CLIMATE / Abrupt Change 1 CLIMATE (WAIS), would one speak of an abrupt climate Abrupt Change change. Stefan Rahmstorf, Potsdam Institute for Climate Paleoclimatic Data Impact Research, Potsdam, Germany A wealth of paleoclimatic data has been recovered 0003 ^ Copyright 2001 Academic Press from ice cores, sediment cores, corals, tree rings, doi:10.1006/rwos.2001.0269 and other sources, and there have been signiRcant advances in analysis and dating techniques. These Introduction advances allow a description of the characteristics of past climatic changes, including many abrupt 0001 High-resolution paleoclimatic records from ice and ones, in terms of geographical patterns, timing and sediment cores and other sources have revealed affected climatic variables. For example, the ratio of a number of dramatic climatic changes that occur- oxygen isotopes in ice cores yields information red over surprisingly short times * a few decades or about the temperature in the cloud from which the in some cases a few years. In Greenland, for snow fell. Another way to determine temperature is example, temperature rose by 5}103C, snowfall to measure the isotopic composition of the nitrogen rates doubled, and wind-blown dust decreased by an gas trapped in the ice, and it is also possible to order of magnitude within 40 years at the end of the directly measure the temperature in the borehole last glacial period. In the Sahara, an abrupt with a thermometer. Each method has advantages transition occurred around 5500 years ago from and drawbacks in terms of time resolution and relia- a relatively green shrubland supporting signiRcant bility of the temperature calibration. Dust, carbon populations of animals and humans to the dry de- dioxide, and methane content of the prehistoric at- sert we know today. mosphere can also be determined from ice cores. 0002 One could deRne an abrupt climate change simply On long timescales, climatic variability through- 0004 as a large and rapid one * occurring faster than in out the past two million years at least has been a given time (say 30 years). The change from winter dominated by the Milankovich cycles in the earth's to summer, a very large change (in many places orbit around the sun * the cycles of precession, larger than the glacial}interglacial transition) occur- obliquity, and eccentricity with periods of roughly ring within six months, is not an abrupt change in 23000, 41000, and 100000 years, respectively. Since climate (or weather), it is rather a gradual transition the middle Pleistocene transition 1.2 million years following the solar forcing in its near-sinusoidal ago, the regular glaciations of our planet follow the path. The term abrupt implies not just rapidity but 100000-year eccentricity cycle; even though this has also reaching a breaking point, a threshold * it only a rather weak inSuence on the solar radiation implies a change that does not smoothly follow the reaching the earth, it modulates the much stronger forcing but is rapid in comparison to it. This phys- other two cycles. The prevalence of the 100000-year ical deRnition thus equates abrupt climate change cycle in climate is thus apparently a highly nonlinear with a strongly nonlinear response to the forcing. response to the forcing that is still poorly under- In this deRnition, the quaternary transitions from stood. The terminations of glaciations occur rather glacial to interglacial conditions and back, taking abruptly (Figure 1). Greenland ice cores show that a few hundred or thousand years, are a prime the transition from the last Ice Age to the warm example of abrupt climate change, as the underlying Holocene climate took about 1500 years, with cause, the earth's orbital variations (Milankovich much of the change occurring in only 40 years. cycles), have timescales of tens of thousands The ice ages were not just generally colder than 0005 of years. On the other hand, anthropogenic global the present climate but were also punctuated by warming occurring within a hundred years is not abrupt climatic transitions. The best evidence for as such an abrupt climate change as long as it these transitions, known as Dansgaard}Oeschger smoothly follows the increase in atmospheric (D/O) events, comes from the last ice age (Figure 2). carbon dioxide. Only if global warming triggered D/O events typically start with an abrupt warming a nonlinear response, like a rapid ocean circulation by around 53C within a few decades or less, fol- change or decay of the West Antarctic Ice Sheet lowed by gradual cooling over several hundred or RWOS 0269 BALA NP GAYATHRI JANE 2 CLIMATE / Abrupt Change thousand years. The cooling phase often ends with Eemian, is so far inconclusive. During the present an abrupt Rnal temperature drop back to cold (&sta- interglacial, the Holocene, climate was much more dial') conditions. Although Rrst seen in the Green- stable than during the last glacial. However, two land ice cores, the D/O events are not a local feature abrupt events stand out. One is the 8200-year event of Greenland climate. Figure 3 shows that subtropi- that shows up as a cold spike in arctic ice cores and cal sea surface temperatures in the Atlantic closely affected the North Atlantic region. The second ma- mirror the sequence of events in Greenland. Similar jor change is the abrupt desertiRcation of the Sahara records have been found near Santa Barbara, Cali- 5500 years ago. There is much evidence from cave fornia, in the Cariaco Basin off Venezuela, and off paintings, Rre remains, bones, ancient lake sedi- the coast of India. D/O climate change is centered ments, and the like that the Sahara was a partly on the North Atlantic and regions with strong atmo- swampy savannah before this time. The best evid- spheric response to changes in the North Atlantic, ence for the abrupt ending of this benign climate with little response in the Southern Ocean or Ant- comes from Atlantic sediments off north-eastern Af- arctica. The &waiting time' between successive D/O rica, which show a sudden and dramatic step-func- events is most often around 1500 years or, with tion increase in wind-blown dust, witnessing a dry- decreasing probability, 3000 or 4500 years. This ing of the adjacent continent. suggests the existence of an as yet unexplained 1500-year cycle that often (but not always) triggers aD/O event. The second major type of climatic Mechanisms of Abrupt Climate event in glacial times is the Heinrich (H) event. Change Heinrich events involve surging of the Laurentide Ice Sheet through Hudson Strait, occurring in the The increased spatial coverage, quality, and time 0008 cold stadial phase of some D/O cycles. They have resolution of paleoclimatic data as well as advances a variable spacing of several thousand years. The in computer modeling have led to a greater under- icebergs released to the North Atlantic during standing of the mechanisms of abrupt climate H events leave tell-tale drop-stones in the ocean change, although many aspects are still in dispute sediments when they melt. Sediment data further and not fully understood. suggest that H events shut down or at least drasti- The simplest concept for a mechanism causing 0009 cally reduce the formation of North Atlantic Deep abrupt climatic change is that of a threshold. Water (NADW). Records from the South Atlantic A gradual change in external forcing (e.g., the and parts of Antarctica show that the cold Heinrich change in insolation due to the Milankovich cycles) events in the North Atlantic were associated with or in an internal climatic parameter (e.g., the slow unusual warming there (a fact sometimes referred to build-up or melting of continental ice) continues as &bipolar see-saw'). until a speciRc threshold value is reached where 0006 At the end of the last glacial, a particularly inter- some qualitative change in climate is triggered. Vari- esting abrupt climatic change took place, the so- ous such critical thresholds are known to exist in called Younger Dryas event (12800}11500 years the climate system. Continental ice sheets may have ago). Conditions had already warmed to near-inter- a stability threshold where they start to surge; the glacial conditions and continental ice sheets were thermohaline ocean circulation has thresholds where retreating, when within decades the climate in the deep water formation shuts down or shifts location; North Atlantic region switched back to glacial con- methane hydrates in the seaSoor have a temperature ditions for more than a thousand years. It has been threshold where they change into the gas phase and speculated that the cooling resulted from a sudden bubble up into the atmosphere; and the atmosphere inSux of fresh water into the North Atlantic itself may have thresholds where large-scale circula- through St. Lawrence River, when an ice barrier tion regimes switch. holding back a huge meltwater lake on the North For the D/O events, Heinrich events, and the 0010 American continent broke. This could have shut Younger Dryas event discussed above, the paleocli- down the Atlantic thermohaline circulation (i.e., the matic data clearly point to a crucial role of Atlantic circulation driven by temperature and salinity differ- ocean circulation changes. Modeling and analytical ences), but evidence is controversial. Alternatively, studies of the Atlantic thermohaline circulation the Younger Dryas may simply have been the last (sometimes called the &conveyor belt') show that cold stadial period of the glacial following a tem- there are two positive feedback mechanisms leading porary D/O warming event. to threshold behavior. The Rrst, called advective 0007 Does abrupt climate change occur only during feedback, is caused by the large-scale northward glacial times? Evidence for the last interglacial, the transport of salt by the Atlantic currents, which in RWOS 0269 BALA NP GAYATHRI JANE CLIMATE / Abrupt Change 3 turn strengthens the circulation by increasing den- cools the northern hemisphere while warming the sity in the northern latitudes.