Singularities that occur in complex systems with multiple thresholds can be assessed with appropriate models. In real systems, however, there always are stochastic elements that influence the behavior of these systems, which are difficult to model. The runaway greenhouse effect, for example, consists of a series of positive feedback loops that result from systemic interactions or can be triggered by stochastic events (Woodwell et al., 1998). Table 19-6 lists examples of such singularities that are triggered by different causes. All of these examples have regional or global consequences. The systemic insights in their behavior generally are based on different simulation approaches. Although local examples (e.g., species extinction) also are abundant in the scientific literature, they are ignored here because climate change does not (yet) seem to be the sole cause, and the processes involved generally are not modeled systematically.
Many model studies (reviewed in Weaver et al., 1993; Rahmstorf et al., 1996) have analyzed the nonlinear response of the worldwide ocean circulationthe so-called conveyor belt. This system transports heat and influences regional climate patterns. One component of this system is the current in the Atlantic Ocean. Warm surface currents flow northward. Heat release and evaporation from the ocean surface lowers the temperature and increases the density and salinity of the water. In the North Atlantic, this denser water sinks at the Labrador and Greenland convection sites and flows back south as deepwater. This so-called North Atlantic THC could slow down or even shut down under climate change (see TAR WGI Chapters 7 and 9).
|Table 19-6: Examples of different singular events
and their impacts.
|Nonlinear response of
|- Changes in thermal and freshwater forcing could result in complete
shutdown of North Atlantic THC or regional shutdown in the Labrador and Greenland Seas. In the Southern Ocean, formation of Antarctic
bottomwater could shut down. Such events are found in the paleoclimatic record, so they are plausible.
|- Consequences for marine ecosystems and fisheries could be severe.
Complete shutdown would lead to a stagnant deep ocean, with reducing deepwater
oxygen levels and carbon uptake, affecting marine ecosystems. Such a shutdown
also would represent a major change in heat budget and climate of northwestern
||WGI TAR Chapters 2.4, 7, and 9|
|Disintegration of West Antarctic
Ice Sheet (WAIS), with subsequent large sea-level rise
|- WAIS may be vulnerable to climate change because it is grounded below sea level. Its disintegration could raise global sea level by 4-6 m. Disintegration could be initiated irreversibly in the 21st century, although it may take much longer to complete.||- Considerable and historically rapid sea-level rise would widely exceed adaptive capacity of most coastal structures and ecosystems.||WGI TAR Chapters 7 and 11; Oppenheimer, 1998|
|Runaway carbon dynamics||- Climate change could reduce the
efficiency of current oceanic and biospheric carbon sinks. Under some conditions, the biosphere could even become a source.
- Gas hydrate reservoirs also may be destabilized, releasing large amounts of methane to the atmosphere.
- These processes would generate a positive feedback, accelerating
buildup of atmospheric GHG
|- Rapid, largely uncontrollable increases in atmospheric carbon concentrations and subsequent climate change would increase all impact levels and strongly limit adaptation possibilities.||WGI TAR Chapter 3; Smith and Shugart, 1993; Sarmiento et al., 1998; Woodwell et al., 1998; Bains et al., 1999; Joos et al., 1999; Katz et al., 1999; Norris and Rohl, 1999; Walker et al. 1999; White et al., 1999|
of continental monsoons
|- Increased GHGs could intensify Asian summer monsoon. Sulfate aerosols partially compensate this effect, although dampening is dependent on regional patterns of aerosol forcing. Some studies find intensification of the monsoon to be accompanied by increase in interseasonal precipitation variability.||- Major changes in intensity and spatial and temporal variability would have severe impacts on food production and flood and drought occurrences in Asia.||TAR WGI Sections 22.214.171.124 and 126.96.36.199; TAR WGII Section 11.5.1; Lal et al., 1995; Mudur, 1995; Meehl and Washington, 1996; Bhaskaran and Mitchell, 1998|
modification of crucial climate-system patterns such as ENSO, NAO, AAO, and AO
|- ENSO could shift toward a more El Niño-like mean state under
increased GHGs, with eastward shift of
precipitation. Also, ENSO's variability could increase.
- There is a growing attempt to investigate changes in other major atmospheric regimes [NAO, Arctic Oscillation (AO), and Antarctic Oscillation (AAO)]. Several studies show positive trend in NAO and AO indices with increasing GHGs.
|- Changing ENSO-related precipitation patterns could lead to changed
drought and flood patterns and changed
distribution of tropical cyclones.
- A positive NAO/AO phase is thought to be correlated with increased storminess over western Europe.
|TAR WGI Section 188.8.131.52; Corti et al., 1999; Fyfe et al., 1999; Shindell et al., 1999; Timmermann et al., 1999|
|Rearrangement of biome distribution as a result of
concentrations and climate change
|- Many studies show large redistribution of vegetation patterns.
Some simulate rapid dieback of tropical forests and other biomes; others
gradual shifts. More frequent fire could accelerate ecosystem changes.
|- All models initially simulate an increase in biospheric carbon
uptake, which levels out later. Only a few
models simulate carbon release.
|White et al., 1999; Cramer et al., 2000|
of international order by
environmental refugees and emergence of conflicts as a result of multiple climate change impacts
|Climate changealone or in
combination with other environmental pressuresmay exacerbate resource scarcities in developing countries. These effects are thought to be highly nonlinear, with potential to exceed
critical thresholds along each branch of the causal chain.
|- This could have severe social effects, which, in turn, may cause
several types of conflict, including scarcity disputes between countries,
clashes between ethnic groups, and civil strife and insurgency, each with
potentially serious repercussions for the security interests of the developed
||Homer-Dixon, 1991; Myers, 1993; Schellnhuber and Sprinz, 1995; Biermann et al., 1998; Homer-Dixon and Blitt, 1998|
The paleoclimatic record shows clear evidence of rapid climatic fluctuations in the North Atlantic region (with possible connections to other regions) during the last glaciation and in the early Holocene (see TAR WGI Section 2.4.3). At least some of these eventsnotably the Younger Dryas event, when postglacial warming was interrupted by a sudden return to colder conditions within a few decades about 11,000 years agoare thought to be caused by changes in the stability of North Atlantic waters. These changes, which are recorded in the central Greenland ice cores and elsewhere, were accompanied by large changes in pollen and other records of flora and fauna in western Europe, indicating that they had widespread effects on European regional climate and ecosystems (Ammann, 2000; Ammann et al., 2000). The likely cause for these fluctuations is changes in the stability of the THC brought about by an influx of freshwater from melting icebergs and/or ice caps (see TAR WGI Section 7.3.7). As discussed in WGI, enhanced greenhouse warming could produce similar changes in stability in the North Atlantic because of warming and freshening of North Atlantic surface waters.
The current operation of THC is self-sustaining within limits that are defined by specific thresholds. If these thresholds were exceeded, two responses would be possible: shutdown of a regional component of the system or complete shutdown of the THC. Both responses have been simulated. A complete shutdown was simulated by Manabe and Stouffer (1993) for a quadrupling of atmospheric CO2 and by Rahmstorf and Ganopolski (1999) for a transient peak in CO2 content. These studies suggest that the threat of such complete shutdown increases beyond a global mean annual warming of 4-5°C, but this is still speculative. It took several centuries until the circulation was shut down completely in both studies. A regional shutdown in the Labrador Sea (while the second major Atlantic convection site in the Greenland Sea continued to operate) was simulated by Wood et al. (1999). Simulated regional shutdown can occur early in the 21st century and can happen rapidlyin less than a decade. Simulations by Manabe and Stouffer (1993) and Hirst (1999) show further the possibility of a shutdown of the formation of Antarctic bottomwater, which is the second major deepwater source of the world ocean.
These simulations clearly identify possible instability for the THC. Determining appropriate threshold values, however, requires analysis of many scenarios with different forcings and sensitivity studies of important model parameters. Stocker and Schmittner (1997), for example, have shown that the THC is sensitive not only to the final level of atmospheric CO2 concentration but also to the rate of change. Rahmstorf and Ganopolski (1999) show that uncertainties in the hydrological cycle are a prime reason for uncertainty in forecasting, whether a threshold is crossed or not (see Figure 19-6). Further parameters are climate sensitivity (high values increase the likelihood of a circulation change) and the preindustrial rate of Atlantic overturning (an already weak circulation is more liable to break down) (e.g., Schneider and Thompson, 2000). These simulations suggest that global warming over the next 100 years could lead to a sudden breakdown of the THC decades to centuries later, which would lead irrevocably to major effects on future generations.
The possible impacts of these circulation changes have not yet been studied systematically. Complete shutdown of the THC would represent a major change in the heat budget of the northern hemisphere because this circulation currently warms northwestern Europe by 5-10°C (Manabe and Stouffer, 1988; Rahmstorf and Ganopolski, 1999). Consequently, shutdown would lead to sudden reversal of the warming trend in this region. The impacts of a regional shutdown would be much smaller but probably still serious. For the European climate, loss of the Greenland Sea branch probably would have a much stronger effect than loss of the Labrador Sea branch because the northward extent of the warm North Atlantic current depends mainly on the former. In either case, the consequences of circulation changes for marine ecosystems and fisheries could be severe (see Section 6.3). Shutdown of the major deepwater sources in the North Atlantic and Southern Ocean would lead to an almost stagnant deep ocean, with as-yet unexplored consequences (e.g., for deepwater oxygen levels, carbon uptake, and marine ecosystems).
Neither the probability and timing of a major ocean circulation change nor its impacts can be predicted with confidence yet, but such an event presents a plausible, non-negligible risk. The change would be largely irreversible on a time scale of centuries, the onset could be relatively sudden, and the damage potential could be very high.
The WAIS contains 3.8 million km3 of ice, which, if released to the ocean, would raise global sea level by 4-6 m. The WAIS has been the subject of attention since analysis of paleodata (Hughes, 1973) and ice sheet models (Weertman, 1974) predicted that such a marine-grounded ice sheet is inherently unstable.
Analysis of ice sediments indicates that in the past 1.3 million years, the WAIS has collapsed at least once (Scherer et al., 1998). It was inferred from marine sediments that the WAIS is still dynamic. Since the last glacial maximum, the grounding line (i.e., the boundary between the floating ice shelves and the grounded ice) has retreated considerably (Hughes, 1998), and this process continues. It probably reflects dynamics that were set in motion in the early Holocene (Conway et al., 1999). This has important implications because it points toward the long equilibration time scales involved in WAIS dynamics.
Fast-flowing ice streams, which feed the shelves from the interior, dominate the discharge of the WAIS (see TAR WGI Section 11.5). These ice flows are constrained at various boundaries. Whereas early studies emphasized the role of ice-shelf boundaries in such ice flow, more recent work points to the importance of different boundaries (i.e., the ice-stream bed, the lateral margins, and the inland endAnandakrishnan et al., 1998; Bell et al., 1998; Joughin et al., 1999; Payne, 1999). With respect to the time scales of an eventual WAIS disintegration, this distinction is crucial because the ice shelves respond to changes in climate within centuries, whereas the conditions at the ice-stream margins and beds have response times on the order of millennia (e.g., McAyeal, 1992). Whether proper incorporation of ice-stream dynamics into ice-sheet models generally eliminates the presumed instability cannot be conclusively resolved. McAyeal (1992), for example, incorporated ice-stream dynamics and deformable bed conditions explicitly into his ice-sheet model and showed that under periodic climate and sea-level forcing (100,000-year cycles), the WAIS collapsed and regrew sporadically throughout a period of 1 million years.
Even if accelerated loss of grounded ice were unlikely to occur over the 21st century, changes in ice dynamics could result in increased outflow of ice into the ice shelves and trigger a grounding-line retreat. An in-water temperature of a few degrees Celsius could cause the demise of the WAIS ice shelves in a few centuries and float its marine-based parts over a period of 1,000-2,000 years (Warner and Budd, 1998). This would produce a sea-level rise of 2-3 m. Huybrechts and de Wolde (1999) evaluate a climate change scenario that stabilizes GHG concentrations at four times the present value in 2150. They show that melting of the WAIS would contribute to 1-m sea-level rise by 2500a rate of rise that would be sustained at 2 mm yr-1 for centuries thereafter. The response of the Greenland ice sheet contributed to several meters of sea-level rise by 3000. Even under this stabilization scenario, melting of the Greenland ice sheet would be irreversible. Both studies, however, simply assume no change in ocean circulation and an immediate warming of water in the sub-ice-shelf cavity with a warming climate. Both assumptions still await full validation.
Global warming projected for the 21st century could set in motion an irreversible melting of the West Antarctic and Greenland ice sheets, implying sustained sea-level rise and irreversible losses. The impacts of complete disintegration of the WAIS and subsequent sea-level rise by 4-6 m, however, have not been fully explored. As summarized by Oppenheimer (1998), the disintegration time scales predicted by models vary widely, between 400-500 years (Thomas et al., 1979) and 1,600-2,400 years (McAyeal, 1992). These time scales correspond to a mean contribution to sea level of 10-15 and 2.5 mm yr-1, respectively. Whereas an estimate in the lower range is approximately equal to the present-day rate of sea-level rise, a value in the middle to high range lies outside human experience and would widely exceed the adaptive capacity of most coastal structures and ecosystems (see Sections 19.3 and 6.5).
Other reports in this collection