Australia: The Land Where Time Began

A biography of the Australian continent 

8.2 ka Event from the Greenland Ice Core

In this paper Thomas presents a collection of chemistry and stable isotope records that are of high resolution from the plateau of the Greenland ice cap during the cold event that occurred at 8.2 ka. Thomas used a composite of 4 records in the study of the cold event that was observed as a 160 year period during which there was a central event that lasted 69 years when there were consistently more than 1 standard deviation below the average for the preceding period. Differences at decadal and shorter time scales were shown from 4 cores in north, south and central Greenland; it is not yet clear if this represents significant spatial differences in response. The change in chemical concentrations for the ions looked at here are small, however, it suggests there were only minor changes in the circulation of the atmosphere for this event. Except for the decrease in the concentration of methane, ice cores from Greenland provide only weak evidence for effects outside the North Atlantic region.

Much has been written about the event that is recorded in palaeoclimate records at approximately 8,200 years before present (BP), that is often called the 8.2 ka or 8 ka event, (Alley & Agustsdottir, 2005; Rohling & Palike, 2005). This was the outstanding climate cooling of the Holocene in the Greenland ice core records, where it was first clearly noted. As observed in the Greenland ice cores the 8.2 ka BP event has been described as being a rapid cooling of 6 ± 2oC at Summit, Greenland (Johnsen et al., 1992; Dansgaard, 1993; Alley et al., 1997). The use of δ15N as a calibration tool in a more recent study confirms that the isotopic anomaly is the result of a change in temperature, with a best estimate of 7.4 K for the temperature change (Leuenberger et al., 1999). It has now been documented that there was a cold, dry event in the North Atlantic regions, though there is still doubt about its extent and cause.

It was proposed (Barber et al., 1999) that it is implied by the pattern of cooling that the transfer of heat from ocean to atmosphere was reduced in the North Atlantic. A decrease in deep water formation and ultimately a cooling at high latitudes (Clark, 2001), would result from an increase in the flux of freshwater. According to Thomas this amplification in freshwater budget in the North Atlantic is generally attributed to the final stages of the deglaciation of the Laurentide and Scandinavian ice sheets.

It was concluded (Barber et al., 1999) that the massive outflow of freshwater that would be required to reduce the formation of deep water originated from glacial lakes of Agassiz and Ojibway, northeastern Canada, which drained catastrophically an estimated 8,470 calendar year BP (⁓7,700 14C years BP) which released 2 x 1014 m3 of lake water over a period of 100 years, at most (Barber et al., 1999; Clark, 2001). Simulations (Clark et al., 2004) of subglacial drainage for Lake Agassiz led to the suggestion that the magnitude of the outbursts was ⁓5 Sv that occurred in less than 6 months. Throughout the Labrador Sea there is correlative evidence that is observed as lower salinities of sea surface waters and increased water stratification (Knudsen et al., 2004).

An atmosphere-sea-ice-ocean model was used (Renssen et al., 2001) to study the mechanism behind the 8.2 ka BP event. They concluded that resulting from a freshwater pulse that was associated with the final stages of the deglaciation of North America weakening of the thermohaline circulation would indeed produce a model response of cooling of the atmosphere which is consistent with the proxy data. A fully coupled atmosphere-ocean global model was used (Legrande et al., 2006) to simulate a short period when the formation of North Atlantic deep water, after a pulse of freshwater, which quantitatively matched palaeoclimate observations. Thomas suggests that if this is indeed the case, then there is a particular significance in the event as the clearest example, under geographic conditions that was similar to those of the present, of the climatic effect of changes in the thermohaline circulation.

There is considerable confusion in the literature, however, what the 8.2 ka event really is. The relatively short duration and rapidity of the 8.2 ka BP event that is recorded in the ice cores from Greenland is often below the temporal resolution of the palaeoclimate data. When dating uncertainties are combined with this it becomes difficult to determine the true change rate, as well as duration and cause of the event at different locations. The contrast between longer anomalies at some sites and the short, high amplitude anomaly around 8.2 ka BP at others were discussed (Alley & Agustsdottir, 2005). It has been pointed out (Rohling & Palike, 2005) that at most locations removed from the North Atlantic, the signals around the 8.2 ka event are relatively small (not outstanding in the Holocene), and of much longer duration (several centuries) then the event that was recorded in the Greenland ice core. The exact relationship between this weak variability and the sharp event that occurred around the North Atlantic is not clear. The description of the 8.2 k event in the literature is not completely clear, even in the region that has a sharp and unique event. Thomas used existing data from 4 Greenland ice cores, as well as new high-resolution isotopic and/or chemical data from 2 of them to define clearly the duration and signature of the 8.2 k event in Greenland ice cores.

The nature and timing of the 8.2 k event


Favourable conditions for the production of annually counted age scales, by the use of various parameters that vary seasonally are provided by high rates of accumulation in all the cores from Greenland, such as 0.24 m ice per year at GISP2 (Meese, 1997) and 0.23 m ice per year at GRIP (Johnsen et al., 1992). The main methods used for the GISP2 core were visual stratigraphy, measurements of electrical conductivity (ECMs), and dust; the minimum of the event is estimated at 8,250 yr BP (Alley et al., 1997). It should be noted that Thomas et al. corrected all calculated ages to a reference “present” of 1950 AD, in spite of different usages that have been employed in the Ice core community. GICC05, the most recent dating of Greenland ice cores (Rasmussen et al., 2006) uses oxygen isotope measurements (Vinther et al., 2006), from the present until the 8.2 ka event, mainly in the high accumulation rate Dye 3 core. According to Thomas et al. the timescales of different ice cores from Greenland can be combined or compared by matching them together by the use of peaks in the ECM, which represent discrete volcanic eruptions (Rasmussen et al., 2006; Vinther et al., 2006). The absolute δ18O minimum for the last 10,000 years in the GRIP ice core was observed at 1,334.05 m depth, which is a calendar date of 8,190 yr BP (8,240 b2k) (before the year 2000 AD), and a counting uncertainty was given as 47 years, from the new GICC05 age scale (Rasmussen et al., 2006). Using a comparison of anomalies in 10Be in the GRIP ice core and 14C in the tree rings (Muscheler et al., 2004) recently estimated the isotope minimum at GRIP to be 8,150 years, with less than 20 years uncertainty. The isotopic minimum was calculated by Thomas et al. to have occurred slightly younger than 8,200 years BP,

The nature of the event in Greenland

The event itself must first be defined in order to determine the true change rate and duration of this event. This has been made difficult by discrepancies in the dating between different palaeoclimate records and ice core records from different sites. Thomas et al. compared the smoothed oxygen isotope data from the GRIP, NGRIP, GISP2, and Dye 3from the Younger Dryas transition to the mid-Holocene; in Fig 2b and subsequent figures in their paper, they zoomed in on the details of the event. The cores were synchronised to the same depth scale in each case by the use of ECM spikes, so they were rather confident that the depths matched to within a few cm.

In Fig 2a of their paper, the event at 8.2 ka BP is seen clearly as the extreme event in the Holocene in all of the 4 cores. There are substantial differences in detail at decadal time scales, though the general shape and duration of the event is similar in all 4 cores. A sharp central spike that is only a decade or so long stands out from the remainder of the event, which makes the GRIP data exceptional. According to Thomas et al. if real, such a sharp event might indicate an even more extreme temperature excursion, or might represent the signal of a freshwater content change in the surface of the ocean, which might be expected if the ocean was flooded with freshwater melt.

They analysed a parallel section of the GRIP core at even higher resolution (1 cm) in order to test this idea further. The existence of the “spike” in the GRIP core is shown by this to be robust, and that it embeds 1-3 years of very low values, which is typical of glacial ice. Analysis of the GISP2 ice core, however, which is 30 km distant from the same section, shows no such spike, at decadal or higher resolution. It could be suggested by this discrepancy between the 2 cores that the isotopic anomaly in the GRIP core was not a period when there extremely cold winters, but more likely an artefact of a build-up of winter snow drift at the GRIP site or snow ablation at the GISP2 site. Whichever it is there is no simular decadal scale event in the GISP2 or either of the other cores, and Thomas et al. did not explore this further. Instead, the event should be seen as a somewhat broader signal around it.

The length of the event

The onset and the end of the event is the next issue to be determined. As a result of different authors using different definitions to define the start and end of the event, this has been a cause of confusion in the literature. Thomas et al. compiled a composite Greenland isotope record for the event in order to tackle the problem. This was done to highlight the common pattern of the signal and they later looked at differences between the sites.

To form the composite record they placed all of the cores onto a GRIP depth scale and used the volcanic ECM peaks as tie points. Throughout the Early Holocene the resolution at which the GRIP data were available was 27.5 cm, and all the cores which are available across the event are at considerably higher resolution than this. Therefore Thomas et al. averaged all of the other datasets to the same resolution, to about 27.5 cm in GRIP depth, approximately 2.5 years. Data across the section that was normalised by Thomas, et al., using the mean and standard deviation of 27.5 cm (GRIP depth) so that they all had the GRIP mean and standard deviation.

Finally, they combined the normalised datasets to give a composite Greenland isotope signal across the event; a 4-point running mean of the 2.5 year resolution data, which represents decadal-scale variability, is also shown (Fig. 4 in their paper). A Greenland spatial average is not represented by this as the cores are not fully representative, though the most common prominent features of the climate change that was occurring at this time should nonetheless be defined by this.

In order to determine when the event started and ended Thomas et al. calculated the mean (34.73‰) and standard deviation (0.83‰) of the GRIP 27.5 cm data for the 1,000 year period between 9,300 and 8,300 PB, because the data from the other sites had been normalised to GRIP this should also represent the statistic for the composite.

They could then define the 2 intervals. The composite 27.5 cm isotope values were considerably below the mean for the preceding 1,000 years between 1324.77 1349.12 m. According to Thomas et al. this can be seen as the maximum extent of the climate cooling anomaly. It represents 159 years on the GICC05 timescale. They also defined a “central event“ during which the composite signal is consistently more than 1 standard deviation below the average of the previous 1,000 years. The core period covers 70 years on the GICC05 timescale.

As defined in this study, the 8.2 ka event is asymmetrical in shape with a considerable amount of variability in the record, as is shown by the presence of spikes that are relatively warm at about 8,200 and 8,160 BP. Climate models of THC weakening have replicated the asymmetrical shape (Manabe & Stouffer, 1995; Renssen et al., 2001) and the relatively warm phase in the early part of the cold event has been observed (Wiersma & Renssen, 2006) in model experiments that show temporal strengthening of the THC approximately 30 years after the introduction of the freshwater perturbation.

There are differences on the decadal and shorter resolution, in spite of similarities in the shape of the low resolution isotope record in each of the Greenland cores, and therefore the assessment of Thomas et al. of the onset of termination depths for the 8.2 ka event would be different in each of the cores. According to Thomas et al. it appears if the early part of the event is most prominent in the NGRIP, and the event was completed slightly earlier at Dye 3 in the south. It might be indicated by this that there was some kind of north-south progression of the event. In order to assess the significance of these findings more duplicate cores in each region would be needed.

Discussion and concluding remarks

In central Greenland the cold event that occurred 8,200 years ago is observed as a period of reduced isotopic ratios that lasted for a period 160.5 years, within which there is a central event of 69 years during which values were significantly lower than the average for the Holocene. When the isotopic record from 4 cores, 2 of which were 30 km apart, were compared, it highlighted the variability between the records at decadal and shorter timescales; though it is yet to be made clear if significant spatial differences are represented by this in response. Smaller changes in chemical composition over Greenland that accompanied this cold event are shown than those that had been previously reported, by re-examination of chemical data, based on a different methodology. According to Thomas et al. The small increase in chloride could reflect a small increase in storminess, or the production of sea ice at the onset of colder conditions around Greenland; changes in conditions in Asia or in the strength of transport could be a reflection of the small increase of calcium. It is worth noting, however, that similar increased Calcium concentrations are also observed in the GISP2 dataset between 5 and 6 ka, which are not accompanied by an isotopic signal, and that there is an increase in the chloride, which is not unusual in the Holocene. It was concluded by Thomas et al. that the evidence for a significant change (outside the Holocene norm), beyond the North Atlantic region during the 8.2 ka event, is weak. Only methane (Blunier et al., 1995), which is not discussed in this paper, shows a clear signal that might be derived from further away.

The 8.2 ka event has often been compared to the Younger Dryas, which was a much larger event (Alley et al., 1997; Alley & Agustsdottir, 2005) which punctuated the termination of the last glacial with a mechanism that has been proposed of flood outburst from the final deglaciation of the Laurentide Ice Sheet (Broecker et al., 1988; Alley, 2000), with a similar mechanism being proposed for the 8.2 ka event.

The estimated change in temperature at the 8.2 ka event is about 40% of that at the Younger Dryas, and the changes in the rate of accumulation and methane display a similar proportionality. The increases in the deposition of chemicals at the transition of the Younger Dryas, however, exceeded 600% for calcium and 200% for chloride.

During the 8.2 ka event the small increase in calcium and chloride (about 5% of that observed in the Younger Dryas suggests that the larger scale atmospheric response during the event was very subdued, and certainly not proportional to the local climatic signal in the North Atlantic.

At present interest in this event is high, which has increased the numbers of palaeoclimate data that are recorded for this period, which is viewed as a suitable test bed for the ability of models to predict the climatic effects of thermohaline circulation changes. It is therefore important that exactly which event is being referred to when palaeoclimate findings are reported. In Greenland there was a definite cold period at 8.2 ka as is indicated by the decrease in isotope ratios and the rates of accumulation that were observed at all 4 deep ice cores that were drilled in Greenland, and by the evidence from 15N in N2 in air bubbles, confirming that there was indeed a temperature decrease. Across northern Europe there are several corroborative records which suggest that these cold conditions spread to locations further than just Greenland. Beyond the North Atlantic, however, the evidence that is being uncovered shows the same sharp event becomes weaker. It has been suggested previously that the event (Alley & Agustsdottir, 2005; Rohling & Palike, 2005) observed as a sudden change in records such as the Greenland one is superimposed on a longer event that was less unique. Thomas et al. concur with this view: while the broad context of the variability of climate is clearly important, they recommend that the term “8.2 ka event” be reserved for signals of a timing and duration that is similar to that reported in this paper. 


  1. Thomas, E. R. (2007). "The 8.2 ka event from Greenland ice cores." Quat. Sci. Rev. 26: 70-81.


Author: M. H. Monroe
Last Updated 03/05/2018
Journey Back Through Time
Experience Australia
Aboriginal Australia
National Parks
Photo Galleries
Site Map
                                                                                           Author: M.H.Monroe  Email:     Sources & Further reading