Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Climate change

Shifts in season

It's cold in winter and hot in summer. But the latest analysis illustrates the need to put observational data at the forefront of attempts to achieve a more detailed understanding of the annual temperature cycle.

It has been known for more than a century1 that increasing the concentration of carbon dioxide in the atmosphere results in an increase in Earth's surface temperature. By contrast, it is only just over a decade since the discovery that CO2 levels also affect the timing of the annual temperature cycle2,3, although the details remain enigmatic.

On page 435 of this issue, Stine and colleagues4 describe how they have updated and extended earlier studies of the annual cycle5,6, using better spatial coverage and more recent data. They have concentrated on the temperate zones because of the dominant annual temperature cycle and, at least in the Northern Hemisphere, the reasonable spatial data coverage in those zones. As well as incorporating some technical improvements, the authors analysed the annual temperature cycle over the oceans.

The annual cycle has two distinct components, amplitude and phase. Stine and colleagues conclude that the amplitude — loosely, half the difference between summer and winter temperatures — has been decreasing over most continental areas and increasing over the oceans. The phase describes the relative timing of the periodic (seasonal) component of temperature. For the most part, the seasons occur earlier over land and later over the oceans, and Stine et al. estimate the terrestrial phase shift to have been 1.7 days between 1954 and 2007. This shift, and the changes in amplitude, are highly anomalous when compared with the data from between 1900 and 1953, implicating human agency as the cause.

The common perception of the timing of the seasons is more complicated because it involves both changes in the annual cycle, discussed here, and the increase in average temperature. (See Figure 1 of the Supplementary Information4 for a graphic description of the different effects.) For example, taking the date from which the temperature usually stays above freezing as marking the start of spring, the increase in average temperature, the smaller seasonal amplitude (which implies warmer winters) and the change in phase all work in the same direction, so the observed effect is large. This is well documented in studies of bird migrations and similar phenomena7,8, where one finds many examples of seasonal patterns shifting to an earlier date by more than a month (Fig. 1). Phase changes have also been invoked to help explain problems ranging from the theory of palaeoclimates9 to changes in sea level10 and even in human mortality11.

Figure 1: Migrating greylag geese.


Spring, and associated phenomena such as bird migrations, now occur earlier than at the start of the twentieth century. Stine and colleagues' data4 quantify the seasonal shift.

Stine et al.4 also compare their observations with the results of a suite of two dozen climate models used by the Intergovernmental Panel on Climate Change (IPCC), and the results are dismaying. Some of these models reproduce the decrease in amplitude, first shown in 1980 (ref. 12), but none predicts, or even reproduces, the change in phase. I have no personal experience with these models, so beyond a general scepticism about complicated models (perhaps best expressed by George Box's dictum, “All models are wrong but some are useful”), I cannot say why they fail. We must remember, however, that although climate models incorporate an amazing variety of effects and get many things right, they are almost certainly missing many more.

As an example, in the mid-1990s I was discussing the phase problem with members of a modelling group and learned that their model had Earth in a circular orbit with no precession. This was astonishing. First, we are trying to measure the effects of CO2 to high accuracy — say 0.01 °C, in a system in which annual temperature extremes routinely exceed ± 50 °C. Second, on an ice-age timescale, the effects of precession are immense, strong enough to be used as a clock. Third, we have known that the orbit is elliptical since Johannes Kepler in the seventeenth century, and about precession since Hipparchus (around 150 bc). The duration of the instrumental temperature record is now 1% or 2% of the 26,000-year precession cycle: when trying to measure small effects it is unwise to ignore large ones.

One should also note the contrast between the enormous computational resources used by the models and the relatively meagre effort required to analyse real data. Thus, work of the type done by Stine et al. is to be applauded. Ignoring the time required to assemble the data and write the programs, it probably took no more than a few seconds of computer time to show effects that were not predicted by any of the models. As Richard Feynman commented, “It doesn't matter how beautiful your theory is, it doesn't matter how smart you are. If it doesn't agree with experiment, it's wrong.”

Where do we go from here? One of many perplexing problems is the year-to-year variations in phase and amplitude in temperature data. These variations are obvious in all of the long-term temperature records and are reasonably consistent with variations in the Sun's magnetic field. We do not understand the subtle influences on climate exercised by solar effects such as the solar wind, the charged particles that flow out from the Sun. Observational evidence for such a coupling has been accumulating for decades, through both palaeoclimate data13 and studies of the upper atmosphere14. However, when one has observed the Sun's acoustic oscillations in barometric pressure15, it is possibly time to pay attention to solar observations.

The solar wind carries much more energy than is available from Edward Lorenz's butterflies, often used to 'explain' purported chaotic behaviour in climate. This raises a philosophical question, as to whether the fascination with 'chaoplexology' in climate research has resulted in a failure to take observations and statistics seriously enough. Climate may be formally chaotic, but so is Earth's orbit16 and this has not prevented people from analysing it in exquisite detail. In my opinion, chaos, fractals, long-memory processes and their ilk should be invoked only when all of the various climate forcings have been carefully studied and all simpler explanations eliminated. We are not even close to meeting that goal.

Finally, independent of any shortcomings in the models, we must remember that the observational evidence for human influence on the climate system is overwhelming. Stine and colleagues' paper4 adds to that evidence. If we do not stop polluting Earth's atmosphere, we may not have enough time left to develop models sophisticated enough to show what is obvious in the data now.


  1. 1

    Arrhenius, S. Phil. Mag. J. Sci. 41, 237–275 (1896).

    CAS  Article  Google Scholar 

  2. 2

    Thomson, D. J. Science 268, 59–68 (1995).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Thompson, R. Int. J. Climatol. 15, 175–185 (1995).

    Article  Google Scholar 

  4. 4

    Stine, A. R., Huybers, P. & Fung, I. Y. Nature 457, 435–440 (2009).| Article |

    ADS  CAS  Article  Google Scholar 

  5. 5

    Mann, M. E. & Park, J. Geophys. Res. Lett. 23, 1111–1114 (1996).

    ADS  CAS  Article  Google Scholar 

  6. 6

    Wallace, C. J. & Osborn, T. J. Climate Res. 22, 1–11 (2002).

    ADS  Article  Google Scholar 

  7. 7

    Schwartz, M. D., Ahas, R. & Aasa, A. Global Change Biol. 12, 343–351 (2006).

    ADS  Article  Google Scholar 

  8. 8

    Parmesan, C. Annu. Rev. Ecol. Evol. Syst. 37, 637–669 (2006).

    Article  Google Scholar 

  9. 9

    Jones, P. D., Briffa, K. R. & Osborn, T. J. J. Geophys. Res. doi:10.1029/2003JD003695 (2003).

  10. 10

    Barbosa, S. M., Silva, M. E. & Fernandes, M. J. Tellus A 60, 165–177 (2008).

    ADS  Article  Google Scholar 

  11. 11

    McGregor, G. R., Watkin, H. A. & Cox, M. Climate Res. 25, 253–263 (2004).

    ADS  Article  Google Scholar 

  12. 12

    Manabe, S. & Stouffer, R. J. J. Geophys. Res. 85, 5529–5554 (1980).

    ADS  Article  Google Scholar 

  13. 13

    Wigley, T. M. L. Solar Phys. 74, 435–471 (1981).

    ADS  Article  Google Scholar 

  14. 14

    Arnold, N. Phil. Trans. R. Soc. Lond. A 360, 2787–2804 (2002).

    ADS  Article  Google Scholar 

  15. 15

    Thomson, D. J., Lanzerotti, L. J., Vernon, F. L., Lessard, M. R. & Smith, L. T. P. Proc. IEEE 95, 1085–1132 (2007).

    Article  Google Scholar 

  16. 16

    Laskar, J., Joutel, F. & Boudin, F. Astron. Astrophys. 270, 522–533 (1993).

    ADS  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Thomson, D. Shifts in season. Nature 457, 391–392 (2009).

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.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing