Perspective Abrupt climate change in the computer: Is it real? Thomas F. Stocker* and Olivier Marchal Climate and Environmental Physics, Physics Institute, University of Bern, Sidlerstrasse 5, CH-3012 Bern, Switzerland Models suggest that dramatic changes in the ocean circulation are responsible for abrupt climate changes during the last ice age and may possibly alter the relative climate stability of the last 10,000 years. mong the archives recording past cli- the air–sea fluxes of heat and freshwater. In freshwater balance of the North Atlantic. By Amate and environmental changes, ice the Pacific there is no such circulation, be- discharging freshwater, the thermohaline cores, marine and lacustrine sediments in cause the surface waters are too fresh: even circulation reduces or collapses completely, anoxic environments, and tree rings have if cooled to the freezing point, they would but the response depends on the location, seasonal to annual resolution. Changes in not acquire enough density to sink down and amplitude and duration of the perturbation dust level (1), snow accumulation (2), sum- establish a large-scale THC. This is caused (11, 12, 14, 15, 17, 18). All three responses mer temperature (3), and indicators of the by a combination of several factors, includ- shown in Fig. 1 can be generated in such productivity of marine life (4) suggest that ing reduced evaporation in the Pacific, models, albeit at different threshold values. some of the climate changes have evolved on fresher surface waters flowing northward, The Younger Dryas cold event (12,700– time scales as short as a few years to decades. and a different basin geometry (8). 11,550 years before present) is the most Such changes appear abrupt in comparison The Atlantic THC is a self-sustaining recent and best documented of the family of with the classical view of climate change. phenomenon and thus prone to instability. abrupt climate changes of the past. Large The understanding of the mechanisms re- If the northward flow of saline tropical amounts of freshwater entered the ocean sponsible for abrupt change has long relied waters decreases (e.g., by increased river before Younger Dryas (19), which may have on qualitative reasoning. However, climate runoff, ice melt, changes in precipitation, triggered this cold event. Models indicate research now increasingly makes use of nu- etc.), the density of high latitude waters is that a full stop (10) or a reduction of the merical models as a tool to interpret and reduced so that these waters are no longer water flux associated with the THC to Ϸ integrate results. Projections of anticipated able to sink: the circulation stops. The result 20% of the modern value (11) both pro- climate change also rely on such models. is a very cool and fresh northern North duce a drop in sea surface temperature in Ϸ Abrupt climate change, as recorded in many Atlantic. This process, first described by the the North Atlantic of 8°C and a substantial paleoclimatic archives (5), provides a chal- late Henry Stommel (9), is now well under- cooling over Greenland of 4–10°C. Al- lenge to these models. Their task is to stood and found by most climate models though the resolution of these coupled elucidate the processes that are involved in (10–12). The thermohaline circulation can ocean–atmosphere models is still coarse, abrupt change. If validated by the past, these thus exhibit more than one stable equilib- there is quantitative agreement with recent models may be able to quantify the likeli- rium, a feature that is typical of a non-linear paleoclimatic estimates of temperature hood of future surprises in the climate physical system. changes during that event (20). system. The concept of hysteresis illustrates the Climate models provide much additional possible responses of the ocean–atmosphere information if they also predict changes in Ocean Circulation and Abrupt Change. Ocean system to perturbations in the surface fresh- the global carbon cycle. Recently, results currents cause significant geographical dif- water balance. More freshwater in the sink- from a climate–carbon cycle model have ferences in the supply of heat to the atmo- ing regions reduces the THC (13–15) (Fig. been directly compared with paleoclimatic sphere (6). Regions around the North At- 1). The system is stable and reacts in a linear data of the Younger Dryas (21). In response lantic Ocean have a mean annual surface air fashion as long as threshold values are not to a meltwater pulse (Fig. 2a), the model temperature that is 5–7°C warmer than crossed (Fig. 1a). An abrupt change with an simulates a collapse of the THC in the those at the same latitude in the Pacific. This amplitude that no longer scales with the Atlantic and a cooling whose timing agrees is due to the thermohaline circulation perturbation occurs if threshold values are well with the Greenland ice core record (THC) of the Atlantic Ocean which is mov- crossed (Fig. 1b). If the initial state of the (Fig. 2b). The amplitude of the cooling is too ing warm, saline tropical waters northward; ocean–atmosphere system is a unique equi- small by a factor of 4 because of the zonal averaging of the model. Slight warming dur- the Gulf Stream is part of this basin-scale librium, the system jumps back to the orig- ing the northern cold event is simulated in circulation. These tropical waters then give inal state once the perturbation has ceased: the far south, evidencing the ‘‘bipolar see- off their heat to the atmosphere north of the abrupt change is reversible. However, if saw’’ (22, 23) between northern and south- Iceland, become denser and sink to the other equilibria exist, the perturbation ern hemispheres (Fig. 2c). The atmospheric abyss, and flow southward in the form of a causes an irreversible change (Fig. 1c). Hys- CO concentration increases during the deep western boundary current (7). The two teresis is well known in climate models, but 2 abrupt event, but the amplitudes do not major locations of deep water formation in its structure is highly model-dependent (16). exceed 30 parts per million by volume (Fig. the Atlantic are in the Greenland-Iceland- A burning question is, Where are we now on 2d); this corresponds well with the CO Norwegian (GIN) Sea and in the Labrador the hysteresis, and what is its structure? 2 changes measured on ice cores (21, 24). Sea, both of which contribute to the forma- tion of North Atlantic Deep Water. The Where Models Agree with the Climate Record. THC is driven by differences in the density Abrupt change is triggered in almost all *To whom reprint requests should be addressed. E-mail: of sea water and therefore is controlled by climate models by perturbing the surface [email protected]. 1362–1365 ͉ PNAS ͉ February 15, 2000 ͉ vol. 97 ͉ no. 4 Downloaded by guest on September 29, 2021 Fig. 1. The ocean–atmosphere system is a nonlinear physical sys- tem that can exhibit hysteresis behavior (13). The upper branch of the hysteresis is characterized by warm North Atlantic sea sur- face temperatures (SST), the lower branch by cold sea surface temperatures. A given perturba- tion (indicated by the blue and red arrows) in the freshwater bal- ance of the North Atlantic (pre- cipitation ϩ runoff Ϫ evapora- tion) causes transitions from an initial state 1 to states 2 and/or 3. Three structurally different responses are possible depending on whether threshold values (dashed line) are crossed: (a) Linear, reversible response. (b) Nonlinear, reversible response. (c) Nonlinear, irreversible response. The stable isotopes of dissolved inorganic because surface waters rich in 13C no longer surface to the abyss. Because the THC carbon in the ocean (␦13C) are further im- sink to depth. This is in agreement with contributes to this transport, a collapsed portant variables in climate models that can marine records that have sufficient resolu- THC would result in an accumulation of 14C be directly compared with the paleoclimatic tion to identify this event (25). By contrast, in the atmosphere. The paleoclimatic data record. In the deep North Atlantic, the ␦13C changes in the Southern Ocean are show the expected increase of 14Catthe climate–carbon cycle model simulates a de- minor, again in agreement with the few beginning of the Younger Dryas, but 14C crease in ␦13C when the THC is reduced available records (26). decreases already a few centuries later, long It should be noted that the simulated before the end of the event. This early changes to a given freshwater flux pertur- decrease could be attributable to a signifi- bation are strongly model-dependent, and cantly reduced production of radiocarbon in mixing parameters like vertical diffusivities the stratosphere or a substantial increase in in the ocean models appear to be most deep water formation in the Southern influential (16). Ocean or elsewhere. No climate model is able to simulate this 14C decrease at present. Where Models Disagree with the Climate Third, a full sequence of abrupt climate Record. Models still have substantial difficul- change such as seen in the Greenland ice ties in simulating important aspects of core record (29) has not yet been simulated abrupt change when confronted with the with climate models. Here the problems are hard evidence from the paleoclimatic ar- twofold: (i) the initialization of such a model chives. All current models respond instan- is an unresolved problem; and (ii) models of taneously to a large perturbation in the continental ice sheets, which will supply freshwater balance, such as that used in Fig. freshwater pulses, are only now being in- 2. Yet, the sea level records tell us that the corporated into comprehensive climate first meltwater peak during the deglaciation models.