcanic forcing) need not be purely detrimental. Instead, an appropriately calibrated stochastic component could mitigate model chaos, improve the coherence of large‑scale modes, and thereby enhance forecast skill. The paper suggests that optimal tuning of noise amplitude and model resolution may be a viable strategy for “noise‑assisted” prediction.

Limitations are acknowledged: Lorenz‑96 is a highly simplified one‑dimensional system with periodic boundaries and a single constant forcing term, which does not capture the full multiscale physics of the atmosphere or ocean. Moreover, the noise is assumed Gaussian and spatially uncorrelated, whereas real geophysical noise often exhibits non‑Gaussian statistics and spatial correlations. Future work is proposed to extend the analysis to multi‑scale models, incorporate non‑Gaussian and spatially correlated stochastic processes, and validate the findings against observational data.

In summary, the paper provides robust numerical evidence that temporally correlated external noise can induce two forms of stochastic resonance—one with respect to noise intensity and another with respect to system size—in the Lorenz‑96 chaotic system. The underlying mechanism is a noise‑driven reduction of chaotic degrees of freedom, leading to enhanced signal coherence. These insights open a pathway toward exploiting stochastic forcing as a tool for improving the reliability of climate and atmospheric forecasts.