The last glaciation (110,000 to 12,000 years ago) was characterized by rapid climate fluctuations that occurred 25 times. The best evidence for these warming events is found in the Greenland ice cores, which go back to the end of the last interglacial period about 110,000 years ago.

In the Northern Hemisphere, these events are rapid warming episodes, typically over decades, followed by gradual cooling over a longer period. For example, about 11,500 years ago, average annual temperatures on the Greenland ice sheet warmed by around 8 °C over 40 years. The cold period following the warming event lasted several hundred years and was reflected in an expansion of the polar front, with ice moving further south across the North Atlantic ocean.

Comparison of evidence from Greenland and Antarctic ice cores suggests that the Northern Hemisphere warming events are related via a see-saw to cool periods in Antarctica and vice versa. The warm periods in the south are called Antarctic Isotope Maxima or AIM.

Comparison of Temperature Variability in Greenland and Antarctica
Comparison of Temperature Variability in Greenland and Antarctica

Comparison of Greenland and Antarctica temperature variability.
''Antarctica temperature proxy (blue, lowercurve); Greenland temperature proxy (red, upper curve).
Antarctic Isotope Maxima (AIM) 8, 12 and 17 are marked. Green horizontal bars denote periods with minimal Antarctic Isotopic Maximum or Arctic warming events activity. Source E. W. Wolff, H. Fischer and R. Röthlisberger, Nature Geoscience Letters, 22 Feb 2009 DOI:10.1038/NGEO442

Previous studies have reported that each northern warming event was accompanied by an increase in atmospheric methane concentrations. The relative timing of warming over Greenland and higher emissions of methane have been difficult to quantify. (It is presumed that the enhanced methane emissions were either from boreal or tropical wetlands.)

There is evidence from several sources that during global warming events there is a close correlation between temperature and greenhouse gas concentration. An open question is whether during these events temperature precedes increasing greenhouse gas concentrations or vice versa. In this article the authors report on a high resolution analysis of ice core data from Greenland to determine and compare temperature and methane concentration during a warming event near the end of the last glacial period and the beginning of the current warm period.

Around 14,500 years ago, a warming period known as the Bølling–Allerød interval led to sudden warming. This period has been seen in proxy records including Greenland ice cores. It was followed by a 1,300 year period of cold climatic conditions and drought which occurred between approximately 12,800 and 11,500 years before the present (BP). This cooling period, referred to as the Younger Dryas, was followed by the warming which led to the present interglacial. Despite the evidence for the Bølling–Allerød warming in many palaeoclimate records, the relative timing of changes in temperature and atmospheric greenhouse gases in different regions of the world remains an open question.

Methane (CH4) is an important greenhouse gas. Methane sources are presumed to be boreal or tropical. The largest methane source is the tropics - the authors estimate that during the Bølling–Allerød warming boreal sources could contribute about 22% of the total methane increase. Because methane is quickly mixed through the atmosphere, methane concentrations in air bubbles can be used to link Greenland and the tropics.


The phasing of global climate change at the onset of the Bølling–Allerød event were studied using air preserved in bubbles in the North Greenland Eemian (NEEM) ice core. Specifically, methane concentrations, which act as a proxy for low-latitude climate, and the nitrogen-15/nitrogen-14 ratio of atmospheric nitrogen, which reflects Greenland surface temperature, were measured over the same interval of time.

Gases become trapped in bubbles at the bottom of the layer of increasingly dense snow known as the firn which ultimately becomes ice. Air bubbles are captured in the firn at a depth of 60–100 meters.

There are several factors that complicate the relative chronology of temperature and greenhouse gas concentrations in ice cores. First of all, air can diffuse up to 60-100 meters in the firn (the layer where snow is gradually compacted into ice) so that the air bubbles captured at any given level in the ice do not correspond to the age of the surrounding ice. Another problem is that methane diffuses faster in the firn than nitrogen. In addition a high resolution analysis needs to account for methane mixing in the atmosphere over a large geographic area relative to its atmospheric lifetime of about 10 years.

Bubbles in Greenland ice core
Bubbles in Greenland ice core

In this study the authors partly avoid the first problem by using nitrogen isotope ratio (ratio of nitrogen-15 to nitrogen-14) in the air bubbles themselves as a temperature proxy. The nitrogen-15/nitrogen-14 ratio of atmospheric nitrogen provides a thermometer for abrupt changes in surface temperature at an ice core site.


The initial increase in both gases starts at 14,698 years before the present (years BP) with a relative uncertainty of 5 years due to the spacing of sample measurements. The increases are easy to detect because the first data point of each increase is much larger than the uncertainty in the previous data points. (The magnitude of the increase exceeds the uncertainty of the measurement by a factor of 5 for methane concentration and 8 for nitrogen-15/nitrogen-14 ratio.)

Two additional factors complicate the timing of the changes in methane concentrations. First, atmospheric mixing and the lifetime of methane smooths the impact of an abrupt increase in methane emissions on atmospheric methane concentration over Greenland. Second, methane moves through the firn more quickly than N2 because of its greater diffusivity. To account for these factors the authors augment the observed data with the results of modeling. They use a simple atmospheric model to account for methane mixing and sinks in the atmosphere. They use a firn model to take into account the relative rates of diffusion of nitrogen and methane in the firn.

The results show that methane emissions and Greenland temperature changed essentially synchronously to within decades. It was not possible to determine whether one led the other.

Temperature and Methane Concentration
Temperature and Methane Concentration
Temperature and methane concentration
Error bars represent the standard deviations of the temperature proxy (blue) and methane (red) measurements. The black vertical line marks the synchronous increase in both gases associated with the abrupt climate change.

Methane could lead temperature

The uncertainty estimate suggest that methane sources could lead Greenland temperature by up to 24 years. This is in agreement with the conclusions of Steffensen, J. P. et al. High-resolution Greenland ice core data show abrupt climate change happens in few years. Science 321, 680–684 (2008) , which found a 10 year( ± 5 years) year lead of tropical regions over high-latitude sites by analyzing the North Greenland Ice Core Project (NGRIP) ice core.

Temperature could lead methane

Alternatively, methane could lag temperature by up to 21 years. This would be partially in agreement with the previous estimates of Severinghaus, J. & Brook, E. Abrupt climate change at the end of the last glacial period inferred from trapped air in polar ice. Science 286, 930–934 (1999) based on lower-resolution and lower-precision ice core gas data from the Greenland GISP2 ice cores in which a 20–80 year lag of tropical methane emissions behind the temperature proxy was found for the Bølling–Allerød warming event.


Three important conclusions can be drawn from these results. First of all, since methane sources are presumed not to be located close of Greenland, synchronous changes in temperature and methane emissions require the rapid transmission of the abrupt Bølling–Allerød climate change over large geographic areas on a hemispheric scale. To determine the cause of such abrupt temperature change requires very precise phase analyses. The uncertainties in estimating methane mixing in the atmosphere and diffusion of gases in the firn complicate the interpretation of very high-resolution ice core data. This introduces uncertainty into the interpretation ice core gas records at time intervals less than decades.

Editor's comments

In the paleoclimate record warming events during which surface temperatures increase by several degrees Celsius can occur very rapidly, over decades. It is not clear what initiates these events and what determines whether the warming is followed by a cooling period or leads to a full interglacial. For example, one proposal is that variations in solar radiation (Milankovitch cycles) are responsible for initiating the warming period, but that ice sheet dynamics determine what follows. Another is that a sudden release of fresh water into the North Atlantic affects North Atlantic ocean currents. Another is that increasing greenhouse gases in the atmosphere lead to global warming. In this article the authors analyze Greenland ice cores corresponding to a warming event during the last years of the last glacial period to shed some light on rapid climate change.


An ice core record of near-synchronous global climate changes at the Bølling transition, Julia L. Rosen, Edward J. Brook, Jeffrey P. Severinghaus, Thomas Blunier, Logan E. Mitchell, James E. Lee, Jon S. Edwards and Vasileios Gkinis, Nature Geoscience 7, 459–463 (2014) doi:10.1038/ngeo2147

For important articles Nature provides a review in the same issue that helps set a broader context for the primary article. In this case the review article was contributed by Eric W. Wolff, a respected researcher in paleoclimatology.
Palaeoclimate: Climate in phase, Eric W. Wolff, Nature Geoscience 7, 397–398 (2014) doi:10.1038/ngeo2165