2.4.6.5 Land-use scenarios
Many CCIAV studies need to account for future changes in land use and land cover. This is especially important for regional studies of agriculture and water resources (Barlage et al., 2002; Klöcking et al., 2003), forestry (Bhadwal and Singh, 2002), and ecosystems (Bennett et al., 2003; Dirnbock et al., 2003; Zebisch et al., 2004; Cumming et al., 2005), but also has a large influence on regional patterns of demography and economic activity (Geurs and van Eck, 2003) and associated problems of environmental degradation (Yang et al., 2003) and pollution (Bathurst et al., 2005). Land-use and land-cover change scenarios have also been used to analyse feedbacks to the climate system (DeFries et al., 2002; Leemans et al., 2002; Maynard and Royer, 2004) and sources and sinks of GHGs (Fearnside, 2000; El-Fadel et al., 2002; Sands and Leimbach, 2003).
The TAR concluded that the use of Integrated Assessment Models (IAMs) was the most appropriate method for developing land-use change scenarios, and they continue to be the only available tool for global-scale studies. Since the TAR, however, a number of new models have emerged that provide fresh insights into regional land-use change. These regional models can generate very different land-use change scenarios from those generated by IAMs (Busch, 2006), often with opposing directions of change. However, the need to define outside influences on land use in regional-scale models, such as global trade, remains a challenge (e.g., Sands and Edmonds, 2005; Alcamo et al., 2006b), so IAMs have an important role to play in characterising the global boundary conditions for regional land-use change assessments (van Meijl et al., 2006).
Regional-scale land-use models often adopt a two-phase (nested scale) approach with an assessment of aggregate quantities of land use for the entire region followed by ‘downscaling’ procedures to create regional land-use patterns (see Box 2.7 for examples). Aggregate quantities are often based on IAMs or economic models such as General Equilibrium Models (van Meijl et al., 2006) or input-output approaches (Fischer and Sun, 2001). Methods of downscaling vary considerably and include proportional approaches to estimate regional from global scenarios (Arnell et al., 2004), regional-scale economic models (Fischer and Sun, 2001), spatial allocation procedures based on rules (Rounsevell et al., 2006), micro-simulation with cellular automata (de Nijs et al., 2004; Solecki and Oliveri, 2004), linear programming models (Holman et al., 2005a, b), and empirical-statistical techniques (de Koning et al., 1999; Verburg et al., 2002, 2006). In addressing climate change impacts on land use, Agent-Based Models (ABMs: see Alcamo et al., 2006b) aim to provide insight into the decision processes and social interactions that underpin adaptation and vulnerability assessment (Acosta-Michlik and Rounsevell, 2005).
Most land-use scenario assessments are based on gradual changes in socio-economic and climatic conditions, although responses to extreme weather events such as Hurricane Mitch in Central America have also been assessed (Kok and Winograd, 2002). Probabilistic approaches are rare, with the exception being the effects of uncertainty in alternative representations of land-use change for hydrological variables (Eckhardt et al., 2003). Not all land-use scenario exercises have addressed the effects of climate change even though they consider time-frames over which a changing climate would be important. This may reflect a perceived lack of sensitivity to climate variables (e.g., studies on urban land use: see Allen and Lu, 2003; Barredo et al., 2003, 2004; Loukopoulos and Scholz, 2004; Reginster and Rounsevell, 2006), or may be an omission from the analysis (Ahn et al., 2002; Berger and Bolte, 2004).
Box 2.7. SRES-based land-use and land-cover characterisations
Future land use was estimated by most of the IAMs used to characterise the SRES storylines, but estimates for any one storyline are model-dependent, and therefore vary widely. For example, under the B2 storyline, the change in the global area of grassland between 1990 and 2050 varies between -49 and +628 million ha (Mha), with the marker scenario giving a change of +167 Mha (Naki´cenovi´c et al., 2000). The IAM used to characterise the A2 marker scenario did not include land-cover change, so changes under the A1 scenario were assumed to apply also to A2. Given the differences in socio-economic drivers between A1 and A2 that can affect land-use change, this assumption is not appropriate. Nor do the SRES land-cover scenarios include the effect of climate change on future land cover. This lack of internal consistency will especially affect the representation of agricultural land use, where changes in crop productivity play an important role (Ewert et al., 2005; Audsley et al., 2006). A proportional approach to downscaling the SRES land-cover scenarios has been applied to global ecosystem modelling (Arnell et al., 2004) by assuming uniform rates of change everywhere within an SRES macro-region. In practice, however, land-cover change is likely to be greatest where population and population growth rates are greatest. A mismatch was also found in some of the SRES storylines, and for some regions, between recent trends and projected trends for cropland and forestry (Arnell et al., 2004).
More sophisticated downscaling of the SRES scenarios has been undertaken at the regional scale within Europe (Kankaanpää and Carter, 2004; Ewert et al., 2005; Rounsevell et al., 2005, 2006; Abildtrup et al., 2006; Audsley et al., 2006; van Meijl et al., 2006). These analyses highlighted the potential role of non-climate change drivers in future land-use change. Indeed, climate change was shown in many examples to have a negligible effect on land use compared with socio-economic change (Schröter et al., 2005b). Technology, especially as it affects crop yield development, is an important determinant of future agricultural land use (and much more important than climate change), contributing to declines in agricultural areas of both cropland and grassland by as much as 50% by 2080 under the A1FI and A2 scenarios (Rounsevell et al., 2006). Such declines in land use did not occur within the B2 scenario, which assumes more extensive agricultural management, such as ‘organic’ production systems, or the widespread substitution of agricultural food and fibre production by bioenergy crops. This highlights the role of policy decisions in moderating future land-use change. However, broad-scale changes often belie large potential differences in the spatial distribution of land-use change that can occur at the sub-regional scale (Schröter et al., 2005b; see also Figure 2.7), and these spatial patterns may have greater effects on CCIAV than the overall changes in land-use quantities (Metzger et al., 2006; Reidsma et al., 2006).