3
Processes That Cause Abrupt Climate Change

Weather changes abruptly from day to day, and there is no basic difficulty in understanding such changes because they involve a “fast” and easily observed part of the climate system (e.g., clouds and precipitation). But mechanisms behind abrupt climate change must surmount a fundamental hurdle in that they must alter the working of a “slow” (i.e., persistent) component of the climate system (e.g., ocean fluxes) but must do so rapidly. Two key components of the climate system are oceans and land ice. In addition, the atmospheric response is a crucial ingredient in the mix of mechanisms that might lead to abrupt climate change because the atmosphere knits together the behavior of the other components. The atmosphere potentially also gives rise to threshold behavior in the system, whereby gradual changes in forcing yield nearly discontinuous changes in response.

A mechanism that might lead to abrupt climate change would need to have the following characteristics:

  • A trigger or, alternatively, a chaotic perturbation, with either one causing a threshold crossing (something that initiates the event).

  • An amplifier and globalizer to intensify and spread the influence of small or local changes.

  • A source of persistence, allowing the altered climate state to last for up to centuries or millennia.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 73
Abrupt Climate Change: Inevitable Surprises 3 Processes That Cause Abrupt Climate Change Weather changes abruptly from day to day, and there is no basic difficulty in understanding such changes because they involve a “fast” and easily observed part of the climate system (e.g., clouds and precipitation). But mechanisms behind abrupt climate change must surmount a fundamental hurdle in that they must alter the working of a “slow” (i.e., persistent) component of the climate system (e.g., ocean fluxes) but must do so rapidly. Two key components of the climate system are oceans and land ice. In addition, the atmospheric response is a crucial ingredient in the mix of mechanisms that might lead to abrupt climate change because the atmosphere knits together the behavior of the other components. The atmosphere potentially also gives rise to threshold behavior in the system, whereby gradual changes in forcing yield nearly discontinuous changes in response. A mechanism that might lead to abrupt climate change would need to have the following characteristics: A trigger or, alternatively, a chaotic perturbation, with either one causing a threshold crossing (something that initiates the event). An amplifier and globalizer to intensify and spread the influence of small or local changes. A source of persistence, allowing the altered climate state to last for up to centuries or millennia.

OCR for page 73
Abrupt Climate Change: Inevitable Surprises As discussed in Chapter 2, the Younger Dryas is the most studied example of an abrupt change and provides insights about possible mechanisms. In many ways the Younger Dryas serves as a defining event that embodies the notion of abrupt climate change. Heinrich and Dansgaard/Oeschger events are generally thought to be governed by physical laws that are closely related to those involved in the Younger Dryas; indeed, many consider the Younger Dryas to be just an unusually big Heinrich or Dansgaard/Oeschger event. However, there are potentially many types of abrupt change, as described in Chapter 2, so we need to consider the full array of possible processes, rather than only those currently in favor for explaining the Younger Dryas and Dansgaard/Oeschger events. Other major events of interest in the paleoclimatic records include decadal Holocene droughts, the dry-moist cycles of North Africa, the Little Ice Age, and the cooling event that occurred about 8,200 years ago (the “8.2K event”). Abrupt shifts in the dominant modes of the modern climate are also being documented, and the active mechanisms in these shifts might be relevant to larger abrupt changes in the distant past or in the future. AN OVERVIEW OF MECHANISMS Understanding and simulating abrupt climate change poses special challenges in climate science. Many climate changes are well described as relatively small deviations from a reference state, often assumed to be in equilibrium with external forcing. Powerful simplified conceptual approaches (“linearizations”) are therefore appropriate and can explain many features in a faithful way. In a linear model, doubling the forcing doubles the response. The linear approach does not hold for abrupt climate change, in which a small forcing can cause a small change or a huge one, so fully nonlinear and transient considerations and simulations are required. In particular, transitions between qualitatively different climate states, as are seen in the paleoclimatic records, require abandoning the “time-slice” perspective in which equilibrium runs of climate models under different forcings are used to try to infer the path of climate change. Three past classical model types have been important in this context because of their ability to simulate paleo abrupt climate change either spontaneously or in response to changes in controlling parameters or external forcing. These are the two-box Stommel model (1961), simple energy-balance models such as that of Sellers (1969), and the Lorenz models (1963, 1990).

OCR for page 73
Abrupt Climate Change: Inevitable Surprises Using a two-box model of the thermohaline circulation (THC) of the ocean, Stommel (1961) showed that the different response times of the ocean surface to heat and freshwater perturbations give rise to multiple equilibria of the THC with very different characteristics. This implies that forcing and other conditions do not uniquely define the state of the climate system and that perturbations beyond thresholds can trigger transitions to other equilibrium states. The model presents some fundamental concepts and demonstrates the major sources of uncertainty in simulating abrupt climate change caused by a change in the THC. Multiple equilibria, however, would not occur if all processes were linear. The nonlinearity of the Stommel model arises because the flow field of ocean water transporting heat and salt is itself a function of temperature and salinity. Nonlinearities and multiple equilibria are the fundamental concepts behind the simulation of abrupt climate change. Simple models of the energy balance of the atmosphere exhibit multiple equilibria. The model of Sellers (1969) is a typical case: a given amount of solar irradiation allows either a cold or a warm planet in equilibrium. The nonlinearity here is introduced by a special formulation of the snow-albedo feedback, which is operative only in a particular range of temperatures. A cold earth is snowy, and snow reflects sunlight and keeps the earth cold. A warmer earth has less snow, absorbs more sunlight, and so stays warmer. Abrupt change can be triggered by variations in total solar output or other parameters that influence the radiative balance, such as snow cover. The Lorenz models (Lorenz, 1963, 1990) provide a highly simplified description of atmospheric circulation. In addition to multiple equilibria in particular ranges of parameters, these models exhibit self-sustained oscillations and chaotic behavior. A system oscillates near one of two preferred centers; abrupt change occurs when the system switches from one mean of oscillation to the other. In a Lorenz model, these transitions occur spontaneously. These examples suggest that abrupt climate change can occur in (at least) three fundamentally different ways: Abrupt climate change can be the response to a rapidly varying external parameter or forcing. If one views only the atmosphere-ocean system, massive sudden discharges of freshwater from disintegrating ice sheets on land would be an example of a sudden external influence. Nonlinearity

OCR for page 73
Abrupt Climate Change: Inevitable Surprises in the atmosphere-ocean system is not a prerequisite for such behavior, whose time scale is dictated essentially by that of the forcing.1 Slow changes in forcing can induce the crossing of a threshold and result in the transition to a second equilibrium of the system. The evolution of such a change would be governed by the system dynamics rather than by the external time scale of the slow change. In considering the whole earth system rather than just the oceans and atmosphere, massive discharges of freshwater from disintegrating ice sheets would be a result of threshold-crossing. Slow melting at the end of the last ice age produced ice-marginal lakes. When the ice margin reached a particular location, such as the path of a former river that the ice had dammed, a threshold was crossed, the ice dam broke, and the water was released rapidly (Broecker et al., 1988). Regime transitions can occur spontaneously in a chaotic system. In this case, external triggers for transitions are not required, so a series of regime changes could continue indefinitely or until slow changes in external forcing or system dynamics removed the chaotic behavior. Oceans Changes in ocean circulation, and especially THC in the North Atlantic, have been implicated in abrupt climate change of the past, such as the Younger Dryas and the Dansgaard/Oeschger and Heinrich/Bond oscillations (Broecker et al., 1988; Alley and Clark, 1999; Stocker, 2000). Today, relatively warm waters reach high latitudes only in the North Atlantic. The high salinity of the Atlantic waters allows them to sink into the deep ocean when they cool, and warmer waters flowing along the surface then replace them. This yields a net heat transport into the high northern latitudes of the Atlantic and northward heat transport throughout the South Atlantic, carrying heat into the North Atlantic (Ganachaud and Wunsch, 2000; see also Plate 4b.) Outburst floods, which would have freshened the North Atlantic and reduced the ability of its waters to sink, immediately preceded the coolings of the Younger Dryas and the short cold event about 8,200 years ago (Broecker et al., 1988; Barber et al., 1999); this suggests causation. Evidence of reduction or elimination of northern sinking of waters during cold times (Sarnthein et al., 1994; Boyle, 2000) provides further support, as does 1   For the whole earth system, a massive nuclear war or the impact of a gigantic meteorite might be considered in this category; such mega-disasters are not the focus of this report because the research agenda is not primarily concerned with the climate system.

OCR for page 73
Abrupt Climate Change: Inevitable Surprises the see-saw relation between Greenland and Antarctic temperatures on millennial scales (Blunier and Brook, 2001; see also Plate 2), which suggests that reduction in heat transport to the north allowed that heat to remain in the south. Those and other considerations focus attention on changes in the THC as one cause of abrupt climate change. However, additional processes presumably were active in the past abrupt changes exemplified by the Younger Dryas, as indicated by the difficulty of fully explaining the paleoclimatic data on the basis of the single mechanism of North Atlantic THC changes. Therefore, the ocean’s role in climate is developed more fully in the following. Water has enormous heat capacity—oceans typically store 10-100 times more heat than equivalent land surfaces over seasonal time scales, and the solar input to the ocean surface for a year would warm the upper kilometer only 1 degree—so the oceans exert a profound influence on climate through their ability to transport heat from one location to another and their ability to sequester heat away from the surface. The deep ocean is a worldwide repository of extremely cold water from the polar regions. If much of this water were brought to the surface in temperate or tropical regions, it could cause substantial cooling that, although transient, could last for centuries. It is not easy to bring cold water to the surface against a stable gradient, though, and this can happen only in special circumstances. Such localized change could, however, have a wider impact through atmospheric teleconnections. Fluctuations in ocean heat transport can also affect climate; for example, an increase in equator-to-pole heat transport would warm the polar regions (melting ice) and cool the tropics. The implications of fluctuation in heat transport by the Atlantic THC have received particular attention, especially as a mediator of Younger Dryas and Dansgaard/Oeschger abrupt change. Deep water forms only in the North Atlantic and around the periphery of Antarctica, where extremely cold, dense waters occur. There is no deep-water formation in the North Pacific, because the salinity is too low to allow high enough density to drive deep convection, despite the low temperatures. By analogy, change in the freshwater balance of the North Atlantic, which might be caused by glacial discharge or warming of the planet through increases in carbon dioxide, potentially can act as a trigger to turn the THC on or off. In contrast, it is not thought that future climate change could turn on deep-water formation in the North Pacific, although there is evidence that at times during ice ages

OCR for page 73
Abrupt Climate Change: Inevitable Surprises the intermediate waters of the Northern Hemisphere Pacific were more ventilated than they are now (Behl and Kennett, 1996). The threshold for deep-water formation cannot be thought of apart from the general circulation of the world ocean, because the density required for North Atlantic surface waters to sink is determined in relation to the “prevailing” deep-water density of the rest of the ocean. This density is determined globally and is intimately linked to processes in the Southern Ocean. Furthermore, the density of surface waters in the North Atlantic is not determined by purely local processes, in as much as the North Atlantic salinity is affected by mixing with transported subtropical Atlantic waters, whose salinity in turn is affected by tropical winds, which can systematically transport moisture out of the Atlantic basin. The freshwater balance of the Atlantic is further affected by melting of glaciers, transport of freshwater by sea ice, and land-surface processes that determine runoff patterns. Several ocean heat-transport mechanisms other than the THC are in operation today. In particular, wind-driven ocean circulation dominates ocean heat transport in the North Pacific (Bryden et al., 1991) and the Indian Ocean (Lee and Marotzke, 1997). Results from relatively simple models suggest that in the absence of a vigorous THC, wind-driven heat and salt transports increase, making up at least in part for the loss of the THC as a transport agent (Marotzke, 1990; Winton and Sarachik, 1993). Cryosphere Land glaciers and sea ice enter into abrupt change mechanisms in many ways. The accumulation of ice on land and the associated ice-albedo feedback are probably too slow to be involved in abrupt climate change. However, because a glacier that is frozen to its substrate can surge if the basal temperature of the ice is raised to the melting point, glacial discharge and decomposition can be rapid (MacAyeal, 1993a,b; Alley and MacAyeal, 1994). Ice-sheet surging certainly would affect sea level, as noted in Chapter 4 with regard to the West Antarctic ice sheet. Surging also may affect the atmospheric flow pattern by changing the elevation of parts of the continental ice sheets (Roe and Lindzen, 2001). Furthermore, rapid glacial discharge can release armadas of icebergs into the ocean, which serve as an important indicator of abrupt climate change; increases in ice-rafted debris are the defining feature of Heinrich events (Broecker, 1994). More importantly, glacial discharge abruptly increases the delivery of freshwater to the ocean (to the North Atlantic, in the case of Heinrich events), with the po-

OCR for page 73
Abrupt Climate Change: Inevitable Surprises tential of modifying the THC. Another class of catastrophic event associated with land glaciers is the formation of large lakes of meltwater, held back only by fragile ice dams. The breaking of an ice dam can lead to the sudden delivery of massive amounts of freshwater to the ocean. As noted above, it is believed that the draining of an ice-dammed lake (Agassiz) was at least partly involved in the initiation of the Younger Dryas event and was most likely responsible for the event about 8,200 years ago (Barber et al., 1999; Broecker et al., 1988; Teller, 1990; Teller, in press). Sea ice, which forms by freezing of ocean water, is an important amplifier of climate forcing. When sea ice forms, it increases the planetary albedo, enhancing cooling. Sea ice also insulates the atmosphere from the relatively warm ocean, allowing winter air temperatures to decline precipitously (as much as tens of degrees C, compared to conditions over open water), and reducing the supply of moisture to the atmosphere, which in turn reduces precipitation downwind. The rectifying and amplifying effects of sea ice are important in connection with changes in the location of North Atlantic deep-water formation. The site of North Atlantic deep-water formation is roughly where a substantial fraction of the heat transported northward in the Atlantic Ocean is deposited. Changes in the location affect the sea-ice margin and can have a net effect on the planetary radiation budget. Formation of sea ice also leads to rejection of very dense brine. This is particularly important around the Antarctic margin. Brine formation there is the major contributor to formation of the world’s deepest ocean water. Sea ice must be considered dynamically. Its movements are rather like those of a viscous fluid in coexistence with ocean water when viewed over a large enough area, but with brittle behavior in smaller regions. The creation of leads, or cracks, in sea ice affects albedo and air-sea exchange, and the movement of sea ice from one place to another has important effects on the global distribution of ice cover. Similar transport issues arise with regard to transport of icebergs discharged from land glaciers; here, the main interest concerns the distribution of freshwater ultimately delivered by the icebergs. Snow cover can also serve as an amplifier of climate change and a source of persistence. Snow-covered land maintains cold conditions because of its high reflectivity and because its surface temperature cannot rise above freezing until the snow melts. There are interesting interactions between snow cover and vegetation. A modest snow cover on a flat surface, such as tundra, suffices to cause high albedo. However, in terrain covered with ever-

OCR for page 73
Abrupt Climate Change: Inevitable Surprises green trees, snow falls to the surface without completely coating the dark canopy, allowing absorption of much solar radiation. Atmosphere The atmosphere is involved in virtually every physical process of potential importance to abrupt climate change. The atmosphere provides a means of rapidly propagating the influence of any climate forcing from one part of the globe to another. Atmospheric temperature, humidity, cloudiness, and wind fields determine the energy fluxes into the top of the ocean, and the wind fields dictate both the wind-driven ocean circulation and the upwelling pattern (of particular interest in the tropics and around Antarctica). Atmospheric moisture transport helps govern the freshwater balance, which plays a crucial role in the THC, and precipitation patterns provide the hydrological driving of glacial dynamics. The atmospheric response to tropical seasurface temperature patterns closes the feedback loop that makes El Niño operate. Atmospheric dust transport can affect the planet’s radiation balance and on longer time scales might affect ocean carbon dioxide uptake via iron fertilization (Mahowald et al., 1999). How do oceans affect the air-temperature pattern? The primary—although by no means the only—effect of oceans on air temperature derives simply from the heat capacity of the oceans and has little to do with horizontal ocean heat transport. Being composed of a fluid that can mix heat vertically, oceans are slow to cool in winter and slow to warm in summer. Land temperature, in contrast, responds rapidly to adjust to changes in the seasonal cycle of solar radiation. The primary winter pattern of surface air temperature thus consists of warm oceans and cold land (see Plate 5). To some extent, that contrast depends indirectly on ocean dynamics, which affect the upper ocean stratification and so the mixed-layer depth and the amount of seasonal heat storage (Plate 5). Complete shutdown of the THC would remove about 8 W/m2 from the Northern Hemisphere extratropical heat budget (Pierrehumbert, 2000). To restore balance, the northern atmosphere-ocean system must cool down until the infrared radiation lost to space is correspondingly reduced. On the basis of a conventional sensitivity factor incorporating water-vapor feedback, the perturbation in heat budget implies an extratropical cooling of about 4°C, and growth of sea ice could amplify this cooling. This is roughly the cooling found in simulations (Seager et al., 2001), in which ocean heat transport was suppressed. Because the Northern Hemisphere THC is pri-

OCR for page 73
Abrupt Climate Change: Inevitable Surprises marily an Atlantic rather than Pacific phenomenon, the heat transport is focused in the Atlantic. The somewhat greater warmth of European coastal regions than of similar latitudes on the Alaskan coast might well be linked to this THC heat transport (Plate 5), although debate continues about where and by how much ocean heat transport warms the atmosphere, and the extent to which changes in oceanic heat transport would be balanced by changes in atmospheric heat transport. Popular treatments sometimes imply that the THC is responsible not only for the temperature difference between Europe and Alaska, but also for the larger difference between Europe and the east coast of North America. This is unlikely because the US Pacific Northwest coast is warmer than the east coast of Asia without an active THC in the North Pacific. Two factors appear to provide the large differences between east and west coasts. First, the prevailing westerly winds pick up some heat during winter while blowing over the ocean, spreading the moderating influence of the ocean downwind. Second, and perhaps more important, the planetary wave pattern—the global-scale sinuous bending of the jet stream—brings Arctic air down over the US New England region and the Asian coast of the Pacific and warmer breezes to the US Pacific Northwest and Europe. The pattern is known to be driven primarily by mountains in the winter (Nigam et al., 1988), and is influenced relatively little by land-sea temperature contrasts. An improved quantitative understanding of the east-west (North America to Greenland to Europe) gradient of temperature fluctuations in past abrupt climate changes is crucial to furthering the understanding of mechanisms. On the other hand, the relative warmth of Norway, which is just to the east of the second major center of heat loss in the North Atlantic (the first being in the Gulf Stream off the coast of the United States), is probably due to the THC; the Pacific lacks similar warmth to that off Norway. A major problem with THC-based theories of the Younger Dryas and Dansgaard/Oeschger events is that atmospheric models yield a predominantly local North Atlantic cooling in response to THC shutdown, with few of the global repercussions that seem to be demanded by the data (see Chapter 2). In fact, the atmosphere alone does not seem able to extend the influence of extratropical influences to the rest of the globe efficiently. Even such a major forcing as the introduction of Northern Hemisphere ice sheets of the last glacial maximum into models has little influence south of the equator (Manabe and Broccoli, 1985; Broccoli and Manabe, 1987). In models generally, North Atlantic cooling affects the strength of the tropical

OCR for page 73
Abrupt Climate Change: Inevitable Surprises Hadley circulation, which leads to some temperature change in the northern half of the tropics, and more important precipitation changes. There have been a number of simulations of the atmospheric response to THC shutdown or to directly imposed North Atlantic sea-surface temperature perturbations. Fawcett et al. (1997) eliminated Nordic sea oceanic heat transport in the GENESIS model and found localized reductions in surface air temperature of as much as 24°C. Even with such an extreme cooling, the temperature perturbation is very localized. It amounts to only 2.8°C over the summit of Greenland versus about 8°C observed (Severinghaus et al., 1998) and 2°C or less over much of Europe, underestimating some changes based on proxy records. As summarized in Ágústsdóttir et al. (1999) and Fawcett et al. (1997), these experiments matched many aspects of changes reconstructed from proxy records (including changes in seasonality of precipitation in central Greenland and other such details) but generally underestimated the magnitude of reconstructed changes except close to the North Atlantic. It is difficult to determine how the oceanic heat transport involved in these experiments compares with the heat carried by the real THC. Manabe and Stouffer (1988, 2000) carried out experiments with a coupled atmosphere-ocean model in which the THC was largely suppressed by a massive artificial injection of freshwater. They found a more moderate Atlantic surface cooling of 6°C with cooling of 1-4°C over Europe and the greatest coolings confined primarily to coastal areas. The atmospheric-response experiments of Hostetler et al. (1999), forced by imposed sea-surface temperature patterns, yield a quite different impression, as the North Atlantic cooling leads to temperature reductions in excess of 4°C over all of Europe and much of Asia. The general locality of atmospheric response in those simulations does not by any means disprove the THC theory of the Younger Dryas and Dansgaard/Oeschger events. Current atmospheric models might be missing some crucial physical feedbacks that allow the real atmosphere to exhibit such large and widespread responses to THC changes. This is an unsettling possibility, in that it suggests that models could also fail to anticipate the threat of surprising and abrupt changes, which might occur in connection with global warming, as indicated in Plate 7 and discussed extensively in Chapter 4. Ocean dynamics could help to extend northern extratropical influences to the tropics and to the Southern Hemisphere, and indeed ocean circulation experiments show some global changes in response to North Atlantic freshwater pulses. A full treatment of the role of the oceans must be carried

OCR for page 73
Abrupt Climate Change: Inevitable Surprises out with coupled atmosphere-ocean models. There have not yet been any successful simulations of the pattern and magnitude of the Younger Dryas or of recurrent Dansgaard/Oeschger events with coupled ocean-atmosphere general circulation models. And there have not been any integrations of coupled ocean-atmosphere models over sufficiently long time scales to determine the natural level of millennial variability in such models, particularly in glacial conditions. Despite considerable success in modeling some aspects of past abrupt climate changes, there is much work still to be done. Fluctuations in the atmospheric carbon dioxide concentration (Petit et al., 1999) have played a crucial role in climate change on the glacial-inter-glacial time scale (Pollard and Thompson, 1997). According to general circulation model results (Manabe and Broccoli, 1985), without the reduced greenhouse effect arising from low carbon dioxide in glacial times, the Southern Hemisphere would have experienced little cooling despite the massive growth of Northern Hemisphere ice sheets. However, carbon dioxide fluctuations are not radiatively significant on the time scale of the Younger Dryas or Dansgaard/Oeschger events, and up to industrial times carbon dioxide had only insignificant fluctuations throughout the Holocene. Methane is an important greenhouse gas and did fluctuate markedly in the course of many abrupt-change events, with changes in low-latitude and high-latitude sources (Brook et al., 1999). The low-latitude source changes are generally thought to be indicative of changes in tropical hydrology, so methane serves as an important indicator of the involvement of the tropics in abrupt climate change (Chappellaz et al., 1997). However, the magnitudes of methane changes were too small to yield appreciable radiative forcings, and the observation that temperature changes in Greenland preceded methane changes (Severinghaus et al., 1998) is not generally consonant with driving by a methane greenhouse effect. Water vapor is an important greenhouse gas, but it is different from carbon dioxide and methane in that its concentration is determined primarily by atmospheric temperature and circulation patterns rather than by sources and sinks (Held and Soden, 2000; Pierrehumbert, 1999). Most of the atmosphere is highly undersaturated in water vapor and thus could hold considerably more than it does at present. The undersaturation is particularly prominent in the tropics, so reorganizations of the atmospheric water-vapor distribution stand as a possibility for amplifying and globalizing abrupt climate change. There are intriguing possibilities for interaction between water-vapor feedback and dust fluctuations, inasmuch as dust and other aerosols can serve as cloud condensation nuclei (Durkee et al., 2000).

OCR for page 73
Abrupt Climate Change: Inevitable Surprises Box 3.2 A Minimal Model for the Thermohaline Circulation For many years, the textbook example for explaining multiple states of the THC has been an idealized two-box model confined to a single hemisphere (Stommel, 1961), although a two-hemisphere version (Rooth, 1982) would be more appropriate to describe the Atlantic THC (Marotzke, 2000). Nevertheless, Stommel’s configuration (Figure 3.2) is useful to classify existing simulations of abrupt climate change involving the THC and to identify some crucial open questions concerning the future stability of the Atlantic’s THC. FIGURE 3.2 decrease as well but then responds with a sudden transition to the reverse mode. This is an abrupt climate change according to the definition used in this document; the response is considerably faster than the change in forcing. As for the real Atlantic and climate system, the crucial question is whether the THC is best characterized by the “lower” or by the “upper” case. Is there a possibility that a temporary perturbation in, say, the freshwater forcing can induce a permanent change in the THC? This would

OCR for page 73
Abrupt Climate Change: Inevitable Surprises Figure 3.2 shows the original set-up of Stommel’s two-box model for the THC. North-south exchange of water between a low-latitude box (left) and a high latitude box (right) is parameterized as a function of the density difference between the two boxes. H denotes the freshwater forcing, the flux of freshwater through the atmosphere from low latitudes to high latitudes or equivalently, as shown here, the flux of salt through the atmosphere from high latitudes to low latitudes. The circulation consists of a volume flux q taken to be proportional to the density difference between high and low latitudes. If q > 0, there is poleward surface flow because high-latitude density is greater than low-latitude density, and vice versa. At low latitudes, the ocean gains heat from and loses freshwater to the atmosphere; the opposite is true at high latitudes. Consequently, both temperature and salinity are higher at low latitudes than at high latitudes. This has opposite effects on density. When q > 0, the temperature difference dominates the density difference and drives the circulation, whereas the salinity difference brakes it, and vice versa. The situation q > 0, with sinking implied at high latitudes, is familiar from the North Atlantic and describes today’s active THC. In a plausible limiting case (Marotzke, 1990), the box temperatures are assumed to be imposed by the atmosphere, as is the surface freshwater exchange. A new addition to the system is horizontal diffusion. It reflects the transport of salinity by the ocean gyres, the systems of ocean circulation characterized by swift currents near the western boundary and slower return flows farther eastward, occurring at virtually the same depth. Understanding the effect of ocean gyres on the stability of the THC is crucial, and the models currently used are likely to distort this influence because of their low spatial resolution. Two cases are considered here. The “standard” case is the classical THC box model with only very weak diffusion; the other case will be designated “diffusive.” In the example shown, the diffusive case has a different proportionality factor relating density differences to flow strength, such that with the same freshwater forcing, the two cases have very similar strengths of the “normal” North Atlantic THC. imply that the real THC displays hysteresis and the possibility of abrupt change. Or would any transition be smooth, on the time scale of the forcing change, and be reversible? Simulations with comprehensive three-dimensional GCMs show both types of behavior. An ocean model with relatively large diffusion, caused by the choice of the numerical scheme, exhibits a transient behavior very similar to the upper dashed curve in Figure 3.4 when a slow freshwater flux perturbation is applied to the North Atlantic (Mikolajewicz and Maier-Reimer, 1994). A less-diffusive coupled model

OCR for page 73
Abrupt Climate Change: Inevitable Surprises Box 3.3 Multiple Equilibria and Model Parameters FIGURE 3.3 Steady-state solutions of the two-box model as a function of the atmosphere-ocean freshwater flux H. Depending on the strength of the horizontal mixing (diffusion) the model exhibits multiple eqilibria for a limited range of freshwater fluxes (bottom curve) or, alternatively, a unique solution (top curve). (See plate 6 for more detail.) exhibits a permanent change in the THC (Manabe and Stouffer, 1988). This aspect has been associated with the quantitatively different vertical mixing in these models (Manabe and Stouffer, 1999). The simple box model above demonstrates that the shape and position of the hysteresis in parameter space depend strongly on the model setup and on values of model parameters. Uncertainties in the parameters translate directly into uncertainties in the hysteresis and therefore uncertainties in the likely response of a system to a perturbation. Investigations using simplified models indicate that mixing schemes and values of mixing parameters are crucial in determining the shape of the hysteresis (Knutti and Stocker, 2000; Schmittner and Weaver, 2001). Unfortunately, mixing in the ocean is not well simulated in current-generation climate models, because of poorly parameterized small-scale processes and insufficient resolution. The location of the present, past, and future states of the climate system on the hysteresis curve, and its shape

OCR for page 73
Abrupt Climate Change: Inevitable Surprises Figure 3.3 (also see Plate 6) shows that the steady states of the two-box model can be calculated analytically and the dependence of the flux q (a measure of the strength of the THC) on the freshwater flux H can be examined. In the standard case (weak diffusion, lower curve), the THC is strongest (value arbitrarily set to 1) when the freshwater flux H vanishes. With higher H, the THC weakens. If H = 0.3, no steady-state solution with q > 0 is possible. However, for H > 0.1, there is a second stable equilibrium, a reverse mode of the THC, with q < 0. This flow pattern strengthens as H increases. For a certain parameter range, here 0.1 < H < 0.3, three equilibria are possible; it is readily shown that the middle one (on the dotted part of the curve) is not a stable solution. In that range of H, the model exhibits hysteresis. The presence of hysteresis is strongly dependent on model parameters. This is shown for a case in which the effect of the horizontal mixing due to gyre transports is increased (strong diffusion). Hysteresis disappears (upper curve); and for progressively increasing freshwater forcing, the THC smoothly approaches zero and—again smoothly—turns into the reverse mode for H = 0.56. For any given freshwater forcing, there is a unique and stable THC. The presence of multiple equilibria of the THC therefore depends strongly on the model formulation, parameterization of processes, and choice of parameter values. and complexity, clearly are among the major unsolved problems in climate dynamics. Results that depend crucially on a specific shape of the hysteresis are likely not to be robust findings at this stage. Whether the nonlinearities giving rise to abrupt change in the simplified models are an artifact of the simplifications or carry over to more complex and realistic systems needs to be investigated. Only more comprehensive and more complete climate models can make a convincing case that the assumptions underlying the nonlinearities in these simple models bear sufficient realism. Simulation of Past Changes of the Thermohaline Circulation Earlier, three fundamental ways of causing abrupt climate change were presented; all have been simulated with coupled GCMs. The first category

OCR for page 73
Abrupt Climate Change: Inevitable Surprises FIGURE 3.4 Response to the thermohaline circulation (THC) to a slow increase and subsequent slow decrease in freshwater forcing for the two cases in Figure 3.3. concerns the response to a sudden external perturbation, such as a disintegrating ice sheet. Manabe and Stouffer (1995) released a short, strong pulse of freshwater (1 Sv = 106 m3/s for 10 years) into the northern North Atlantic in their coupled model. The THC responded instantaneously with a reduction of almost 70 percent within a decade but recovered to the original strength within less than 200 years. Although the THC shut down almost completely, a threshold was not crossed, and no transition to a new equilibrium was simulated. The model does, however, have a second equilibrium that is achieved if the perturbation is much stronger (Manabe and Stouffer, 1999). It appears, therefore, that the behavior of the model is qualitatively similar to the standard case of the Stommel two-box model (see Plate 6, orange line). Depending on the amplitude of the perturbation, the coupled GCM thus exhibits either a transition to a new stable state or a reversible change (which might still be abrupt and of large amplitude). Manabe and Stouffer (1999) also showed that simply increasing the vertical diffusivity in their model led to a situation with a single equilibrium; this implies that processes other than those discussed above could qualitatively change the hysteresis structure of the model. Behavior similar to that in Manabe and Stouffer’s (1995) model was observed by Schiller et al. (1997), albeit because of a different forcing. They

OCR for page 73
Abrupt Climate Change: Inevitable Surprises forced their model with a freshwater flux qualitatively similar to that in Figure 3.4. This coupled model appears to have a unique equilibrium, and the simulated changes evolve roughly on the time scale of the forcing. Any abruptness in such a model will therefore be generated only through the forcing and not through the dynamics of the model itself. The transient behavior of this complex model is qualitatively comparable with that of the diffusive Stommel model (Figure 3.4, upper dashed curve). The complete, temporary shutdown of the THC causes a massive cooling in the North Atlantic and weak warming in the South Atlantic. The climatic behavior of opposite sign during a full THC shutdown has been termed the bipolar seesaw (Broecker, 1998; Stocker, 1998), and is a phenomenon that was probably active during most of the last glacial period (Blunier and Brook, 2001) (see Plate 2). The second type of abrupt change in climate models occurs when slow changes in the forcing push the model beyond a threshold and induce a transition to a second equilibrium. This behavior is simulated in ocean-only models that include simplified formulations of the atmosphere (Rahmstorf, 1994, 1995) and in models of reduced complexity (Aeberhardt et al., 2000; Schmittner and Weaver, 2001). Periodic forcing can also trigger abrupt change in a model that has multiple equilibria, provided that the forcing amplitude covers both branches of the hysteresis. Ganopolski and Rahmstorf (2001) showed that weak forcing is sufficient if the model’s hysteresis has a narrow shape. As emphasized earlier, such results are not considered robust, because they depend strongly on the structure of the hysteresis and the location of the initial state on it. Tuning and choices of parameters can strongly influence the transient behavior under a given forcing (Schmittner and Weaver, 2001) (Figure 3.4). The third possibility for the occurrence of abrupt climate change in model simulations is a spontaneous regime transition similar to the paradigm of the Lorenz (1963) model. This concept has been explored in simplified models (Timmermann and Lohmann, 2000) but was recently also found in a control run of a coupled atmosphere-ocean GCM (Hall and Stouffer, 2001). Annual mean surface air temperature, averaged over a region extending from southwestern Greenland to Iceland, showed natural interannual variability of around ±1°C (Figure 3.5, top). During the 15,000-year integration, one event occurred in which the temperature underwent a rapid decrease of about 4°C within a few decades (Figure 3.5, bottom). The event was triggered by an atmospheric pattern of persistent northwesterly flow in the region. This induced a southwestward Ekman flow in the surface ocean,

OCR for page 73
Abrupt Climate Change: Inevitable Surprises FIGURE 3.5 Model of surface air temperature averaged over an area spanning from southwestern Greenland to Iceland found in a 15,000-year control integration of a coupled atmosphere-ocean Global Circulation Model. The top shows the entire series, while the bottom highlights a 500-year period that contains an abrupt cooling event. (Hall and Stouffer, 2001). enhanced the East Greenland Current, and brought freshwater into the northern North Atlantic. Only then did the THC respond with a weakening that amplified the initial perturbation. In this example, changes in the THC were a response to rather than a cause of the cooling event. Although the result is intriguing, its robustness is poorly understood; the lack of intermediate-size events during the long integration is a cause for concern. Some Open Questions Even if one accepts that the THC played a central role in past abrupt climate change, the above examples of recent studies underline how our understanding of past rapid changes is still limited. In particular, the question of what triggered abrupt climate change, such as the Younger Dryas, has not yet been answered, unless we assume that the climate system essentially exhibits spontaneous regime transitions of the THC. In the following, we attempt to break down that overarching question into smaller, more manageable topics. The problem contains some fundamental aspects, as epitomized by the two different hysteresis structures in Plate 6, indicating that the whole spectrum of approaches—theory, modeling, and observations—will be needed.

OCR for page 73
Abrupt Climate Change: Inevitable Surprises The conceptual discussion above has been framed entirely in terms of Stommel’s hemispheric model of the THC, but the likely importance of the bipolar see-saw alone indicates that this view must be broadened. Two seemingly unrelated questions help to focus the discussion: First, what is the role of convective mixing in the THC? Second, if salinity is so important in determining the global THC pattern, why does the North Atlantic gain its high-latitude surface density mainly through heat loss (Schmitt et al., 1989)? The first question is motivated by the strong observational (Munk, 1966; Munk and Wunsch, 1998) and numerical (e.g., Bryan, 1987; Marotzke, 1997) evidence that the strength of the THC is strongly controlled by the vigor of vertical mixing in the stratified regions of the ocean. This is because the deepwater that upwells in these regions must be heated by mixing to maintain the near-surface higher temperatures. In contrast, the THC strength is quite insensitive to the efficiency of convective mixing (that is, the speed with which the ocean eliminates dense water overlying lighter water), and the THC could even be strong in the absence of convective mixing, according to the single-hemisphere GCM of Marotzke and Stott (1999). The answer to the first question leads the way to answering the second. Marotzke (2000) has argued that the results from single-hemisphere models should be viewed as applying to the global integral of the various THC branches and that the magnitude of the global integral obeys different laws from the distribution of the grand total over various competing deepwater formation sites. A number of idealized and more realistic ocean GCMs have shown that varying the freshwater flux forcing leaves the global integral of deep sinking nearly constant but the strength of North Atlantic sinking considerably changed (Tziperman, 1997; Klinger and Marotzke, 1999; Wang et al., 1999). Hence, a reduction in Northern Hemisphere THC would be associated with an increase in Southern Hemisphere THC. Thus, the globally integrated THC (sum of all branches) is rate-limited by vertical mixing and the gross pole-equator density contrast, which is dominated by the pole-equator temperature contrast (not by salinity). It follows that convection is basically driven by heat loss and that it occurs predominantly at high latitudes. That so much oceanic deepwater is formed in the North Atlantic, and not in any of the other competing high latitude regions, is determined by the North Atlantic’s high surface salinity and hence high surface density. These, in turn, are strongly influenced by the freshwater flux forcing; moreover,

OCR for page 73
Abrupt Climate Change: Inevitable Surprises the deepwater formation is intimately related to the very deep convective mixing occurring there. This two-step procedure—considering first the global integral of THC strength and then its distribution over different regions—helps to sort out a number of conceptual questions, but others arise. Most important is the role of the Antarctic Circumpolar Current, especially the wind-induced upwelling there (Toggweiler and Samuels, 1995); this has only recently begun to be addressed theoretically (Gnanadesikan, 1999). The interaction of the North Atlantic THC with the other oceans is poorly understood, both conceptually and from observations (e.g., Whitworth et al., 1999). Extending one’s perspective beyond the Stommel model also calls into question the long-held tenet that freshwater forcing necessarily weakens the THC. Rooth’s (1982) interhemispheric box model suggests that the Atlantic THC actually increases with increased freshwater flux (Rahmstorf, 1996; Scott et al., 1999; Marotzke, 2000). This is confirmed for the Atlantic branch of the THC in an idealized global GCM, as long as one considers the equilibrium response (Wang et al., 1999). Under a faster increase, as would be expected with increased greenhouse gases or might have occurred with outburst floods or ice-sheet surges in the past, the THC did weaken. If the equilibrium response can be interpreted as reflecting the THC’s response to very slowly varying atmospheric moisture flux, as might plausibly have happened during the glaciations, the assumed overall glacial weakening of the THC, as opposed to the shorter-lived and stronger proposed weakening of the THC in specific events during the glacial, could be explained. Given confirmation that the North Atlantic surface densities and the resulting convective activity do matter, the question arises whether we understand, and can observe, what they are influenced by. The drivers could be oceanic transports of freshwater (e.g., Aagard and Carmack, 1989) and heat, local surface fluxes, or remote influences, such as the water-vapor transport from the Atlantic to the Pacific (Zaucker and Broecker, 1992; Schmittner et al., 2000; Latif et al., 2000). In addition, one should consider the evolution of surface density in the Southern Ocean, in particular the Weddell Sea, because it is not well observed and the processes that link it with North Atlantic densities are not well understood. There is a huge gap in our conceptual understanding linking changes in convective activity, in the North Atlantic or elsewhere, to the THC and the northward heat transport. Regionally, changes in properties appear to occur rapidly but are poorly understood (e.g., Sy et al., 1997). On a larger

OCR for page 73
Abrupt Climate Change: Inevitable Surprises scale, the THC (and hence heat transport) can respond to changes in external forcing much faster than on the millennial time scale of thermodynamic equilibration of the deep ocean, but it is not clear which of the various plausible, identifiable time scales—ranging from months to decades—is most relevant. The classical picture (Kawase, 1987; Döscher et al., 1994) suggests that the deep circulation is set up through wave processes on time scales of months to a few years, but it has also been argued that advection by the deep western boundary current must be important, and this implies time scales of decades (Marotzke and Klinger, 2000). The potential predictability and prediction of the THC raise the thorny issue of assessing the quality of numerical simulations of future climate evolution. The fundamental problem is best understood when juxtaposed with daily weather forecasting. Through thousands of forecasts based on model simulations, it has been established that weather predictions have “skill”; that is, they improve upon a naïve baseline. But the time needed to test a weather forecast typically is a day—we will know tomorrow night whether tomorrow’s picnic gets rained out. An analogous accumulation of evidence of forecast skill is impossible when the prediction lead time is decades and more. How, then, can we state with well-defined confidence what is likely in store? An abrupt THC change is likely to have consequences massive enough that its probability should be estimated, if only crudely. In lieu of establishing true forecast skill, the models used for future climate-change scenarios should pass all the tests posed by available data. For example, one would reasonably have more confidence in climate models of future evolution that were able also to simulate the complicated climate history, including the rapid changes, of the last glacial interval. However, it is difficult to establish the “transferability” of putative modeling success of the past. Although being able to simulate the Younger Dryas is neither strictly necessary nor sufficient for being able to predict the THC’s evolution under global warming, it would enhance confidence considerably. Another strategy would be to simulate an ensemble of possible future climates and ascribe probabilities to them; an example is discussed in Chapter 4. Still, the problem remains of assessing whether the probabilities have the right order of magnitude. A further limit to predictability might arise from chaotic dynamics. In the global model of Wang et al. (1999), the timing and presumably the critical threshold of a THC collapse were fundamentally unpredictable to within a factor of 2. Different realizations of statistically identical random perturbations in the wind field in Wang et al. (1999), mimicking different (unpre-

OCR for page 73
Abrupt Climate Change: Inevitable Surprises dictable) weather histories, led to collapse times of that varied between 250 and 500 years. These results have not been confirmed (or challenged) with more sophisticated models, and the random perturbations might have been unrealistically large, but the possibly chaotic nature of THC changes needs to be addressed. Considering the known limitations of climate models, it is not currently possible to ascribe probabilities to future abrupt climate changes. Given the possible large impacts, simulation of abrupt climate change caused by a collapse in the THC constitutes a “Grand Challenge” problem in computational earth science. A dedicated supercomputer could be used to test whether the THC shows the possibility of threshold behavior, abrupt change, and hysteresis in a climate model that in its ocean components resolves the most important features of the ocean circulation. The conceptual models discussed here demonstrate that the outcome is open: on the one hand, high-resolution models are expected to have stronger gyre transport of heat and salinity, which appears to work in favor of the diffusive paradigm; on the other hand, a high-resolution model would be expected to exhibit a greater degree of internal variability, increasing the chance of self-induced threshold-crossing—if indeed a threshold exists. The need for substantially increased computational resources is clear.