Benzi, Parisi, Sutera and Vulpiani (Tellus 34, 10, 1982) proposed that the dominant ~100,000-year (100 kyr) peak in Quaternary paleoclimate spectra arises from a stochastic resonance: a weak periodic modulation of the solar constant by Earth-orbital (Milankovitch-band) forcing, about 0.1% in amplitude and 100 kyr in period, periodically modulates the barrier separating two stable global climates of a zero-dimensional energy balance model, while internal stochastic fluctuations — with variance in the narrow window 0.14–0.22 K² per year — trigger glacial–interglacial transitions synchronized with the forcing. The authors demonstrated the effect only in this idealized single-variable model and concluded that the mechanism 'seems to warrant further investigation.' The analytic theory of stochastic resonance in bistable systems is now mature, and the mechanism has been demonstrated in a coupled ocean–atmosphere model for millennial-scale glacial variability, but the original question is unresolved: does stochastic resonance, driven by observationally justified noise, actually generate the ~100 kyr glacial–interglacial cycles of the real climate system in realistic coupled models and in quantitative agreement with paleoclimate records, or are the cycles a deterministic nonlinear response of the ice–climate system to orbital forcing?

Background

Oxygen-isotope records (δ18O, the ratio of heavy to light oxygen isotopes in fossil plankton and in ice, which track global ice volume and temperature) from deep-sea cores show that the Quaternary period (the last ~2.6 million years) was dominated by alternating glacial and interglacial climates with global temperature swings of order 10 K. The power spectrum of these records contains a dominant narrow peak near a period of 10510^5 years, with smaller peaks near 40,000 and ~21,000 years (Hays et al., 1976). Milankovitch theory attributes these frequencies to quasi-periodic variations of the Earth's orbit — the eccentricity of the orbit (~100 kyr band), the obliquity (tilt) of the spin axis (~41 kyr), and the precession of the longitude of perihelion (~21–23 kyr) — which weakly modulate the seasonal and geographical distribution of incoming solar radiation. A long-standing difficulty, often called the 100,000-year problem, is that the orbital forcing near 100 kyr is far too weak, in linear response terms, to account for the observed ~10 K amplitude; moreover, the dominant rhythm changed from ~41 kyr to ~100 kyr about 800–1200 thousand years ago (the mid-Pleistocene transition, MPT) without a corresponding change in the orbital forcing itself.

The simplest model class addressing this question is the zero-dimensional (globally averaged) energy balance model of Budyko–Sellers type: a single equation for the globally averaged temperature T balancing incoming solar radiation against outgoing thermal radiation, Rin=μ(1−α(T))QR_{in}=\mu(1-\alpha(T))Q versus Rout=ε(T)σT4R_{out}=\varepsilon(T)\sigma T^4, where α(T) is the temperature-dependent albedo (reflectivity, dominated by ice and cloud cover), ε(T) an effective emissivity, σ the Stefan–Boltzmann constant, Q the mean solar input, and μ a dimensionless modulation of the solar constant. Because the albedo feedback (ice and cloud cover reflecting more sunlight as the planet cools) can outweigh the stabilizing Planck response (thermal radiation increasing with temperature) over a range of T, such models can possess two stable steady 'climates' separated by roughly 10 K with an unstable state between them; the stable branches are identified with glacial and interglacial states. Hasselmann (1976) showed that fast, unpredictable weather and ocean fluctuations can be modeled, on slow climate timescales, as white noise (a Wiener process: random forcing with no correlation time) driving the slow climate variables. Sutera (1981) then showed that in a bistable energy balance model such noise produces random glacial–interglacial transitions whose mean waiting time can be of order 10510^5 years — but the waiting-time distribution is exponential, so noise alone yields no spectral periodicity.

Stochastic resonance, introduced for this problem by Benzi, Parisi, Sutera and Vulpiani (1982) and independently by Nicolis (1982), adds the third ingredient: a weak periodic forcing. In a bistable system the mean escape time over the barrier depends exponentially on the ratio of barrier depth to noise variance (Kramers-type escape). A small periodic modulation μ(t)=1+0.0005cos⁡(ωt)\mu(t)=1+0.0005\cos(\omega t) with a 10510^5-year period — of the order of the actual eccentricity-driven change in annual-mean insolation — periodically raises and lowers the barrier separating the two climates by a factor of order two while moving the stable states themselves by only a few tenths of a kelvin. For an intermediate window of noise variance, the escape probability is negligibly small when the barrier is high and close to one when it is low, so the random transitions lock in phase with the forcing and the spectrum develops a sharp 10510^5-year peak of full ~10 K amplitude. In the 1982 simulations (stable climates at 280 K and 290 K) this resonance appeared for noise variance between 0.14 and 0.22 K² per year, with the displayed case at about 0.15 K² per year, described as compatible with historical estimates of global temperature fluctuations. Companion papers by the same group and by Nicolis developed the mechanism and a heuristic resonance condition, and stochastic resonance subsequently became a major topic of statistical physics, with a mature theory of escape-time statistics, synchronization, and spectral amplification (reviewed by Gammaitoni et al., 1998).

In modern terms the 1982 mechanism asks whether noise plays an essential, amplifying role in the glacial cycles, and it faces live competitors. Earth-system models of intermediate complexity (EMICs: coupled climate models with reduced spatial resolution but explicit ice-sheet and carbon-cycle components) simulate 100-kyr cycles and the MPT as deterministic, strongly nonlinear responses of the ice–climate–carbon system to orbital forcing, without any resonance (Ganopolski & Calov, 2011; Willeit et al., 2019). Noise-assisted transitions and coherence resonance (noise-induced near-periodicity without, or with only weak, periodic forcing) have been analyzed in low-order and conceptual models (Ditlevsen, 2009, 2010; Ditlevsen et al., 2020; Bosio et al., 2023). Deterministic excitation of internal relaxation oscillations (self-sustained slow–fast cycles) has been proposed to reproduce the timing of glacial terminations with no stochastic parameterization at all (Pierini, 2023). And the most recent direct continuation of the 1982 program — a stochastic-resonance model with greenhouse-gas-modulated, non-stationary bistability — reproduces both the emergence of the glacial cycles and the MPT (Xu et al., 2025). Dansgaard–Oeschger events (abrupt millennial-scale warmings recorded in Greenland ice during the last glacial period) provide the one setting where stochastic resonance has been demonstrated in a coupled ocean–atmosphere model with plausible noise (Ganopolski & Rahmstorf, 2002) and tested against data (Alley et al., 2001).

Problem Statement

The question left by the 1982 paper is whether the stochastic-resonance mechanism — cooperative amplification of weak, ~100-kyr-period orbital forcing by internal stochastic variability in a climate possessing glacial and interglacial attractors — is an operative mechanism of the real climate system's Quaternary glacial–interglacial cycles, rather than an artifact of a zero-dimensional, single-variable model with a tuned noise level. The authors concluded that the mechanism 'seems to warrant further investigation'; the open problem is whether it withstands that further investigation for the real climate system. Resolving the general question requires addressing all three of the following concrete lines of evidence, which are its joint content rather than optional examples:

(1) Realistic-model demonstration or exclusion: whether spatially resolved, coupled atmosphere–ocean–ice-sheet climate models driven by realistic orbital forcing and observationally justified stochastic forcing (in magnitude, spectral color, and state dependence) exhibit stochastic-resonant glacial–interglacial transitions of the observed ~10 K amplitude and ~100-kyr rhythm — as opposed to producing the cycles by deterministic nonlinear response or not reproducing them at all.

(2) Noise quantification: whether the internal stochastic forcing of the actual climate system, from identified physical sources (weather, ocean variability, volcanic and other short-term perturbations), has the magnitude and statistical structure that the barrier structure of the climate system requires for resonance — the 1982 zero-dimensional model needed a variance within 0.14–0.22 K² per year against its barrier modulation, and the analogous requirement must be evaluated for realistic models.

(3) Observational test: whether the resonance signature — amplified narrow-band spectral power near 10510^5 years, phase locking of glacial–interglacial transitions to the orbital forcing, and dependence of the rhythm on noise amplitude — quantitatively matches paleoclimate records.

These questions are posed against the background of the competing live mechanisms (deterministic nonlinear response of the ice–climate–carbon system, noise-assisted transitions, coherence resonance, deterministic excitation of internal relaxation oscillations).

References

  1. Benzi, R., Parisi, G., Sutera, A., & Vulpiani, A. (1982). Stochastic resonance in climatic change. Tellus 34(1), 10–16. DOI: 10.1111/j.2153-3490.1982.tb01787.x (open-access copy: DOI 10.3402/tellusa.v34i1.10782). https://doi.org/10.1111/j.2153-3490.1982.tb01787.x
  2. Sutera, A. (1981). On stochastic perturbation and long-term climate behaviour. Quarterly Journal of the Royal Meteorological Society 107(451), 137–151. DOI: 10.1002/qj.49710745109. https://doi.org/10.1002/qj.49710745109
  3. Benzi, R., Sutera, A., & Vulpiani, A. (1981). The mechanism of stochastic resonance. Journal of Physics A: Mathematical and General 14(11), L453–L457. DOI: 10.1088/0305-4470/14/11/006. https://doi.org/10.1088/0305-4470/14/11/006
  4. Nicolis, C. (1982). Stochastic aspects of climatic transitions—response to a periodic forcing. Tellus 34(1), 1–9. DOI: 10.1111/j.2153-3490.1982.tb01786.x. https://doi.org/10.1111/j.2153-3490.1982.tb01786.x
  5. Benzi, R., Parisi, G., Sutera, A., & Vulpiani, A. (1983). A theory of stochastic resonance in climatic change. SIAM Journal on Applied Mathematics 43(3), 565–578. DOI: 10.1137/0143037. https://doi.org/10.1137/0143037
  6. Hasselmann, K. (1976). Stochastic climate models Part I. Theory. Tellus 28(6), 473–485. DOI: 10.1111/j.2153-3490.1976.tb00696.x. https://doi.org/10.1111/j.2153-3490.1976.tb00696.x
  7. Hays, J. D., Imbrie, J., & Shackleton, N. J. (1976). Variations in the Earth's orbit: pacemaker of the ice ages. Science 194(4270), 1121–1132. DOI: 10.1126/science.194.4270.1121. https://doi.org/10.1126/science.194.4270.1121
  8. Gammaitoni, L., Hänggi, P., Jung, P., & Marchesoni, F. (1998). Stochastic resonance. Reviews of Modern Physics 70(1), 223–287. DOI: 10.1103/RevModPhys.70.223. https://doi.org/10.1103/RevModPhys.70.223
  9. Alley, R. B., Anandakrishnan, S., & Jung, P. (2001). Stochastic resonance in the North Atlantic. Paleoceanography 16(2), 190–198. DOI: 10.1029/2000PA000518. https://doi.org/10.1029/2000PA000518
  10. Ganopolski, A., & Rahmstorf, S. (2002). Abrupt glacial climate changes due to stochastic resonance. Physical Review Letters 88(3), 038501. DOI: 10.1103/PhysRevLett.88.038501. https://doi.org/10.1103/PhysRevLett.88.038501
  11. Ditlevsen, P. D. (2009). Bifurcation structure and noise-assisted transitions in the Pleistocene glacial cycles. Paleoceanography 24(3). DOI: 10.1029/2008PA001673. https://doi.org/10.1029/2008PA001673
  12. Ditlevsen, P. D. (2010). Extension of stochastic resonance in the dynamics of ice ages. Chemical Physics 375(2–3), 403–409. DOI: 10.1016/j.chemphys.2010.05.022. https://doi.org/10.1016/j.chemphys.2010.05.022
  13. Das, M., & Kantz, H. (2020). Stochastic resonance and hysteresis in climate with state-dependent fluctuations. Physical Review E 101(6), 062145. DOI: 10.1103/PhysRevE.101.062145. https://doi.org/10.1103/PhysRevE.101.062145
  14. Ditlevsen, P. D., Mitsui, T., & Crucifix, M. (2020). Crossover and peaks in the Pleistocene climate spectrum; understanding from simple ice age models. Climate Dynamics 54(3–4), 1801–1818. DOI: 10.1007/s00382-019-05087-3. https://doi.org/10.1007/s00382-019-05087-3
  15. Bosio, A., Salizzoni, P., & Camporeale, C. (2023). Coherence resonance in paleoclimatic modeling. Climate Dynamics 60(3–4), 995–1008. DOI: 10.1007/s00382-022-06351-9. https://doi.org/10.1007/s00382-022-06351-9
  16. Ganopolski, A., & Calov, R. (2011). The role of orbital forcing, carbon dioxide and regolith in 100 kyr glacial cycles. Climate of the Past 7(4), 1415–1425. DOI: 10.5194/cp-7-1415-2011. https://doi.org/10.5194/cp-7-1415-2011
  17. Willeit, M., Ganopolski, A., Calov, R., & Brovkin, V. (2019). Mid-Pleistocene transition in glacial cycles explained by declining CO2 and regolith removal. Science Advances 5(4), eaav7337. DOI: 10.1126/sciadv.aav7337. https://doi.org/10.1126/sciadv.aav7337
  18. Leloup, G., & Paillard, D. (2022). Influence of the choice of insolation forcing on the results of a conceptual glacial cycle model. Climate of the Past 18(3), 547–558. DOI: 10.5194/cp-18-547-2022. https://doi.org/10.5194/cp-18-547-2022
  19. Pierini, S. (2023). The deterministic excitation paradigm and the late Pleistocene glacial terminations. Chaos 33(3), 033108. DOI: 10.1063/5.0127715. https://doi.org/10.1063/5.0127715
  20. Xu, T., Katul, G. G., & Hu, S. (2025). Stochastic resonance elucidates the emergence and periodicity transition of glacial cycles. Geophysical Research Letters 53(1), e2025GL118862. DOI: 10.1029/2025GL118862. https://doi.org/10.1029/2025GL118862
  21. Benzi, R. (2007). Stochastic Resonance: from climate to biology. arXiv:nlin/0702008. https://arxiv.org/abs/nlin/0702008