To main QEN pages

To new maps homepage

Sudden climate transitions during the Quaternary

By Jonathan Adams, MS 6335, Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA


The time span of the Quaternary has been punctuated by a large number of relatively sudden climate transitions. Most of these seem to have taken a few centuries at most, and it is looking increasingly possible that many occurred over just a few decades or even several years. The clearest information is available on the large Younger Dryas-to-Holocene change around 11,500 years ago, which seems to have occurred over several decades, in a series of steps. It is quite likely that the speed of this change is representative of other similar but less well-studied climate transitions during the last 130,000 years, involving sudden cold events (Heinrich events/stadials), warm events (Interstadials) and the beginning and ending of long warm periods, such as the Eemian interglacial. However, detailed analysis of the climate record will be necessary before it is possible to say confidently what timescale these other events occurred on. At the very most, these changes seem to have occurred over a few centuries. Significant rapid climate transitions also occurred during the present (Holocene) interglacial, with cold and dry phases occurring on a 1500-year cycle. Again, the timescale of the transitions is uncertain, but a decade-to-century timescale seems most likely from the resolution of the records. In the past few centuries, smaller transitions (such as the ending of the Little Ice Age) also seem to have occurred over only a decade or two. Whether climate can undergo even greater cooling during an interglacial is uncertain; however some apparently rapid and very severe cold phases seem to have occurred during previous interglacials. In general it seems that climate tends to undergo most of its changes in sudden jumps rather than incremental century-on-century changes.

Various mechanisms, involving changes in ocean circulation, dust fluxes from the land surface, and changes in snow cover and water vapour content of the atmosphere, have been invoked to explain the sudden regional and global transitions which occur over a decade-to-century timescale. There is the worrying possibility that human-induced climate change could occur in a sudden jump, taking a few decades or less. One scenario, backed up by model simulations, is that the North Atlantic circulation could suddenly either weaken or intensify due to a change in river runoff or melting of ice. This might, for example, send Europe and perhaps the whole world, into a short-lived ice age.

Broad climate variability; the background of oscillations on the timescale of tens of thousands of years

Climatic variability on the timescale of tens of thousands of years has turned out to be a predominant pattern in recent earth history. The two and a half million years known as the Quaternary have been marked by large numbers of global climate oscillations, between warmer and cooler conditions. This trend of oscillations appears to be merely the continuation of a pattern of variability extending back well into the Tertiary period and possibly beyond (Kennett 1995). During the timespan of the Quaternary, the length and the amplitude of these climate cycles has increased (Crowley & North 1991).

A large global interglacial-glacial-interglacial climate oscillation has been recurring on approximately a 100,000 year periodicity for about the last 900,000 years, though each individual cycle has had its own idiosyncrasies in terms of the timing and magnitude of changes. As is usually the case with the study of the past, data become less precise and in shorter supply as one goes back further in time, and only a few sketchy outlines are known for the earliest cycles (Winograd et al. 1997). Because it offers a relatively detailed climate record from the land, from the oceans and from the ice cores, the most recent large climate oscillation spanning the last 130,000 years has been the subject of most study. This interval has seen the global climate system switch (Fig.1) from warm interglacial to cold glacial conditions, and back again. In terms of stratigraphic naming, this oscillation begins with the previous Eemian interglacial (Isotope Stage 5e), spans the colder last glacial period (Stages 5d-2), and the present Holocene Interglacial beginning around 11,000 years ago (Isotope Stage 1). There are some signs that similar events and processes were at work in previous glacial-interglacial cycles over the past 900,000 years, and the general background of continuous (but less dramatic) variability in the earth's climate system is thought to extend well back into the Tertiary, if not beyond (ref.).

Even for this most recent Eemian-to-Holocene phase there is still too much ambiguity in terms of the errors in geological dating techniques, in the gaps in the record, and in the slowness of responses by indicator species, to know precisely when certain events occurred and whether the climate changes were truly synchronous between different regions. The general picture summarized here roughly reflects the present consensus gained from ice cores, deep ocean cores, and terrestrial and lake sediments around the world. This is a consensus amongst the scientific community which itself is subject to sudden jumps when new data are presented, or as more thorough reanalyses of previous data come forth; for this reason, this review is liable to be significantly out of date within only a few weeks or months of being written!

The record of decade-timescale changes during the last 130,000 years

The Eemian. The Eemian interglacial (Fig.1) seems to have begun around 130,000 years ago with a rapid global warming (of uncertain duration) taking the earth out of an extreme glacial phase, into conditions generally warmer than those of today (ref.). There are various indications of climate instability during the Eemian itself, but they are controversial. The initial indications of the GRIP ice core evidence (DAnsgaard et al. 1993, Taylor et al. 1993) were that the 15,000-to-10,000 year span of the Eemian was punctuated by a large number of short-lived cold events, at least in the Greenland area but possibly elsewhere in the world. This conclusion was based on analysing the isotope chemistry of the ice (which formed in annual layers like the growth rings of a tree) to infer the climatic conditions which prevailed in the Greenland region as the ice was being layed down as snow.

Many of the putative cold events that show up in the ice core involved a rapid (taking a few decades or less) decline in temperatures from conditions warmer than today to conditions more closely resembling a full glacial phase. These brief cold events lasted anything between several decades and several thousand years before ending suddenly, with no clear pattern in terms of either their timing or duration.

A second ice core (GISP2) taken from another point on the Greenland ice cap has given a superficially similar, but significantly different, picture of the climate history (Grootes et al. 1993). For the last 115,000 years, every detail of the two ice cores is identical, suggesting that they have both faithfully recorded the climate history back to that point in time. For the period before about 115,000 years ago both cores do show evidence of dramatic climate instability, but the timing of the events is different. This suggests that in fact neither of the two ice cores is correct during most of the Eemian Interglacial, and that the only reason they show evidence of instability is that the oldest ice layers at the base of the ice sheet have become churned up as it flowed over the uneven bedrock beneath. The 'cold climate' events in the midst of the warm-climate Eemian could then be just older ice from the previous full-glacial cold phase that has been folded and up-faulted into the warmer-climate Eemian layers. Although at present it is perhaps too soon to say that neither ice core is correct in any of its details regarding Eemian climate instability, they cannot both be right in all of their details. There are presently plans to test the validity of these long Greenland cores by drilling into the West Antarctic ice sheet, which also accumulates snow rapidly enough to give a long detailed annual record extending back to the Eemian.

If the Eemian was indeed as unstable as the ice core records suggest it might have been, one would expect to see evidence of very frequent rapid climate changes hitting the forests of Europe during this period. In general, such evidence is conspicuously absent from the various long pollen records that extend through this period, and from sea surface temperature indicators in cores in the north Atlantic ocean. Thouvenay et al. (1994) have suggested that they had found evidence for such changes in lake sedimentology from central France during the Eemian, but there appears to be some doubt as to whether their lake record actually extends back this far (they might be seeing variability from the later early glacial stage).

However, evidence for a single sudden cool event during the Eemian is clearly present in a more solidly dated pollen record from a lake in central Europe studied by Field et al. (1994), from loess sedimentology in central China (Zhisheng & Porter 1997), and from certain ocean sediment records in the northern Atlantic (ODP site 658). These three sources of evidence each suggest a single cold and dry event (causing a several-degree decline in Atlantic surface temperature, and on land opening up the west European forests to give a mixture of steppe and trees) near the middle of the Eemian, about 121,000 y.a. It was followed by a return to warm conditions similar to the present. This cold event does not seem to correspond closely in time with any particular cold interval visible in the disturbed GRIP or GISP2 ice core record, but it does show that the Eemian was subject to some rapid changes. How long this mid-Eemian cold event lasted (though several centuries seems likely according to preliminary estimates), and whether it came on or ended suddenly over only a few decades, is unclear at present. Note that a North Atlantic core studied by Adkins et al. (1997) seems to give a contradictory picture, with the final cooling event at the end of the Eemian (dated from other long records to about 110,000 y.a., occurring about 118,000 years ago. Whether this 'end of Eemian' cooling is in fact the same as the within-the-Eemian cooling event seen from the other records remains to be seen.

Sudden transitions after 115,000 years ago. The Eemian interglacial seems to have ended in a sudden cooling event about 110,000 years ago (though the chronology is uncertain), recorded from Ice cores, ocean sediment cores and pollen records from across Eurasia. From a relatively high resolution core in the North Atlantic, Adkins et al. (1997) suggest that the final cooling event took less than 400 years, but the limitations on the resolution of this record leave open the possibility that it was actually a much more rapid event.

Following the end of the Eemian, a large number of other sudden changes and short-lived warm and cold events have been documented. The most extreme of these are the warm interstadials and the cold Heinrich events, which are most prominent in the ice-core record of Greenland and the pollen records of Europe, suggesting that they were most intense in the North Atlantic region.

Interstadials. Sudden and short-lived warm events occurred many times during the generally colder conditions that prevailed between 110,000 and 10,000 years ago. First picked up as brief influxes of warm climate plants and insects into the glacial tundra zone of northern Europe, they are known as 'interstadials' to mark them off from the cold phases or 'stadials'. The interstadials show up strongly in the Greenland ice core records. For the time period between 115,000 and 14,000 years ago, 24 of these short lived warm events have so far been recognized from the Greenland ice core data (where they are called 'Dansgaard-Oeschger events'), although many lesser warming events also occurred (Dansgaard et al. 1993). Similar short-lived warm phases also seem to appear in the eastern Pacific (Behl & Kennett 1996), western Siberia, and possibly also in central China (ref.s). The duration of each interstadial can be counted from the annual layers that have accumulated in the ice, or (rather less precisely) from the thickness of sediment accumulated in an ocean bed core. Ice core and ocean data suggest that they both began and ended suddenly, though in general the 'jump' in climate at the start of an interstadial was followed by a more gradual decline involving a stepwise series of smaller cooling events and often a fairly large terminal cooling event which returned conditions to the colder 'glacial' state (e.g. Ramussen et al. 1997). From the ice core evidence from Greenland, it seems that the warming into each interstadial occurred over a few decades or less, and the overall duration of some of these warm phases may have been just a few decades, while others vary in length from a few centuries to nearly two thousand years. According to the GISP2 ice core data, interstadials seem to show a pattern in their timing (see Yang et al., in JGR-atmospheres, October, and Mayewski et al. in JGR special GISP2 issue due out in November).

Heinrich events. Opposite in sign to the Interstadials are extreme and short-lived cold events, known as 'Heinrich events'. These also occur against the general background of the glacial climate, and they represent the brief expression of the most extreme glacial conditions. The several massive ice-rafting events (Heinrich 1988) (Table 1) also show up in the Greenland ice cores as a further 3-6 deg.C drop in temperature from the already relatively cold glacial climate, and many of these events also been picked up as particularly cold and arid intervals in Europe (ref.). It is thought that at least some of these events also affected the climate further afield from Greenland - giving cold, arid conditions, as far away as central China and Antarctica (ref.). Preliminary data (C. Whitlock..web page) from marine sediments off California and Oregon, pollen records from Pacific Northwest lakes, and glacial records from western North America reveal climate events that appear to be related to the Heinrich events of the North Atlantic. Thus it appears that these may have been global events although the climate shift may have been smaller outside the North Atlantic region. The last Heinrich event (known as H1) sensu stricto occurred just after the Last Glacial Maximum and seems to mark the extreme cold and aridity that occurs in many parts of the world around 17,000-15,000 years ago. The Younger Dryas cold phase (see below) may also be regarded as a 'Heinrich' climate event (it is sometimes now referred to as H0), and as it has been studied in considerable detail it may give clues to a general pattern true to all Heinrich climate events. The detailed timescale on which these most of the ice rafting and climate change events began and ended is uncertain, though each changes clearly took less than centuries (and quite possibly decades) to occur. By analogy with the well-studied Younger Dryas event (below), Heinrich events may be regarded as possibly beginning and ending with sudden climate 'jumps' taking just a few decades. However, this idea remains tentative without further work.

Table 1

Timing of major Heinrich events during the last 130,000 years

YD or H0 12.0 ka (calendar age)

H1 16.5

H2 23.0

H3 29.0

H4 37.0

H5 50.0?

H6 70.0?

YD: Younger Dryas

H: Heinrich event

Lower amplitude background variability In the Greenland ice core and North Atlantic sediment records there is also a general background of more frequent but lower amplitude changes, involving rapid but relatively small temperature fluctuations. This 'background' variability is only just beginning to be studied, and the detailed speed with which changes occurred is uncertain. By analogy with the Younger Dryas and other relatively well-studied climate changes, they may be regarded as 'possible' human-lifetime scale climate changes. A recent analysis of North Atlantic ocean sediments suggests that at least during the last 30,000 years or so, temperatures in this region varied on a 1500-year cycle, with relatively sudden cooling events at the coldest point of each cycle (Bond et al. 1997).

Recently, detailed studies of the sequence of events in ocean sediments and ice cores shows that smaller ice-sheet surges and cooling events occurred around the rim of the North Atlantic about every 7,000-10,000 years on average, in the time interval between about 70,000 and 10,500 years ago. In the later stage of this time span, after about 38,000 y.a., they were even more frequent at about every 1,000-3,000 years, apparently following a mean cycle length of about 1500 years (Bond et al. 1997).

The cold Heinrich events seem to have been extreme runaway instances of these cyclic cooling events, while warm interstadials tended to begin at the warmest part of the cycle. A similar 1500-year cycle seems have operated during the Holocene (Bond et al. 1997), and both may share a common climate mechanism - perhaps an 'oscillator' in the North Atlantic deepwater formation system (see below).

The Younger Dryas. The Younger Dryas cold event at about 12,900-11,500 years ago seems to have had the general features of a Heinrich Event, and may in fact be regarded as the most recent of these. The sudden onset and ending of the Younger Dryas has been studied in particular detail in the ice core and sediment records, and it might represent a pattern that has prevailed with other Heinrich events and other climate transitions during the last 100,000 years or so. A new detailed study of two Greenland ice cores (GRIP and GISP2), just published in Science (Taylor et al. 1997), suggests that the main Younger Dryas-to-Holocene warming took several decades in the Arctic, but was marked by a series of sudden steps in warming, each taking less than 5 years. About half of the warming was concentrated into a single period of less than 15 years. A rapid global rise in methane production at the same time suggests that the warming and moistening of climate (causing more methane output from swamps and other biotic sources) was a globally synchronized change. The water vapour content of the atmosphere is one likely 'messenger' in this global transition, by virtue of its effect as a greenhouse gas (see below). However, the detailed chronology of different environmental indicators within the cores also suggests that changes in lower latitude temperature and dust flux from the continents preceded the change in Greenland temperatures that relates closely to the northern thermohaline circulation, and that dust and haze might have been a key factors in triggering or amplifying the global transition. According to the Greenland ice-cores, conditions remained slightly cooler than present for a while after the main warming period; 'normal' Holocene warmth was not reached for a further 1500 years (up until around 10,000 calendar years ago).

It is not yet clear if the general pattern of the transition between the Younger Dryas and Holocene is representative of other rapid warming and cooling events in the past 110,000 years. Not all of these events have been studied in as much detail as the Younger Dryas, but those transitions which have been well studied using very high-resolution records seem to have occurred over only a few decades (needs ref.s here).

Other possibly sudden climate transitions since the start of the Holocene. Following the sudden start of the Holocene about 11,500 years ago, there have been a number of sudden, widespread climate changes recorded from the palaeoclimatic record around the world. The most striking of these is a sudden cooling event, about 8,200 years ago and giving cool, dry conditions lasting perhaps 200 years before a rapid return to conditions warmer (and generally moister) than the present. This event is detectable in the Greenland ice cores, where the cooling seems to have been about half-way as severe as the Younger Dryas-to-Holocene difference. No detailed assessment of the speed of change involved seems to have been made within the literature (though it should be possible to make such assessments from the ice core record), but the short duration of these events at least suggests changes that took only a few decades to occur. What was apparently the same event shows up in records from North Africa across Southern Asia, as a phase of markedly more arid conditions involving a failiure of the summer monsoon rains. Cold and/or aridity also seems to have hit northernmost South America, eastern North America and parts of NW Europe. Smaller, but also sudden and widespread, changes to drier or moister conditions have also been noted for many parts of the world for the second half of the Holocene, since about 5,000 years ago. One particularly strong arid event occurred about 4,000 years ago across northern Africa and southern Asia.

Bond et al. (1997) have recently found evidence that at least in the North Atlantic region, and possibly globally, the Holocene climate has been affected by a strong 1500-year cycle, with sudden cold events lasting a century or two during the 'low' part of each cycle. Generally the coldest point of the cycle was about 2 deg.C cooler than the warmest part, representing a fairly substantial change in climate. It is uncertain whether these climate cycles extended around the world or were generally confined to the region around the North Atlantic, but the 8,200 y.a. event (which fits in as one of the cold events of this 1500-year pattern) does seem to have been almost global and it may merely have been the most extreme of these.

Different sources seem to suggest differing speeds and intensities for Holocene climate events. According to chemical indicators of windblown sea salt in the GISP2 ice core, the Little Ice Age - which began in late Medieval times and ended in the early 1800's - may have been the most rapid and largest change in polar circulation during the Holocene (O'Brien et al. 1997) (C. Wake pers. comm.). If so, this would seem to imply that an event which was clearly intense in some regions (such as the dry phase around 8,200 years ago across so many low and mid-latitude regions), was not relatively so important in other regions (such as near to the poles).

Other sudden climate jumps from the more recent and more distant past

The ocean sediment evidence of Bond et al. (1997) suggests that various cool events recorded from the Holocene record fit into a general 1500-year temperature cycle which is recorded in the North Atlantic; if so, there may be several other widespread cool events which have not yet been detected in records from the continents. These 1500-year cycles, involving about a 2 deg.C change in mean surface temperature of the North Atlantic, would be major events if they were to suddenly affect the present-day world with its high population and finely balanced food production. These periodic Holocene cool phases seem to have come on rapidly, taking less than a century and possibly just a few decades.

Recently interpreted evidence from the GRIP2 ice core (above) also suggests that the most intense phases of the Little Ice Age came on and ended suddenly, over just a few decades. The Little Ice Age was another climate oscillation (fairly small by comparison with many of the events recorded in ice cores and sediment records) which gave cooler conditions over the lands around the North Atlantic between about 700 and 200 years ago.

Much larger changes in climate seem to have occurred during previous interglacial phases (Winograd et al. 1997). For example, a quite severe cold and arid event may have affected Eurasia (and possibly other parts of the world) during the Eemian Interglacial about 121,000 years ago. Again, whether the onset and ending of this event was as rapid as only a few decades is not known at present. Other relatively sudden cool and arid phases (occurring against a background of similar-to-present conditions) seem to have affected some of the previous interglacials before about 200,000 years ago. Again, the speed with which these climate transitions occurred does not seem to have been discussed in the ice-core literature, but the possibility that these changes occurred over only a few decades must be considered a possibility.

Other smaller changes are observed in the detailed Greenland ice cap record, but it is important to note that not all the rapid changes observed in the Greenland ice cap correspond to large climate changes elsewhere. For example, a warming of 4 deg.C per decade was observed in an ice core from northern Greenland for the 1920's (Dansgaard et al. 1989), but this corresponded to a global shift of 0.5 deg.C or less. For this reason it is always desirable to have sources of evidence from other regions before invoking a broad, dramatic climate shift. What this relatively recent climate shift does suggest though, is that the climate system tends to undergo most of its changes in sudden jumps, even if those changes are relatively small against the background of those seen during the Quaternary. This is further evidence that if and when the next climate shift occurs, it will not be a gradual century-on-century change but rather a sudden step-function that will begin suddenly and occur over a decade or two.

The mechanisms behind sudden climate transitions. It is still unclear how the climate on a regional or even global scale can change as rapidly as present evidence suggests. It appears that the climate system is more delicately balanced than had previously been thought, linked by a cascade of powerful mechanisms that can amplify a small initial change into a much larger shift in temperature and aridity. At present, the thinking of climatologists tends to emphasise several key components:

North Atlantic circulation as a trigger or an amplifier in rapid climate changes. The circulation of the north Atlantic Ocean is presently seen as playing a major role in either triggering or amplifying rapid climate changes in the historical and recent geological record. The North Atlantic has a peculiar circulation pattern; the north-east trending Gulf Stream carries warm and relatively salty surface water from the Gulf of Mexico up to the seas between Greenland, Iceland and Norway. Upon reaching there, the surface water cools off and (with the combination of being cooler and relatively salty) becomes dense enough to sink into the deep ocean. The 'pull' exerted by this dense sinking water is thought to help maintain the strength of the warm Gulf Stream, ensuring a current of warm tropical water into the north Atlantic that sends mild air masses across to the European continent.

If the sinking process in the north Atlantic were to diminish or cease, the weakening of the warm Gulf Stream would mean that Europe had colder winters. However, the Gulf Stream does not give markedly warmer summers in Europe - more the opposite in fact - so a shutting off of the mild Gulf Stream air masses does not in itself explain why summers also become colder during sudden cooling events (and why ice masses start to build up on land due to winter snows failing to melt during summer). In the north Atlantic itself, sea ice would form more readily in the cooler winter waters due to a shut-off of the Gulf Stream, and for a greater part of the year the ice would form a continuous lid over the north Atlantic. A lid of sea ice over the North Atlantic would last for a greater proportion of the year; this would reflect back solar heat, leading to cooler summers on the adjacent landmass as well as colder winters. With cooler summers, snow cover would last longer into the spring, further cooling the climate by reflecting back the sun's heat. The rapid result of all this would be a European and west Siberian climate that was substantially colder (because the warm Gulf Stream air was diverted away by the shutting down of the North Atlantic circulation, and by a high-pressure region formed over the sea ice lid) and substantially drier (because the air that reached Europe would carry less moisture, having come from a cold sea ice surface rather than the warm Gulf Stream).

After an initial rapid cooling event, the colder summers would also tend to allow the snow to build up year-on-year into a Scandinavean ice sheet, and as the ice built up it would reflect more of the sun's heat, further cooling the land surface, and giving a massive high pressure zone that would be even more effective at diverting Gulf Stream air and moisture away from the mid-latititudes of Europe. This would reinforce a much colder regional climate.

The trigger for a sudden 'switching off' of deep water formation in the North Atlantic could take the form of an exceptionally wet year on the landmasses which have rivers draining into the Arctic sea (Siberia, Canada, Alaska). Ocean circulation modelling studies suggest that a relatively small increase in freshwater flux to the Arctic Sea could cause deep water production in the north Atlantic to cease. During glacial phases, the trigger for a shut-off could be the sudden emptying into the northern seas of a lake formed along the edge of a large ice sheet on land (for instance, the very large ice-dammed lake that existed in western Siberia), or a diversion of a meltwater stream into the path of the Gulf Stream (as seems to have occurred as part of the trigger for the Younger Dryas cold event). A pulse of fresh river water would dilute the dense, salty Gulf Stream and float on top, forming a temporary lid that stopped the sinking and pulling of water that drives the Gulf Stream. The Gulf Stream could weaken or switch off altogether, breaking the 'conveyer belt' and allowing a sea ice cap to form, preventing the Gulf Steam from starting up again. Theoretically, the whole process could occur very rapidly, in the space of just a few decades or even several years. The result could be a very sudden climate change to colder conditions, as has happened many times in the area around the North Atlantic during the last 100,000 years.

The sudden switch could also occur in the opposite direction, for example if warmer summers caused the sea ice to melt back to a critical point where the sea ice lid vanished and the Gulf Stream was able to start up again. Indeed, following an initial cooling event the evaporation of water vapour in the tropical Atlantic could result in an 'oscillator' whereby the salinity of Atlantic Ocean surface water (unable to sink into the north Atlantic because of the lid of sea ice) built up to a point where strong sinking began to occur anyway at the edges of the sea ice zone. The onset of sinking could result in a renewed northward flux of warm water and air to the north Atlantic, giving a sudden switch to warmer climates, as is observed many times within the record of the last 130,000 years or so.

If the Gulf Stream switched off, it would not only affect Europe. Antarctica would be even colder than it is now, because much of the heat that it does receive ultimately comes from Gulf Stream water that sinks in the north Atlantic, travels in a sort of river down the western side of the deep Atlantic Basin and then resurfaces just off the bays of the Antarctic coastline. Even though it is only a few degrees above freezing when it reaches the surface, this water is much warmer than the adjacent Antarctic continent, helping to melt back some of the sea ice that forms around Antarctica. The effect of switching off the deepwater heat source would be cooler air and a greater sea ice extent around Antarctica, reflecting more sunlight and further cooling the region. However, the north Atlantic deep water takes several hundred years to travel from its place of origin to the Antarctic coast, so it would only produce a direct effect a few centuries after the change occurred in the North. It is not known what delay was present in the correlated climate changes between the north Atlantic region and Antarctica, but it is generally thought that other (relatively indirect) climate mechanisms, such as greenhouse gases in the atmosphere, linked these two far-flung regions and produced rather more closely synchronised changes.

The idea of Gulf Stream slowdowns as a mechanism in climate change is not merely theoretical. There is actually evidence from the study of ocean sediments that deepwater formation in the north Atlantic was diminished during the sudden cold Heinrich events and other colder phases of the last 130,000 years, including the Younger Dryas phase. The process also 'switched on' rapidly at times when climates suddenly warmed around the north Atlantic Basin, such as at the beginning of interstadials or the beginning of the present interglacial (Ramussen et al. 1997). Decreasing deep water formation occurred at times when the climate was cooling towards the end of an interstadial, and it diminished suddenly with the final cooling event that marked the end of the interstadial (Ramussen et al. 1997), and over a period of less than 300 years at the beginning of the Younger Dryas.

Other direct observations from the last few decades also suggest that deepwater formation off Iceland can slacken slightly in response to a run of wet years around the Arctic Sea, with detectable effects on the European climate. It seems that during other relatively cold phases that do not approach the extreme conditions of the Heinrich events, such as the Little Ice Age event of the last millennium, deep water formation remained in place but that the sinking water was not as dense as it is at present and that a smaller volume was produced. Sinking more gently and in smaller quantities, it would have exerted less of a 'pull' on the Gulf Stream circulation, and hence there would have been been a diminished heat flux northwards from the warm Equatorial Atlantic waters. During the colder glacial phases, deep water formation in the present areas between Greenland, Iceland and Norway would have ceased due to a thick cap of sea ice (though there is evidence it occasionally opened up to let Gulf Stream water through to the sea between Iceland and Norway, this did not result in much deepwater formation and so the pull and the northward heat flux seems to have been small). Instead, during the most intense cold phases the deepwater formation area seems to have moved to the south of the British Isles, at the edge of the extended sea ice zone. Even here, it seems to have been weaker than at present, producing relatively small quantities of rather dilute deepwater. This was probably because the whole surface of the Atlantic Ocean (even the tropics) was cooler; with less evaporation from its surface, even the water that did reach northwards was less briney (and thus less dense), so less able to sink when it reached the cold edge of the sea ice zone. An initial slowdown of north Atlantic circulation may sometimes have been the initial trigger for a set of amplifying factors (see below) that rapidly led to a cooling of the tropical Atlantic, reinforcing the sluggish state of the glacial-age Gulf Stream.

Broader changes in temperature and rainfall over much of the world are thought likely to have occured as a result of a switching on or off of the north Atlantic circulation, and these changes would result in amplification by the feedback mechanisms suggested below. As evidence of such a broader link to global climate, over recent years changes in the monsoon-belt climates of Africa and Asia have also been observed to occur in association with decadal-scale phases of weaker north Atlantic circulation. By extrapolation, it is generally thought that bigger changes in the north Atlantic circulation would result in correspondingly larger changes in climates in the monsoon belts and in other parts of the world.

More about deep ocean circulation and climate

In addition to this relatively direct effect of deepwater on North Atlantic and Antarctic climate, other subtle effects on global climate would be expected to result from a sudden change in north Atlantic circulation, or indeed they may themselves trigger a change in the north Atlantic circulation by their effects on atmospheric processes. These include the interaction with global carbon dioxide concentrations, dust content and surface reflectivity.

Carbon dioxide and methane concentration as a feedback in sudden changes. Analysis of bubbles in ice cores shows that at the peak of glacial phases, CO2 was about 30% lower than during interglacial conditions. This is thought to be due to some change in plankton activity or ocean circulation patterns that occurs under colder climates, drawing more carbon down out of the atmosphere once climate began to cool. The lower carbon dioxide concentrations resulting from this would cool the atmosphere, and allow more snow and ice to accumulate on land. Relatively rapid changes in climate, occurring over a few thousand years, could have resulted from changes in the atmospheric CO2 concentration. The actual importance of carbon dioxide in terms of the climate system is unknown, though computer climate simulations tend to suggest that it directly cooled the world by less than 1 deg.C on average, but due to amplification of this change by various factors within the climate system such as the water vapour content, the resulting change in global climate could have been more than 2 deg.C (ref.)

A problem with invoking carbon dioxide as a causal factor in sudden climate changes is that it generally seems to have varied too slowly, following on the timescale of millennia what often occurred on the timescale of decades. Methane, a less important greenhouse gas, was also 50% lower during glacial phases, probably due to reduced biological activity on the colder, drier land surfaces. However, it does seem to have increased rapidly in concentration in association with changes in climate, reaching its normal Holocene levels in around 150 years or less during the global climate warming at the end of the Younger Dryas, around 11,500 years ago (Taylor et al. 1997). Such sudden rises in methane concentration were probably not important in affecting climate; the warming effect of a 50% change in methane would have been much less than an equivalent change in CO2, because methane is at such a low overall concentration in the atmosphere. It has been suggested that another mechanism, involving sudden and short-lived releases of massive amounts of methane from the ocean floors, could sometimes have resulted in rapid warming phases that do not leave any trace in terms of raised methane levels in the ice core data, where the trapped gas bubbles generally only indicate methane concentrations at a time resolution of centuries rather than the few years or decades that such a 'methane pulse' might last for. However, more recently obtained ice cores from areas where the ice sheet built up particularly rapidly (Chapellaz et al. 1993, Taylor et al. 1997) show a more detailed time resolution of the record of methane concentration in the atmosphere. These records fail to show any evidence of sudden 'bursts' of methane.

Surface reflectivity (albedo) of ice, snow and vegetation. The intensely white surface of sea ice and snow will reflect back much of the sun's heat, hence keeping the surface cool. In general the ice cover on the sea, and the snow cover on the land, have the potential to set off rapid climate changes because they can either appear or disappear rapidly given the right circumstances. The effect of sea ice cover has already been mentioned above as an integral part of the mechanism by which north Atlantic circulation changes may modify climate. Ice sheets are more permanent objects which, whilst they reflect a large proportion of the sunlight that falls upon them, take hundreds of years to melt or build up because of their sheer size. When present, sea ice or snow can have a major effect in cooling regional and global climates, but with a slight change in conditions (e.g. just a slightly warmer summer) they will each disappear rapidly, giving a much greater warming effect because sunlight is now absorbed by the much darker sea or land cover underneath. In an unusually cold year, the opposite could happen, with snow staying on the ground throughout the summer, itself resulting in a cooler summer climate. A runaway change in snow or sea ice could thus be an important amplifier or trigger for a major change in global temperature. It is possible that by slow changes over millennia or centuries, the climate could be brought to a break point involving a runaway change in snow and ice reflectivity over a few decades. These slow background changes might include variations in the earth's orbit (affecting summer sunlight intensity), or gradual changes in carbon dioxide concentration, or in the northern forest cover which affects the amount of snow that is exposed to sunlight.

It is possible that the relatively long-lived ice sheets might occasionally help bring about very rapid changes in climate, by rapidly 'surging' outwards into the sea and giving rise to large numbers of icebergs that would reflect back the sun's heat and raidly cool the climate. The intensely cold Heinrich events that punctuated the last ice age were initially thought to be caused by sudden slippage of the Laurentide ice sheet that covered most of Canada (ref.), sending out the grit-laden icebergs that are detected from North Atlantic sediments. However, it now seems that all the separate ice sheets around the north Atlantic surged outwards simultaneously, and that their outwards movement probably thus represents a secondary response to an initial climate cooling (e.g. a change in the deepwater formation system in the north Atlantic) rather than the initial trigger (ref.). This does not mean that ice surges and ice bergs were irrelevant in the extreme cold of Heinrich events; by their albedo effects they may have helped to intensify and temporarily stabilize a cooling event that would have occurred anyway. However, this may have occurred decades or centuries after the initial 'step function' event associated with the rapid cooling.

Another, possibly neglected, factor in rapid regional or global climate changes may be the changes in the albedo of the land surface that result from changes in vegetation or algal cover on desert and polar desert surfaces. An initial spreading of dark-coloured soil surface algae or lichens following a particularly warm or moist year might provide a 'kick' to the climate system by absorbing more sunlight and thus warming the climate, and also reducing the dust flux from the soil surface to the atmosphere (see below). Larger vascular plants and mosses might have the same effect on the timescale of years or decades. The recent detailed analysis of the ending of the Younger Dryas by Taylor et al. 1997, suggests that warming occurred around 20 years earlier in lower and mid latitudes, perhaps due to some initial change in vegetation or snow cover affecting land surface albedo. Some of the earlier climate warming events during the last 130,000 years show similar signs of changes in dust flux followed by changes in high-latitude temperature (Raymo pers. comm. 1997). (needs ref.s).

Water vapour as a feedback in sudden changes. Water vapour is a more important greenhouse gas than carbon dioxide, and as its atmospheric concentration can vary rapidly, it could have been a major trigger or amplifier in many sudden climate changes. For example, a change in sea ice extent or in carbon dioxide, would be expected to affect the flux of water vapour into the atmosphere from the oceans, possibly amplifying climate changes that would otherwise have occurred anyway. Large rapid changes in vegetation cover might also have added to these changes in water vapour flux to the atmosphere. In a recent (1997) lecture presentation the geologist W.S. Broecker has suggested that water vapour may act as a global 'messanger', co-ordinating rapid climate changes, many of which seem to have occurred all around the world fairly simultaneously, or in close succession. Broecker notes the evidence for large changes in the water vapour content of the atmosphere in terms of changes in the 18O content of tropical high Andean ice cores (Thompson et al. 1995), suggesting that the air's overall content of water vapour was about half of what it is at present..

Dust and particulates as a feedback in sudden changes. Particles of mineral dust, plus the aerosols formed from fires and from chemicals evaporating out of vegetation and the oceans, may also be a major feedback in co-ordinating and amplifying sudden large climate fluctuations. Ice cores from Greenland (Taylor et al. 1997), Antarctica and tropical mountain glaciers, and deep ocean cores, show greater concentrations of mineral dust during colder phases. This suggests that there was more dust around in the world's atmosphere during cold periods that during warm phases. It seems that the atmospheric content of dust and sulphate particles changed very rapidly, over just a few decades, during sudden climate transitions in the Greenland ice core record (Taylor et al. 1997). The drier and colder the world gets, the more desert there is and the higher the wind speeds, sending more desert dust into the atmosphere where it may reinforce the cold and dryness by forming stable 'inversion' layers that block sunlight and prevent rain-giving convective processes. A run of wet years in the monsoon belt could trigger rapid revegetation of desert surfaces by algae or vascular plants, and a sudden decrease in the amount of dust blown into the atmosphere. Less dust could help make conditions still warmer and wetter, pushing the climate system rapidly in particular direction (though dust and other particles might actually tend to warm the surface if they blow over lighter-coloured areas covered by snow or ice; Overpeck et al. 1997).

Seasonal sunlight intensity as a background to sudden changes. A major background factor in some, but not all, sudden climate switches seems to have been the set of 'Milankovitch' rhythms in seasonal sunlight distribution. Although this factor changes gradually over many thousands of years, it may take the earth's climate to a 'break point' at which other factors will begin to amplify change into a sudden transion.

In one of the Milankovitch rhythms, the shape of the earth's orbit shifts from more elliptical to more nearly circular. In another the degree of tilt of the earth's axis changes, and in the third the timing of the seasons changes relative to the earth's elliptical track nearer and further from the sun. These rhythms, respectively work on 21,000, 42,000 and 100,000-year timescales, alter the relative amount of solar radiation reaching the earth's Northern and Southern Hemispheres during summer and winter. Times when summer sunlight in the Northern Hemisphere is strong (but when the winter sunlight is correspondingly weak) tend to be the times when the rapid global transition from glacial to interglacial conditions occurs. This is thought to be due to the effects of summer temperatures on various of the factors mentioned above; for example, it ensures melting back of snow and sea ice in summer, helping the earth to absorb more solar radiation and thus to heat up further. It is probably the combination of amplifying factors that brings the earth out of glacial and into interglacial conditions. These big glacial-interglacial transitions tend roughly to follow the 100,000-year timescale, when the three different rhythms (and possibly other poorly understood factors such as the internal structure of ice-sheets) line up to give a big increase in northern summer warmth, but the lesser individual rhythms can also be detected in the temperature record on the 19,000 and 42,000-year timescales, and in fact the timing of interglacial onset tends to more closely follow multiples of the 19,000 year cycle than an exact correspondence to the 100,000 year cycle (Raymo, pers. comm.).

It is important to note, however, that most of the very rapid climate transitions during the last 100,000 years do not show any clear association in timing with the background Milankovitch rhythms. In these cases their ultimate trigger must lie in other factors, probably a combination of many processes that sometimes line up to set the climate system on a runaway course in either the direction of cooling or warming.

Could dramatic decade-timescale climate transitions occur in the near future?

From present understanding of the record of the last 130,000 years, at least a few large climate changes certainly occurred on the timescale of individual human lifetimes, the most well-studied and well-established of these being the ending of the Younger Dryas, and various Holocene climate shifts. Many other substantial shifts in climate took at most a few centuries, and they too may have occurred over a few decades, but the time resolution in the climate record is either not available, or has not yet been studied in enough detail. It will take time before the meticulous work of logging year-by-year changes in long ice cores and lake records is able to give a relatively complete picture of when, and exactly how quickly, rapid climate changes occurred. There are many 'suspected' decade-timescale climate changes from the past (just as the Younger Dryas was until recently a 'suspected' but distinctly unproven decadal-scale climate shift), but very few 'proven' ones. Greater knowledge of how frequently such sudden events have occurred, and under what general circumstances, before a greater understanding of them can be reached.

It is difficult to say what the risks are of a sudden switch in global or North Atlantic region climate might be, because the mechanisms behind all past climate changes (sudden or otherwise) are incompletely understood. Not even knowing how often decade-timescale changes occurred in the recent geological past, we are handicapped in trying to find mechanisms which might explain them and be used for forecasting future events. Even if one knew everything there was to know about past climate mechanisms, it is likely that we would still not be able to forecast such events confidently into the future. This is because the system will have been influenced by probabilistic events (due to the chaotic nature of the ocean-climate system, with runaway changes coming from miniscule differences in initial conditions), so it is not justifiable to talk in terms of what 'definitely' will or will not happen in the future, even though the public and policymakers are looking for certainities. All that one can reasonably do is set out what the current understanding is, acknowledging that this understanding is limited and may turn out to be wrong in certain key respects, and then talk in terms of probablities of particular events occurring.

At the outset, there is the possibility that most of the climate instability seen in the recent geological past is not relevant to our immediate future, because it represents a fundamentally different system; a 'glacial' state. Most of the rapid climate transitions during the last 130,000 years seem to have occurred against the background of a world with a larger northern ice sheet extent than at present, perhaps indicating that in this glacial mode the climate is predisposed to be more unstable than in our present interglacial state. Even the sudden and widespread early-to-mid Holocene arid event (8,200 y.a.) occurred at a time when large parts of the Laurentide ice sheet remained unmelted over Canada, and it may correlate with the draining of a large meltwater lake eastwards into the north Atlantic.

However, there were at least some rapid climate transitions which occurred when ice sheet extent was no greater than at present, such as the apparently widespread late Holocene cool/arid event around 3,800 y.a., and another cool event around 2,600 y.a. (although the time taken for onset of these later Holocene changes in regional and global climates does not yet seem to have been determined in the literature).

However, various large full-interglacial climate changes during the Holocene and certain earlier interglacials (e.g. the Eeemian and the Holstein Interglacials in Europe; Winograd et al. 1997)) that show up in the Greenland ice cap do seem to correlate with genuinely large climate shifts in Europe and elsewhere, taking conditions from temperate to boreal or even sub-arctic. Whether they occurred over decades, centuries or thousands of years, they offer a worrying analogue for what might happen if greenhouse gas emissions continue unchecked. Judging by its past behaviour under both glacial (e.g. the ending of the Younger Dryas) and interglacial conditions (e.g. the various Holocene climate oscillations leading up to the 20th century), climate has a tendency to remain quite stable for most of the time and then suddenly 'flip'; at least sometimes over just a few decades, due to the influence of the various triggering and feedback mechanisms discussed above. Such observations suggest that even without anthropogenic climate modification there is always an axe hanging over our head, in the form of random very large-scale changes in the natural climate system; a possibility that policy makers should perhaps bear in mind with contingency plans and international treaties designed to cope with sudden famines on a greater scale than any experienced in written history. By starting to disturb the system, humans may simply be increasing the likelihood of sudden events which could always occur anyway.

Another source of evidence seems to underline the potential importance of sudden climate changes in the coming centuries and millennia: computer modelling studies of the (still incompletely understood) north Atlantic deepwater formation system suggest that it is indeed sensitive to quite small changes in freshwater runoff from the adjacent continents, whether from river fluxes or meltwater from ice caps. Some scenarios in which atmospheric carbon dioxide levels are allowed to rise to several times higher than at present result in increased runoff from rivers entering the Arctic Basin, and a rapid weakening of the Gulf Stream, resulting in colder conditions (especially in winter) across much of Europe. One recently published study suggests that just doubling the amount of carbon dioxide in the atmosphere could be enough to set off such a change (ref.). Whilst these are only preliminary models, and thus subject to revision as more work is done, they do seem to point in the same direction as the ancient climate record in suggesting that sudden shutdowns or intensification of the Gulf Stream circulation might occur under full interglacial conditions, and be brought on by the disturbance caused by rising greenhouse gas levels. To paraphrase W.S. Broecker; 'Climate is an ill-tempered beast, and we are poking it with sticks'.


Adkins J.F., Boyle E.A., Keigwin L & Cortijo E. (1997). Variability of the North Atlantic thermohaline circulation during the last interglacial period. Nature v.390 p.154-156.

Behl, R. J., and Kennett, J. P., 1996, Brief interstadial events in the Santa Barbara basin, NE Pacific, during the past 60 kyr. Nature, v. 379, p. 243-246.

Bond G., Showers W., Cheseby M., Lotti R., Almasi P., deMenocal P., Priore P., Cullen H., Hajdas I. & Bonani G. (1997). A pervasive millenial-scale cycle in North Atlantic Holocene and Glacial climates. Science v.278 p.1257-1265.

Broecker W.S., "Will Our Ride into the Greenhouse Future be a Smooth One?" GSA Today 7(5):1-7 (May 1997).

Chappellaz J.M., Blunier T., Ratnaud D., Barnola J.M., Schwander J. & Stauffer B. (1993). Synchronous changes in atmospheric CH4 and Greenland climate between 40 and 8 kyr BP. Nature v.366 p.443-445.

Crowley T.J. & North G.R. (1991). Palaeoclimatology. Oxford University Press, Oxford.

Dansgaard et al. (1989). Ref.

Dansgaard, W. et al. (1993) Evidence for general instability of past climate from a 250-kyr ice-core record. Nature 364, 218-220

Heinrich H. (1988) Origin and consequences of cyclic ice rafting in the northeast Atlantic Ocean during the past 130,000 years. Quaternary Research, v.29, p.142-152.

Field et al. (1994) Ref.

Grootes et al. (1993), Nature

Kennett J.P. (1996). A review of polar climatic evolution during the Neogene, based on the marine sediment record. p.49-64. In; Vrba E., Denton G.H., Partridge T.C. & Burckle L.H. Palaeoclimate and Evolution, with emphasis on human origins. Yale University Press, New Haven.

O'Brien et al. (1994), Science 270, p.1962-1964

Field et al (1994), Nature Ref.

Overpeck et al. (1997) Ref.

Ramussen T.L., van Weering T.C.E. & Labeyrie L. (1997). Climatic instability, ice sheets and ocean dynamics at high northern latitudes during the last glacial period (58-10 KA BP). Quaternary Science Reviews v.16 p.73-80.

K. C. Taylor, P. A. Mayewski, R. B. Alley, E. J. Brook, A. J. Gow, P. M. Grootes, D. A. Meese, E. S. Saltzman, J. P. Severinghaus, M. S. Twickler, J. W. C. White, S. Whitlow, G. A. Zielinski (1997). The Holocene-Younger Dryas Transition Recorded at Summit, Greenland. Sciencev.278 p.825-827.

Taylor, K.C. et al (1993) The 'flickering switch' of late Pleistocene climate change. Nature 361, 432-436.

Thompson, L. G., and E. Mosley-Thompson. 1987. Evidence of abrupt climatic damage during the last 1,500 years recorded in ice cores from the tropical Quelccaya ice cap, Peru. Pp. 99-110 in Abrupt Climatic Change--Evidence and Implications, W. H. Berger and L. D. Labeyrie, eds.

Thompson, L. G., Mosley-Thompson, E., Davis, M. E., Lin, P.-N., Henderson, K. A. , Cole-Dai, J., Bolzan, J. F., and Liu, K.-B., 1995, Late glacial stage and Holocene tropical ice core records from Huascarán, Peru: Science, v. 269, p. 46-50.

Thompson, L. G., E. Mosley-Thompson, W. Dansgasrd, and P. M. Grootes. 1986. The "Little Ice Age" as recorded in the stratigraphy of the tropical Quelccaya ice cap. Science 234:361-364.

Thouvenay et al. (1994).

Winograd I.J., Landwehr J.M., Ludwig K.R., Coplan T.B. & Riggs A.C. (1997). Duration and structure of the past four interglaciations. Quaternary Research v.48 p.141-154.

Zhisheng A. & Porter S.C. (1997). Millenial-scale climatic oscillations during the last interglaciation in central China. Geology v.25 p.603-606.

References cited on the QEN pages, including those cited here

More on sudden jumps in climate, and the causes behind them

A detailed set of pages on ice core records of gases

You can contact me at;

Document last updated 19th December 1997.