Using Cellular Automata for Integrated Modelling of Socio-environmental Systems

1) Guy Engelen , 2) Roger White , 1) Inge Uljee and 1) Paul Drazan
1) RIKS Research Institute for Knowledge Systems; P.O. Box 463; 6200 AL Maastricht; The Netherlands
2) Memorial University of Newfoundland; St. John’s; NF; Canada A1B 3X9

Abstract
Cellular automata provide the key to a dynamic modelling and simulation framework that integrates socioeconomic with environmental models, and that operates at both micro and macro geographical scales. An application to the problem of forecasting the effect of climate change on a small island state suggests that such modelling techniques could help planners and policy makers design more effective policies --policies better tuned both to specific local needs and to overall socio-economic and environmental constraints.


IntroductionTown and country planners face the difficult task of dealing with a world that is complex, interconnected,and ever-changing. Coastal zone management, urban land-use planning, and the design of policies forsustainable economic development all pose the problem of dealing with systems in which natural andhuman factors are thoroughly intertwined. There is growing scientific evidence that a purely macroscopicapproach to these problems does not suffice, because spatial and organizational details are important inunderstanding the dynamics of such systems (Allen and Lesser, 1991; Kauffman, 1993; Langton, 1992;Nicolis et al. 1989).At the descriptive level, the need for spatial detail is attained in Geographical Information Systems. But,in order to put forward effective measures for changing --or maintaining-- the organization of socio-what is where why it iseconomic and environmental systems, it is necessary not only to know but alsothere. These systems must be understood and managed as coherent dynamic entities, so that systemintegrity is maintained. We present here a dynamic modelling framework and encompassing decisionsupport shell that is capable of integrating socio-economic and environmental factors at a variety of scales,while representing spatial dynamics with a high level of geographical detail. This modelling framework isquite general in terms of the situations to which it can be usefully applied. But we will present it here inthe form of an example --an application concerning the impact of climate change on a small island state.An example: Exploring the Impact of Climate Change on a Small Island.macro-scaleShifting climate conditions, expressed at the in terms of changes in temperature,precipitation, and storm frequency, are likely to affect productivity levels, demand patterns, and exportsand imports, and will probably cause migration of people and their activities as well (see e.g. Alm et al.,micro-scale1993). But all of these effects are actually expressed, on the ground, as phenomena. Forexample, an increase in the total export demand for a particular agricultural product will normally meanthat more land will be required. But the consequences will be very different depending on whether theland is found by converting existing agricultural land or by clearing forested land, especially if the latter iseasily eroded or is itself ecologically significant. Furthermore, changes in productivity that may occur asother activities are displaced onto more marginal land, or as erosion causes loss of fertility, will in turnhave repercussions on the macro-level economics. In other words, the spatial details of land use areimportant in understanding the impact of macro-level changes.No one model is capable of capturing the whole range of these phenomena, from those operating on aworld scale down to those that threaten strips of beach or affect individual fields. For example, spatialinteraction based models, consisting of sets of linked dynamic equations, are useful for representingspatial and temporal dynamics at regional scales (White, 1977; Engelen and Allen, 1986; Pumain et al.,1989), but become computationally impractical when much spatial detail is required (White and Engelen,1993). On the other hand, models capable of dealing with extreme geographical detail, such as thoseavailable in Geographical Information Systems, lack the dynamics required to represent the processesoperating in the system (Brimicombe, 1993). One solution to the problem is to make use of a modellingframework consisting of two linked components --one for macro-level processes and another for thoseoperating at the micro-level. Both components exchange results continuously during the simulation andget the data relevant for their level of detail from the same geographical database, ideally a GIS (Figure1).macro-levelAt the , the modelling framework integrates several component sub-models, representing thenatural, social, and economic sub-systems. These are all linked to each other in a network of mutual,reciprocal, influence (Figure 1, top). The macro level allows for the use of regionalised representationsand for different types of mathematical formulations, thus permitting a more or less detailed modelling ofvarious aspects of the sub-systems as required for specific applications. For the case of the smallprototypical Caribbean Island, the macro-level is modelled as a single point in interaction with the worldoutside.