Global change

Climate's astronomical sensors

Article metrics

A re-evaluation of the relationship between Earth's orbital parameters, ice-sheet extent and ocean circulation sets further puzzles for those trying to disentangle cause from effect in long-term climatic changes.

Earth's climate 'feels' the slow changes in the parameters of our orbit around the Sun. The great ice sheets of the Northern Hemisphere are one sensor, in that they are sensitive to the amount of solar energy they receive in summer. Lisiecki et al.1 (page 85 of this issue) provide evidence that ocean dynamics also responds to orbital changes, and not just in the north.

Much of our life is controlled by the rhythms of days and seasons — not surprisingly, given that the Sun is our ultimate source of energy. Earth's atmosphere senses the rhythms of days and seasons, too, but both atmosphere and oceans may respond to the much longer astronomical cycles that affect incoming solar radiation. In 1976, Hays et al.2 described how they tackled this problem. They collected deep-sea sediments in the Southern Ocean, dated them according to depth, and analysed the oxygen-isotope composition of the calcium carbonate remains of foraminifera preserved in the sediments. This quantity is a proxy for ice-age conditions: isotopic composition indicates whether climate at any time was glacial — with large ice sheets in the Northern Hemisphere and low temperatures in the deep oceans — or interglacial, as today. Hays et al. then plotted this measure against time to estimate the frequency spectrum. Several of the dominant glacial oscillation periods they found corresponded perfectly to the astronomical periods calculated analytically by Berger3: 19,000 and 23,000 years for climatic precession; 41,000 years for changes in obliquity.

So, what are precession and obliquity? Earth revolves around the Sun following an elliptic figure. The climatic-precession parameter tells us what time in the year we reach perihelion — that is, the closest point to the Sun, when Earth is globally exposed to the maximum amount of incoming solar radiation. Perihelion is presently reached on 3 January; it will be reached in July in 11,000 years and again in January in 22,000 years. Obliquity is the angle between the Equator and Earth's orbital plane. Changes in this angle are responsible for the seasons: the larger it is, the more energy the polar areas receive in summer. Neither precession nor obliquity modifies the total amount of energy reaching Earth in one full year. Only eccentricity — the orbital deviation from the circular — does that, but the effect is so small that it is neglected in most theories. Eccentricity does, however, modulate the amplitude of the effect of precession with periods of 100,000 and 400,000 years3.

Given that astronomical cycles hardly modify the global amount of incoming solar energy, the climate's astronomical sensors must be sensitive to the seasonal and spatial distribution of this energy. In that respect, the response of ice sheets immediately comes to mind. The amount of ice melting every year depends on the amount of solar energy absorbed during the warm season; the total ice mass is therefore expected to decrease when obliquity is high and perihelion is reached around summer. As early as 1876, John Murphy suggested that summer insolation could control glacial cycles4. History records the name of Milutin Milankovitch5 as the father of this theory, however, because of the firm mathematical foundation he provided for it (although he missed some crucial aspects of the ice sheets' response6).

But candidate astronomical sensors other than ice sheets have been proposed, most notably in two papers7,8published as part of the SPECMAP project, which aims to rationalize the chronology represented by different palaeoclimate records. These papers downgraded the Milankovitch mechanism to a second-order effect, and attributed the prime cause of glacial–interglacial cycles to the response of Arctic sea ice to northern summer insolation. This Arctic response would have led to the development of northern ice sheets through a somewhat convoluted causal pathway involving circulation in the North Atlantic Ocean and changes in the concentration of atmospheric carbon dioxide.

Several drawbacks have been identified in the SPECMAP model9, but Lisiecki et al.1 have delivered the coup de grâce. They began by noting that SPECMAP was not supported by good palaeoenvironmental records of the deep-ocean circulation. They instead used 30 archives of a naturally occurring isotopic indicator (the isotopic ratio of carbon in foraminifera shells) known to be sensitive to the distribution of water masses in the ocean. The archives are sufficiently broadly distributed geographically to provide a good idea of the global ocean circulation dynamics over the past 250,000 years.

Lisiecki et al. then essentially replicated the SPECMAP analysis procedure: band-pass filtering of time-series data to isolate the fraction of the signal thought to respond to precession and obliquity, and then assessing how this signal lagged the orbital elements (Fig. 1). The surprising result is that, when obliquity is high, the Atlantic Ocean tends to be dominated by deep water of Nordic origin — the opposite of the SPECMAP prediction. Moreover, when Earth is near its perihelion at the time of summer in the Northern Hemisphere, the Atlantic seems to be dominated by water of southern rather than northern origin.

Figure 1: Illustration of the linear signal analysis used by Lisiecki et al.1.

This example considers the output of a simple dynamical system (green, simulated sea-level11) forced by a known input (blue, summer insolation in the Northern Hemisphere12). Both signals are filtered to extract their variance in a given frequency band (here, around 21,000 years, which corresponds to Earth's climatic precession). It is then verified that their phases, estimated by means of a Hilbert transform, are coherently related to each other. Such is the case here, with output lagging input by 1,500 to 6,500 years (90% confidence). This procedure confirms that the input effectively controls the system. But it does not guarantee that it is the cause of the large cycles in the output signal. In the artificial case tested here, these cycles are known to be autonomous: they would occur even without external forcing. The forcing simply acts as a clock, which has the effect of improving output predictability. Likewise, Lisiecki et al.1 show that obliquity and precession control ocean circulation, but not the extent to which glacial cycles depend on this external forcing.

With this, the view that the Arctic is the main 'front-end' orbital sensor of the climate system becomes hard to defend: the two orbital configurations (high obliquity and Northern Hemisphere perihelion) have the similar effect of increasing summer insolation in the Arctic. How could they lead to opposite ocean responses? Lisiecki et al.1 remark that things would be easier to explain if the ocean responded to summer insolation of the Southern Ocean, but we are left with conjectures to explain the mechanism. This is a challenge for those running general circulation models of the ocean and atmosphere.

Finally, another challenge merits mention. Lisiecki et al. used linear time-series analysis techniques to decipher the influence of orbital elements on climate. This is perhaps good enough to point out first-order effects, but climate is a nonlinear system. For example, it took about 100,000 years to build the big ice sheets that existed on the Earth of our mammoth-chasing ancestors, but those ice sheets largely disappeared within 10,000 years. That's not typical of a linear system. In fact, we are still unsure that orbital variations are necessary to explain glacial–interglacial cycles10,11.

We need a more systematic way of developing and applying nonlinear statistical models to test our understanding of the slow dynamics of climate. The task is not straightforward — how, for instance, do we rigorously account for dating uncertainties in sediments without becoming trapped in circular arguments (palaeoclimate scientists know this as the 'orbital tuning' problem)? Yet it is the only way to answer the crucial question of how far ahead glacial–interglacial cycles can be predicted.


  1. 1

    Lisiecki, L. E., Raymo, M. E. & Curry, W. B. Nature 456, 85–88 (2008).

  2. 2

    Hays, J. D., Imbrie, J. & Shackleton, N. J. Science 194, 1121–1132 (1976).

  3. 3

    Berger, A. Nature 268, 44–45 (1977).

  4. 4

    Murphy, J. J. Q. J. Geol. Soc. Lond. 32, 400–406 (1876).

  5. 5

    Milankovitch, M. Canon of Insolation and the Ice-age Problem (Engl. transl. of original 1941 publ., Narodna biblioteka Srbije, Belgrade, 1998).

  6. 6

    Weertman, J. Nature 261, 17–20 (1976).

  7. 7

    Imbrie, J. et al. Paleoceanography 7, 701–738 (1992).

  8. 8

    Imbrie, J. et al. Paleoceanography 8, 699–735 (1993).

  9. 9

    Ruddiman, W. F. Quat. Sci. Rev. 25, 3092–3112 (2006).

  10. 10

    Tziperman, E., Raymo, M. E., Huybers, P. & Wunsch, C. Paleoceanography 21, doi:10.1029/2005PA001241 (2006).

  11. 11

    Saltzman, B. & Maasch, K. A. Trans. R. Soc. Edinb. 81, 315–325 (1990).

  12. 12

    Berger, A. J. Atmos. Sci. 35, 2362–2367 (1978).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Crucifix, M. Climate's astronomical sensors. Nature 456, 47–48 (2008) doi:10.1038/456047a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.