2 Evaluating urban land use simulation models

Compared to natural or agricultural landscapes, urban systems are strongly influenced by both the social and the natural dimension. As mentioned in the introduction, urban landscapes are coupled with human–nature systems (Liu et al., 2007), with many interlinkages between the human sphere – first and foremost demography and economy – land use and the environment. Figure 1* provides a very general but comprehensive overview on the major components of an urban landscape: the major driving force for change is the human sphere, which creates pressures on the state of the land use, which again will have effects on the environment, its natural resources and ecosystems.

Figure 1: Main components (human sphere, land use, natural resources) and relationships (1 – 4) which describe human–nature interactions in urban regions: (1) Impact of human sphere on land use, (2) feedback of land use on human sphere, (3) impact of land use on environment (including ecosystems) and (4) feedback of environment on human sphere.

The human sphere characterises the socio-economic system of cities: it comprises variables such as population (development), households, spatial planning and governance, the real estate market, commercial activities and infrastructure, including transportation. Specifically, the human sphere includes human decision making and actions upon land use. The land use component itself comprehends all types of typical urban land uses such as residential, industrial, commercial, transport and recreation. The third component contains natural resources, such as ecosystems, biodiversity, soil functions and water resouces (cf. again Figure 1*). We set up these feedback loops between the three dimensions/components of the urban system discussed above: (1) the impact of the human sphere on land use, (2) the feedback (= reverse to the impact function) of land use on the human sphere, (3) the impact of land use on the environment and (4) another feedback of the environment on the human sphere. All relationships are labelled in Figure 1*, respectively. Furthermore, a short Section (4.4) deals with the scale-specific causal feedbacks between local and regional scale, insofar as they are covered by the models investigated. The evaluation of the feedback loops includes (1) the identification of a respective formal representation of the respective causalities in the model and (2) whether or not they have an impact on other model components again/vice versa. In order to structure the review and to give brief overviews of the models under review, we summarised the findings of the analysis of each of the models in Table 1, which provides comprehensive information about the main purpose and major components classified according to Figure 1*.