This section discusses some of the caveats of climate scenario development and focuses on the need for consistency in representing different physical aspects of the climate system. It does not discuss the many possible inconsistencies with respect to socio-economic issues in scenario development. Chapter 3 of the TAR WG II (Carter and La Rovere, 2001) and Chapter 2 of the TAR WG III (Morita and Robinson, 2001) provide a detailed treatment of these issues. Three common inconsistencies in applying climate scenarios are discussed, concerning the representation of ambient versus equivalent CO2 concentrations, biosphere-ocean-atmosphere interactions and time lags between sea level rise and temperature change.
The climate system consists of several components that interact with and influence each other at many different temporal and spatial scales (see Chapter 7). This complexity adds further constraints to the development of climate scenarios, though their relevance is strongly dependent on the objectives and scope of the studies that require scenarios. Most climate scenarios are based on readily available climate variables (e.g., from AOGCMs) and, where these are used in impact assessments, studies are often restricted to an analysis of the effects of changes in climate alone. However, other related environmental aspects may also change, and these are often neglected or inadequately represented, thus potentially reducing the comprehensiveness of the impact assessment. Furthermore, some feedback processes that are seldom considered in AOGCM simulations, may modify regional changes in climate (e.g., the effect of climate-induced shifts in vegetation on albedo and surface roughness).
Concurrent changes in atmospheric concentrations of gases such as CO2, sulphur dioxide (SO2) and ozone (O3) can have important effects on biological systems. Studies of the response of biotic systems require climate scenarios that include consistent information on future levels of these species. For example, most published AOGCM simulations have used CO2-equivalent concentrations to represent the combined effect of the various gases. Typically, only an annual 1% increase in CO2-equivalent concentrations, which approximates changes in radiative forcing of the IS92a emission scenario (Leggett et al., 1992), has been used. However, between 10 and 40% of this increase results from non-CO2 greenhouse gases (Alcamo et al., 1995). The assumption that CO2 concentrations equal CO2-equivalent concentrations (e.g., Schimel et al., 1997; Walker et al., 1999) has led to an exaggeration of direct CO2 effects. If impacts are to be assessed more consistently, proper CO2 concentration levels and CO2-equivalent climate forcing must be used. Many recent impact assessments that recognise these important requirements (e.g., Leemans et al., 1998; Prinn et al., 1999; Downing et al., 2000) make use of tools such as scenario generators (see Section 18.104.22.168) that explicitly treat atmospheric trace gas concentrations. Moreover, some recent AOGCM simulations now discriminate between the individual forcings of different greenhouse gases (see Chapter 9, Table 9.1)
The biosphere is an important control in defining changes in greenhouse gas concentrations. Its surface characteristics, such as albedo and surface roughness, further influence climate patterns. Biospheric processes, such as CO2-sequestration and release, evapotranspiration and land-cover change, are in turn affected by climate. For example, warming is expected to result in a poleward expansion of forests (IPCC, 1996b). This would increase biospheric carbon storage, which lowers future CO2 concentrations and change the surface albedo which would directly affect climate. A detailed discussion of the role of the biosphere on climate can be found elsewhere (Chapters 3 and 7), but there is a clear need for an improved treatment of biospheric responses in scenarios that are designed for regional impact assessment. Some integrated assessment models, which include simplifications of many key biospheric responses, are beginning to provide consistent information of this kind (e.g., Alcamo et al., 1996, 1998; Harvey et al., 1997; Xiao et al., 1997; Goudriaan et al., 1999).
Another important input to impact assessments is sea level rise. AOGCMs usually calculate the thermal expansion of the oceans directly, but this is only one component of sea level rise (see Chapter 11). Complete calculations of sea level rise, including changes in the mass balance of ice sheets and glaciers, can be made with simpler models (e.g., Raper et al., 1996), and the transient dynamics of sea level rise should be explicitly calculated because the responses are delayed (Warrick et al., 1996). However, the current decoupling of important dynamic processes in most simple models could generate undesirable inaccuracies in the resulting scenarios.
Climate scenario generators can comprehensively address some of these inconsistencies. Full consistency, however, can only be attained through the use of fully coupled global models (earth system models) that systematically account for all major processes and their interactions, but these are still under development.
Other reports in this collection