|Working Group II: Impacts, Adaptation and Vulnerability|
|Other reports in this collection|
1.4.2. Detection and Attribution of Impacts to Climate Change
Many observed changes in ecosystems, animal (e.g., butterfly and bird patterns) and plant (e.g., timing of flowering) species, and physical systems (e.g., glaciers or river runoff) have been associated with observed changes in climate (not necessarily anthropogenic changes) in recent decades (high confidence). Moreover, as described in Chapters 5 and 19, such observed changes often are in the directions expected as a response to climate stimuli, based on understandings expressed in the literature about biophysical processes that govern responses to climate (e.g., Root and Schneider, 2001; Root et al., 2001). This consistency has led the authors of such studies to conclude that surface temperature trends of recent decades are likely to be discernible at regional scales through observed changes in biological and physical entities for several systems (varying confidence, depending on which specific system is considered; see, e.g., Chapters 5 and 19). From these observed responses to the relatively small climate changes observed to date (as compared to changes projected for the next century), it is concluded that many environmental systems can be highly sensitive to climate change. However, determination of a potential causal relationship between the response of a physical or biological system to observed recent climatic changes does not imply that regional climate changes were a direct result of anthropogenic global climatic trends, although the latter are likely to have had significant influence on many regional trends. Working Group II does not focus on evaluating the likelihood that regional observed climatic variations are caused by anthropogenic climate changes; detection and attribution assessment of climatic changes is primarily a Working Group I activity. As noted, however, Working Group II does address attribution of observed changes to environmental systems to observed climate changes, even if the connection to possible anthropogenic climate changes is not specifically addressed here.1.4.3. Key Determinants of Impacts
Early studies concentrated on impacts caused by changes in global mean temperature. Often these studies were carried out at a few elevated temperatures—typically, 2°C and 4°C (corresponding to the bulk of the range of IPCC Working Group I SAR equilibrium temperature rise expected for a doubling of CO2 concentration above pre-industrial levels). Global mean temperature still is a significant variable, serving as a modulus of change against which to compare climate sensitivities and impacts. In addition to mean quantities, however, other characteristics of climate measures, such as climate variability or runs of unusually warm weather, have become important variables for analysis (e.g., Mearns et al., 1984; Colombo et al., 1999), as has specification of changes in regional temperatures, sea level, and precipitation. These expanded measures of climatic change are routinely included in recent impact studies (e.g., IPCC, 1998). Less often considered are changes in extreme events (but see Table SPM-1, WGII TAR Summary for Policymakers), despite their potential importance.
It is essentially undisputed that a sustained 2°C temperature change occurring in a decade would have a more profound impact than one occurring over a century. The effect of rates of change on impacts is still under active investigation (see Chapter 19). Early results have suggested that rates of change exceeding the ability of ecosystems to migrate would be particularly damaging (see Chapter 5). Adaptation of coastal dwellers to rapid climatic changes or a high background “noise level” of natural variability would be more difficult relative to slowly occurring changes or smoothly varying climates (e.g., West et al., 2001). Finally, as noted by IPCC Working Group I (1996a, p. 7), “nonlinear systems, when rapidly forced, are particularly subject to unexpected behavior.” In other words, the adaptability of various decision agents would be reduced if any change is unexpected; thus, a rapid rate of change is more likely to generate “surprises” that inhibit effective adaptation by natural and managed systems. Table 1-1 describes several extreme events that can substantially influence the vulnerability of sectors or regions to climatic changes (see also Table SPM-1, WGII TAR Summary for Policymakers).
Economists also have suggested that the transient stage of moving from one equilibrium climate to another could cause the greatest economic impacts, even if the static impacts of the new equilibrium climate were small (Nordhaus and Boyer, 2000).
Climate sensitivity—the globally averaged response of the surface temperature to a fixed doubling of CO2—is based on static or equilibrium calculations in which the climatic model is allowed to reach a steady state after the CO2 increase is applied. The real Earth, on the other hand, is being forced by a time-evolving forcing of GHGs and other global change forcings; this, combined with the time-evolving response of the climate system to any forcing, means that the amount of global climatic warming, as well as the time-evolving patterns of climatic changes, are likely to be different during the transient phase of climatic change than in equilibrium. Recent studies of climate change impacts have made use of transient or time-dependent scenarios of climate change that are derived from fully coupled, ocean-atmosphere general circulation models (AOGCMs). These studies indicate that many systems would be notably affected (see Chapter 19)—some adversely and some beneficially—by changes in climate within the next 2 to 3 decades (high confidence). Farther into the 21st century, as radiative forcing on the climate builds, the magnitude of adverse impacts would increase, the number and scale of many beneficial effects would decrease (Chapter 19), and the probability that adverse impacts would predominate would increase (high confidence). Transient scenarios are just entering the climate impacts literature, which unfortunately tends to lag the climate effects literature by several years; thus, much of the impacts literature still is based on equilibrium climate change scenarios. To the extent possible, Working Group II has assessed literature that uses transient scenarios. It is important to use transient scenarios as much as possible because the climate effects literature suggests that static calculations (typically, CO2 held fixed at double pre-industrial concentrations) do not produce the same time-evolving regional patterns of climatic changes as do transients—and because, of course, the actual Earth is undergoing a transient response to anthropogenic forcings.
Most studies of climate change impacts have focused on changes in mean climate conditions. However, global climate change is likely to bring changes in climate variability and extreme events as well. This is relevant here because decisionmakers often consider hedging strategies to be prepared for the possibility of low-probability but high-consequence events—a risk management framework. Features of projected changes in extreme weather and climate events in the 21st century include more frequent heat waves, less frequent cold spells (barring so-called singular events), greater intensity of heavy rainfall events, more frequent midcontinental summer drought, greater intensity of tropical cyclones, and more intense El Niño-Southern Oscillation (ENSO) events (Table SPM-1, WGII TAR Summary for Policymakers).
A small number of studies have investigated the potential impacts of hypothesized changes in climate variability and/or extreme events. Results of these studies, coupled with observations of impacts from historical events (e.g., Chapter 8), suggest that changes in climate variability and extremes are likely to be at least as important as changes in mean climate conditions in determining climate change impacts and vulnerability (high confidence). The literature suggests that omission of changes in extreme events and/or climate variability will yield underestimates of climate change impacts and vulnerability. In its assessment of potential vulnerabilities and adaptation options, Working Group II has focused on the interactions of natural climate variability and anthropogenic change and the potential for “win-win” adaptation options that would increase resilience to both phenomena.
In many environmental fields, there are thought to be thresholds below which only minor effects occur. Critical levels in acid rain are one example (Brodin and Kuylenstierna, 1992). These kinds of thresholds also are possible in climate change and are incorporated into some models as “tolerable” levels that must be exceeded before significant impacts occur (Hope et al., 1993).
However, in climate change, thresholds have been proposed that are much more complicated. Below the threshold, there may be some impacts, but they will be smoothly varying with the change in climate. Some positive effects might even be observed in some regions or sectors for a small global warming, giving the impression that there is little impact. Above the threshold, however, potentially damaging events may occur. For example, most models show (by 2100) a weakening of thermohaline circulation that transports warmer water to the North Atlantic (see TAR WGI Summary for Policymakers) but only very low confidence that there will be full collapse of the thermohaline circulation by 2100—although some rapid greenhouse buildup scenarios suggest that emissions during the 21st century could trigger a collapse in the following century (e.g., Rahmstorf, 1999; Schneider and Thompson, 2000). Likewise, only very low confidence is given to the prospect of substantial collapse by 2100 of the West Antarctic Ice Sheet (TAR WGI Summary for Policymakers). Other examples of potential threshold phenomena can be found in the literature for regional situations. For example, Wang and Eltahir (2000) demonstrate that rainfall in the Sahel region of Africa can have several equilibrium values, depending on the level of disturbance to vegetation cover. For vegetation removal of less than a threshold value, the system recovers within a few years. For vegetation removal above a threshold, however, there is a new steady-state rainfall regime that is much reduced from “normal” conditions. These thresholds may be, as characterized previously, a result of rapid transient forcing of the climate system, in terms of altered radiative properties of the atmosphere or characteristics of the land surface. Although such threshold events remain somewhat speculative, their impacts clearly would be more severe than smoothly varying (and thus more adaptable) events. Some thresholds in impacts, however, are much less speculative, such as hospital admissions for heat conditions above a threshold temperature—and these threshold temperatures vary regionally as there is some acclimatization to heat stress (Chapter 9)—or species living near mountaintops that would be forced out of existence, even by smooth climatic warming, because they reached the threshold of having no place to move up into (e.g., Still et al., 1999).
Sometimes the expression “threshold” is used as an approximation when the response actually is more likely to be smooth but strongly nonlinear. The release of methane from gas hydrates trapped in deep sediments and the health impacts of thermal stress would be examples of this category. Working Group II assesses potential thresholds for ecological and human systems.
By definition, it is difficult to give examples of the surprises that might be created under a changed climate. Such surprises, however, can make even the most careful calculation of impacts extremely inaccurate, as noted previously. Surprises have been classified by many authors in many contexts (see Schneider et al., 1998, for a review of the literature and many citations). In particular, low-probability events—or those whose probability is difficult to assess—often are labeled rhetorically as “surprises,” even though the event has been classified or identified as known. Strictly speaking, such events are more accurately called “imaginable surprises;” true surprises are wholly unexpected events. Another useful category is “imaginable conditions for surprise” (Schneider et al., 1998), where the specific event in question is unexpected but a set of conditions that increases the likelihood of surprises can be assessed; increasing the rate of forcing of the climatic system is one example, as noted in Section 126.96.36.199.
Investigations into climate change and its potential consequences have begun to highlight the importance of strongly nonlinear, complex, and discontinuous responses. These types of responses, called singularities, can occur at all temporal and spatial scales of systems influenced by climate change (high confidence can be given to the likelihood that some such singularities will occur, but low confidence usually is assigned to any specific example of a possible abrupt event; see Chapter 19). Strongly nonlinear responses are characterized by thresholds—which, if exceeded by a stimulus, result in substantially greater sensitivity to further stimulus or dramatic change, explosive growth, or collapse. Complex responses involve interactions of many intricate elements that yield outcomes that are not easily predicted. Examples of these types of responses include coral bleaching, collapse of fish stocks, disease outbreaks, changes in fire and other disturbance regimes in vegetation systems, crop failure, malnutrition and hunger, and collapse of pastoral communities. Advances in our understanding of these types of responses are largely qualitative, but they are important in understanding the character of dangers posed by climate change. Omission of potential nonlinear and complex responses from climate change impact assessments is expected (well-established, but incomplete) to yield underestimates of impacts (see Chapters 5 and 19). Because of the magnitude of their potential consequences, large-scale discontinuous responses warrant careful consideration in evaluations of climate change dangers. Working Group II points to the potential for such occurrences and their potential consequences for human and natural systems, but it is unable to provide detailed assessments of potential effects, given the paucity of information in the literature.
Other reports in this collection