Advantages: Aggregate analyses synthesize climate change impacts in an internally consistent manner, using relatively comprehensive global indicators or metrics. These often are expressed in U.S. dollars (e.g., Tol, 2001b) or other common metrics such as changes in vegetation cover (Alcamo et al., 1998). This enables direct comparisons of impacts among sector systems and regions and with other environmental problems and emission control costs. Some aggregate analyses have assessed differences in relative impacts in developed and developing regions of the world and have shown that regional differences in impacts may be substantial.
Disadvantages: Aggregate analyses lack richness of detail. Partly this is inherent because aggregation explicitly seeks to synthesize complex information. Partly this is because aggregate analyses tend to rely on reduced-form models. Condensing the diverse pattern of impacts into a small number of damage indicators is difficult. Some metrics may not accurately capture the value of certain impacts; for example, nonmarket impacts such as mortality and loss of species diversity or cultural heritage often are not well captured in monetization approaches, and change in vegetation cover may not clearly indicate threats to biodiversity. Other complicating issues concern comparison of impacts across time (impact today and several generations from now) and between regions (e.g., impact in developing and developed countries), as well as how much importance to assign to different effects. In addition, many aggregate studies examine a static world rather than a dynamic one and do not consider the effects of changes in extreme events or large-scale discontinuities. The aggregation process is not possible without value judgments, and different ethical views imply different aggregate measures across socioeconomic groupings and generations (see Azar and Sterner, 1996; Fankhauser et al., 1997). Choice of discount rates can affect valuation of damages. In addition, general shortcomings that affect all reasons for concern are particularly prominent in aggregate analysis (e.g., accounting for baseline development, changes in variability and extreme events, and costs and benefits of adaptation).
Uncertainties: Uncertainties include whether all climate change impacts (positive and negative) are included, the implications of various aggregation and valuation methods, and implicit or explicit assumptions of methods, including possible mis-specifications of nonlinearities and interaction effects.
Research Needs: The next generation of aggregate estimates will have to account better for baseline developments, transient effects, climate variations, and multiple stresses. Further progress also is still needed in the treatment of adaptation. A broader set of primary studies on impacts in developing countries and nonmarket sectors would reduce the need for difficult extrapolation. More work also is needed on the ethical underpinnings of aggregation and on alternative aggregation schemes. Work on reflecting information from the other reasons for concern into the aggregate approach is underway, but proceeding slowly.
Advantages: Integrated assessment frameworks or models provide a means of structuring the enormous amount of and often conflicting data available from disaggregated studies. They offer internally consistent and globally comprehensive analysis of impacts; provide "vertical integration" (i.e., cover the entire "causal chain" from socioeconomic activities giving rise to GHG emissions to concentration, climate, impacts, and adaptations); provide "horizontal integration" (i.e., account for interlinkages between different impact categories, adaptations, and exogenous factors such as economic development and population growth); and allow for consistent treatment of uncertainties. IAMs have been used primarily for benefit-cost and inverse (or threshold) analyses. The latter have the advantage of being directly related to Article 2 because they define impacts that may be considered "dangerous" (through specification of thresholds related to, e.g., harm to unique and threatened systems or the probability of large-scale discontinuities).
Disadvantages: The main disadvantages with most IAMs are those associated with aggregate approaches: reliance on a single or a limited number of universal measures of impacts. These may not adequately measure impacts in meaningful ways. This is partly because IAMs rely on reduced-form equations to represent the complexities of more detailed models. Their usefulness is highly dependent on how well they are able to capture the complexities of more disaggregated approaches. Some of the IAMs used for benefit-cost analyses have considered large-scale irregularities (e.g., Gjerde et al., 1999), but inclusion of such outcomes is preliminary. Few have accounted for loss of or substantial harm to unique and threatened systems. Although inverse (or threshold) approaches allow researchers to overcome these problems, the disadvantages of this kind of analysis include the difficulty of explicitly specifying thresholds and combining them within and across sectors and regions.
Uncertainties: Uncertainties are the same as those for the aggregate approach or for unique and threatened systems, depending on the structure and objectives of the model. This also would include the effects of different assumptions, methods, and value choices.
Research Needs: Among the biggest challenges facing integrated assessment modelers (see Weyant et al., 1996) are developing a credible way to represent and value the impacts of climate change; a credible way to handle low-probability but potentially catastrophic events; a credible way to incorporate changes in extreme weather events; and realistic representations of changes in socioeconomic and institutional conditions, particularly in developing countries. In addition, they must decide how to allow explicitly for effects of different value choices, systems, and assumptions; how to quantify uncertainties; and how to credibly incorporate planned adaptation, including costs and limitations.
Advantages: Extreme events are recognized as major contributors to the impacts of climate variability now and to potential impacts of climate change in the future. Thus, realistic climate change impact assessments must take them into account even though they may change in complex wayssuch as in frequency, magnitude, location, and sequences (e.g., increased variability may lead to more frequent floods and droughts). Better understanding of changes in extreme events and adaptation measures for coping with them also will help in coping with present variability.
Disadvantages: Extreme events are more difficult to model and characterize than average climates. Changes in extreme events will be complex and uncertain, in part because extremes occur in a chaotic manner even in the present climate. Large data series are needed to characterize their occurrence because, by definition, they are rare events. This means that long time scale model simulations are needed to develop relevant statistics from long time slices or multiple realizations. Extreme events need to be considered in terms of probabilities or risks of occurrence rather than predictions. This chaotic element adds to other sources of uncertainty. It means that engineering or other design standards based on climatology that normally use long data series of observations will need a synthetic data set that simulates potential changes in future climate. It also makes adaptation to changes in extremes more difficult because planned adaptation must rely on necessarily uncertain projections into the future from theory and thus requires greater faith in the science before the information will be acted on.
Uncertainties and Research Needs: Better knowledge of the behavior of extremes will require long or multiple simulations at finer spatial and temporal scales, to capture the scale, intensity, and frequency of the events. Some types of extreme events (e.g., hail and extreme wind bursts) are poorly simulated at present; others, such as ENSO and tropical cyclones, are extremely complex and only now are beginning to be better simulated. Arguments for changes in their behavior are still often largely theoretical, qualitative, or circumstantial, rather than well based in verified models. Moreover, much more work is needed on how they will affect natural and human systems and how much of the recent trend to greater damages from extreme events is related to changes in exposure (e.g., greater populations, larger investments, more insurance cover, or greater reporting) rather than changes in the number and intensity of those extremes. More work is needed on how best to adapt to changes in extreme events, especially on how planners and decisionmakers can best take information on projected changes in extremes into consideration. This may be done best by focusing on projected change in the risk of exceeding prescribed natural, engineering, or socioeconomic impacts thresholds.
Advantages: Consideration of strongly nonlinear or even disruptive effects accompanying climate change is a critical component of the "dangerous interference" debate. The basic idea is to corroborate any non-negligible probability for high-consequence impacts that may be triggered by human climate perturbations. The political process to avoid high-consequence impacts may be facilitated by the global scope of such effects (e.g., disintegration of the WAIS generating a planetary sea-level rise of approximately 5m). Inclusion of extreme events in the analysis helps, in general, to pursue all other reasons for concern in a realistic way because irregular impacts may dominate impacts on unique and threatened systems, distributional impacts, and aggregate impacts.
Disadvantages: This is an emerging area of research, facing several serious challenges because of the complexity of nonlinear interactions to be considered. The prevailing lack of knowledge is reflected in use of the term "surprises" for disruptive events. The potentials for climate change-induced transformations of extreme events regimes and for large-scale discontinuities in the Earth system are still highly uncertain. The search for irregularities might turn out to be futile and distract scientific resources from other important topics, such as the distributional aspects of regular climate change impacts.
Uncertainties and Research Needs: By definition, uncertainties are most severe in this realm of impact research. At present, there is no way of estimating the probabilities of certain disruptive events or assigning confidence levels to those probabilities. As a consequence, a strong research program should be launched that combines the best paleoclimate observations with the strongest simulation models representing full and intermediate complexity.
Looking across the different analytic approaches (implicitly, the different reasons for concern), it is clear that to a great extent they complement and in many respects do not overlap each other. Combining these approaches into an integrated framework is the ambition of IAMs, at least in principle. However, this process is just starting. Because observed evidence has not been incorporated in the other analytic approaches, impacts to unique and threatened systems have not been accounted for in aggregate and IAM approaches, they are difficult to sum, and large-scale irregular impacts have only begun to be addressed, it does not appear to be feasible yet to combine these approaches into a comprehensive analytic approach. Thus, those who are seeking to implement climate policies must currently do their own integration of information from the alternative lines of inquiry.
Other reports in this collection