A novel explanation for the long-term temperature record in Antarctic ice cores invokes local solar radiation as the driving agent. This proposal will prompt palaeoclimate scientists to pause and to go back to basics. See Letter p.91
Analysis of ice cores retrieved from the Antarctic and Greenland ice sheets is one of the main sources of our understanding of past climate. A component of that understanding is that, on timescales of 20,000 years and more, climate change in Antarctica is determined by the amount of solar radiation (insolation) reaching high northern latitudes in summer. On page 91 of this issue, Laepple et al.1 call into question some of the evidence for that view.
Precisely dated polar ice cores have allowed examination of the 'bipolar see-saw' relationship of air temperatures between the hemispheres on millennial timescales2, as well as of longer-term, glacial–interglacial climate change paced by variations in Earth's orbit — the Milankovitch forcing of ice ages3. In these studies, the use of isotopes that are stable in water, in the form of the ratios of oxygen and deuterium isotopes, is well established. These ratios constitute the fundamental proxy measurements for estimating past temperatures from ice cores at both poles2,3,4.
Because ice cores consist of ice, the stable-isotope ratios in the ice stem from those contained in precipitation (snow, which becomes compacted to ice). In other words, if there is no precipitation, no isotopic signal remains in the ice core. This simple principle has been acknowledged in interpreting the Greenland ice-core record5. Subsequent studies6,7 have described how changes in the seasonal pattern of precipitation during glacial–interglacial cycles have significantly biased the isotopic temperature record in Greenland. But it was thought that the effect in Antarctica was probably minor because of its comparatively stable precipitation seasonality.
Laepple and co-authors1 apply this idea of precipitation seasonality to the Antarctic ice-core record. However, they do not deal with changes in seasonal patterns, as the previous studies did, but instead consider the situation in which seasonality is itself unchanging and in which snow accumulation over inland Antarctica is maximal in winter and minimal in summer. This seasonality in snowfall has various causes, such as the strong radiative cooling that induces clear-sky precipitation and increased moisture transport in winter, and sublimation of ice into water vapour in summer.
By assuming that this seasonal pattern of snow accumulation has persisted throughout glacial–interglacial cycles, and that the local air temperature has fluctuated according to the present-day relationship between temperature and insolation, the authors1 produce an accumulation-weighted insolation signal as a record of temperatures in Antarctic ice cores. They find that it has the opposite phase to the orbital-precession component (determined by long-term changes in the orientation of Earth's rotational axis) of the local summer-insolation signal — and so, surprisingly, that it is in phase with summer-insolation intensity in the Northern Hemisphere.
If the Antarctic local temperature is determined by local insolation, the precession component in the ice-core temperature signal should be out of phase with Northern Hemisphere insolation, because the precession component is out of phase between the two hemispheres. However, the precession component filtered from the isotopic temperature record in the Antarctic ice cores is coherent and in phase with the Northern Hemisphere insolation intensity3 — seemingly supporting the Milankovitch theory, according to which southern climate is driven by insolation changes at high northern latitudes.
But does the close phasing necessarily support a causal relationship? Perhaps not. Laepple and co-authors1 have rethought how the signals of temperature change are produced. Their accumulation-weighted insolation record suggests that a precession rhythm synchronized with — but not caused by — the Northern Hemisphere could be generated if the local temperature fluctuated in line with local insolation conditions in the Southern Hemisphere. The unveiling of this 'pseudo-rhythm' strikes at the foundation of temperature estimates gleaned by analysing isotope ratios in ice cores. Does it mean, as Laepple et al. suggest, that the evidence from Antarctic ice cores cannot be used to support or refute the Milankovitch theory?
This theory is supported not just by temperatures inferred from Antarctic ice cores, but also by sea surface temperatures recorded in sediment cores from the Southern Ocean. In these cores, the orbital-precession rhythm is often found to be in phase with summer insolation in the Northern Hemisphere and therefore opposing the local summer insolation8. The seasonality of snow accumulation does not affect sediment processes in the ocean. Furthermore, the existence of shorter (millennial timescale) but strong bipolar see-saw connections between the two hemispheres implies that there are indeed mechanisms for the interhemispheric propagation of climate signals through the ocean and/or atmosphere2. There is no reason to believe that such mechanisms have not operated over longer timescales.
A caveat regarding the results themselves is that Laepple and colleagues' insolation-based air-temperature estimate shows a rather small amplitude (around 0.7 °C peak to peak) compared with that derived from ice cores (3 °C peak to peak). This is probably because the authors' use of local insolation as the temperature proxy means that they assume zero insolation during winter (polar night) throughout glacial–interglacial cycles. They themselves acknowledge this point, and suggest that other factors not accounted for in their approach may explain the discrepancy.
Nevertheless, we must now consider that the orbital-precession rhythm in Antarctic ice cores can partly be attributed to local conditions. In the same way that an ill-fitting piece of a jigsaw puzzle can be disconcerting, this pseudo-rhythm will be discomfiting to those who study palaeoclimate and climate dynamics. 'Is the signal I see really created by climate change?', is a question they will have to ask themselves. And they will need to take a hard look at the principles on which their data are founded. The relationship between the isotopes in water and air temperature, for instance, is based on geographical (spatial) observations only. But its temporal variability has not been confirmed at any ice-core drilling sites in inland Antarctica, even by observations on an annual timescale. Sometimes, in science as in life, it is necessary to pause in order to make progress.
About this article
Journal of the Air & Waste Management Association (2013)