Cellular Strategies for the Simulation of Human Spatial Systems

MACMILLAN, W.D. (bmacmill@nta.geog.ox.ac.uk), University of Oxford, School of Geography, Mansfield Road, Oxford OX1 3TB, U.K.

Key Words: multi-agent simulation, cellular modelling, irregular networks, scale, economic agency

Numerical systems in which the state of each cell in a 2-D or 3-D grid is a function of its previous state(s) and that of its neighbours are well suited to the description of hydrological, geomorphological, and atmospheric processes (see Burrough, 1998); however, they are of limited value in representing human spatial systems. Multiagent simulation has the advantage of separating the treatment of populations of agents from that of the territory that they occupy (see Benenson, 1998). Helpful though this is, it is not sufficient, on its own, to allow a thorough exploration of many economic and social processes. Part of the difficulty lies in the fact that economic and social actions are not generally confined to neighbourhoods. This applies both in a spatial sense, where action at a distance is important and, in time, where actions may occur over different time horizons. The spaces of human geographical activity tend to be complex (see Cliff and Haggett, 1998, on the use of multidimensional scaling to produce metrics based on diffusion rates). The complexity has a number of sources, one of the most important of which can be thought of as the vector structure of human landscapes. Some work has been done using vector structures in the form of regular networks, which is broadly analogous to cellular modelling in terms of strategy (see Peeters et al., 1998) but this approach is still in its infancy.

The purpose of this paper is to explore the scope and limitations of cellular modelling (and the related approaches referred to above) with respect to spatial economic processes. It looks at questions of spatial and temporal scale, the treatment of economic agency, the replacement of conventional approaches to pricing and allocation by iterative, agent-based simulation, and the range of problem types that might be tackled by a cellular strategy. It also looks at the problem of devising a visual vocabulary for human spatial simulation in a variety of contexts.

References

Benenson, I. (1998), Multiagent Simulations of Residential Dynamics in the City, Computers, Environment and Urban Systems, 22(1), pp. 25-42.

Burrough, P. (1998), Dynamic Models and Geocomputation, in Geocomputation: A Primer (P. Longley, S. Brooks, R. McDonnell, and W. MacMillan, eds.), Wiley.

Cliff, A. D. and P. Haggett. (1998) On Complex Geographical Space: Computing Frameworks for Spatial Diffusion Processes, in Geocomputation: A Primer (P. Longley, S. Brooks, R. McDonnell, and W. MacMillan, eds.), Wiley.

Peeters, D., J.F. Thisse, and I. Thomas (1998), Transportation Networks and the Location of Human Activities, Geographical Analysis, 30(4), pp. 355-371.