Climate Change 2001: Impacts, Adaptation and Vulnerability

3.7. Representing Interactions in Scenarios and Ensuring Consistency

3.7.1. Introduction

There is great diversity in the scenarios adopted in impact assessments. This diversity is valuable in providing alternative views of the future, although it can hamper attempts to summarize and interpret likely impacts by introducing inconsistencies within or between studies. Moreover, there are certain key dependencies in climate change science that have resulted in time lags and inconsistencies in the application of scientific results between different research areas. This has been reflected in the IPCC process (see Table 3-6). Thus, although TAR WGI reviews recent projections of future climate, these results are not yet available to the impacts community to prepare and publish their analyses, on which the TAR WGII assessment is based. Instead, most impact studies have relied on earlier, more rudimentary climate projections. Similarly, the simplified assumptions used in climate model simulations about changes in radiative forcing of the climate from changing GHG and aerosol concentrations represent only a limited subset of plausible atmospheric conditions under a range of emissions scenarios reviewed by TAR WGIII.

Creation of comprehensive scenarios that encompass the full complexity of global change processes and their interactions (including feedbacks and synergies) represents a formidable scientific challenge. This section addresses some components of this complexity. First it treats generally accepted biogeochemical processes; second, it addresses emerging climate-system processes; and third, it reviews rarely considered interactions between anthropogenic and natural driving forces. Finally, the importance of comprehensiveness and compatibility in scenario development is discussed.

3.7.2. Representing Processes and Interactions in Scenarios

3.7.2.1. Generally Considered Interactions

Emissions of greenhouse gases have increased their atmospheric concentrations, which alter the radiative properties of the atmosphere and can change the climate (see TAR WGI Chapters 3-8). Determination of atmospheric concentrations from emissions is not straightforward; it involves the use of models that represent biogeochemical cycles and chemical processes in the atmosphere (Harvey et al., 1997; TAR WGI Chapters 3-5). Several atmosphere-ocean interactions are considered in defining the future transient response of the climate system (Sarmiento et al., 1998; TAR WGI Chapter 8). For the purposes of scenario development, CO₂ occupies a special role, as a greenhouse gas (IPCC, 1996a) and by directly affecting carbon fluxes through CO₂ fertilization and enhanced water-use efficiency (see Section 3.4.2). These direct responses are well known from experimentation (Kirschbaum et al., 1996). Biospheric carbon storage is further strongly influenced by climate, land use, and the transient response of vegetation. All of these interactions define the final CO₂ concentrations in the atmosphere and subsequent levels of climate change (see Table 3-7).

The early simple climate models that were used in the IPCC's First and Second Assessment Reports all emphasized the importance of CO₂ fertilization but few other biogeochemical interactions (Harvey et al., 1997). Inclusion of more realistic responses of the carbon cycle in climate scenarios still is an evolving research area (Walker et al., 1999), but most interactions now are adequately represented.

3.7.2.2. Less Considered Interactions

Interactions between land, vegetation, and the atmosphere have been studied extensively in deforestation and desertification model experiments (Charney et al., 1977; Bonan et al., 1992; Zhang et al., 1996; Hahmann and Dickinson, 1997). Changes in surface characteristics such as snow/ice and surface albedo and surface roughness length modify energy, water, and gas fluxes and affect atmospheric dynamics. These interactions occur at various scales (Hayden, 1998), but although their importance is well appreciated (Eltahir and Gong, 1996; Manzi and Planton, 1996; Lean and Rowntree, 1997; Zeng, 1998) they still generally are ignored in scenario development.

Climate modeling studies (e.g., Henderson-Sellers et al., 1995; Thompson and Pollard, 1995; Sellers et al., 1996) suggest an additional warming of about 0.5°C after deforestation on top of the radiative effects of GHG, but these effects are not necessarily additive on regional scales. Betts et al. (1997) concur that vegetation feedbacks can be significant for climate on regional scales. More recent studies, however, tend to predict smaller changes, partly as a result of the inclusion of more interactions such as the cloud radiative feedback. Field experiments show large changes in surface hydrology and micrometeorological conditions at deforested sites (Gash et al., 1996). On the other hand, observations have not provided direct evidence of changes in overall climate in the Amazon basin (Chu et al., 1994) or in Sahel surface albedo (Nicholson et al., 1998), but the available data series are too short to be conclusive.

Table 3-8: Summary of scenarios adopted in an assessment of global impacts on five sectors (Parry and Livermore, 1999).

Scenario Type (up to 2100)	Ecosystems^a	Water Resources^b	Food Security^c	Coastal Flooding^d	Malaria Risk^e

Socioeconomic/technological
- Population	—	X	X	X	X
- GDP	—	—	X	X	—
- GDP per capita	—	—	X	X	—
- Water use	—	X	—	—	—
- Trade liberalization	—	—	X	—	—
- Yield technology	—	—	X	—	—
- Flood protection	—	—	—	X	—

Land-cover/land-use change	—	—	—	X	—

Environmental
- CO₂ concentration	X	—	X	—	—
- Nitrogen deposition	X	—	—	—	—

Climate
- Temperature	X	X	X	—	X
- Precipitation	X	X	X	—	X
- Humidity	X	X	—	—	—
- Cloud cover/radiation	X	X	—	—	—
- Windspeed	—	X	—	—	—
- Diurnal temperature range	X	—	—	—	—

Sea level	—	—	—	X	—

^a White et al. (1999) and see Chapter 5. ^b Arnell (1999) and see Chapters 4 and 19. ^c Parry et al. (1999) and see Chapters 5 and 19. ^d Nicholls et al. (1999) and see Chapters 6, 7, and 19. ^e Martens et al. (1999) and see Chapters 8 and 18.

Palaeoclimatic reconstructions, using empirical data and model results, provide better opportunities to study vegetation-atmosphere interactions. Climate models that incorporate dynamic vegetation responses simulate larger vegetation shifts for changed past climates than expected by the orbitally forced climate effect alone. For example, an additional 200-300 km poleward displacement of forests simulated for 6,000 ky BP in North America was triggered by changes in surface albedo (Kutzbach et al., 1996; Texier et al., 1997; Ganopolski et al., 1998). However, these shifts are not observed in all model experiments (e.g., Broström et al., 1998). Other modeling results suggest that oceans also play a prominent role (Hewitt and Mitchell, 1998). Thus, vegetation-ocean-climate interaction seems to be important in defining regional climate change responses.

Most vegetation models used in scenario development are equilibrium models (i.e., for a given climate they predict a fixed vegetation distribution). The latest dynamic vegetation models attempt to include plant physiology, biogeochemistry, and land surface hydrology (e.g., Goudriaan et al., 1999), and some explicitly treat vegetation structure and succession. Foley et al. (1998) coupled one such model to a GCM and found that the most climatically sensitive zones were the desert/grassland and forest/tundra ecotones. These zones also tend to be exposed to large disturbances and natural climate variability (Schimel et al., 1997b). In another model experiment, Zeng and Neelin (1999) found that interannual and inter-decadal climate variability helps to keep the African savannah region from getting either too dry or too wet, through nonlinear vegetation-atmosphere interactions. Few of these models contain simulations of disturbances, such as fire regimes (Crutzen and Goldammer, 1993; Kasischke and Stocks, 2000), which rapidly alter vegetation patterns and influence vegetation responses. Unfortunately, hardly any of these insights are included routinely in scenario development.

Other reports in this collection