Human Geography Applications

Cellular automata have been employed in the study of a wide range of geographic phenomena. The plethora of applications is illustrative of the usefulness of the approach. Human geography applications are largely focused on issues relating to urban geography and behavioral geography.

Models of urbanization and urban growth feature prominently in the literature. Cells are generally used to represent areas of urbanization or land parcels, with cell boundaries matched to pixels in remotely sensed images. States can be relatively simple, with binaries representing urbanization, for example. State specification can, however, be much richer, formulated as layers of land use or activity (synonymous with attribute layering in a GI system), or as fuzzy membership. A variety of transition rules have been employed as representative of the forces of urbanization. These include urbanization based on agglomeration through diffusion limited aggregation; land use transition based on potential for development; physics driven mechanics; data driven rules from sta tistical analysis; data trained rule sets from artificial neural networks; and spatial heuristics. In some cases, different rule sets are employed as what if experiments to test the varying influence of urbanization processes. In this sense, cellular automata are used as artificial laboratories to test theory, as tools to think with.

Cellular automata have also been used to model other aspects of human geography. Work by Schelling and Sakoda with simple chessboard models is among the early antecedents of cellular automata simulations of sociospatial segregation. Cells are equipped with states that correspond to ethnicities, and rules based on preferences for co location in space are used to show how small biases can quickly lead to large scale segregation from random initial conditions, and to investigate the tipping point at which sociospatial segregation begins. More recent models of residential mobility follow similar schemes, with rules designed to introduce more elaborate mechanisms based on spatial choice and dissonance.

David O’Sullivan has also developed a model of gentrification dynamics based on cellular automata. Cells are used to represent gentrifiable real estate and rules are introduced to test rent gap hypotheses from the theoretical literature.

Automata based work on traffic simulation is largely agent based, but several applications have been developed using cellular automata. Cells are used to represent vehicles and pedestrians in these instances, with movement formulated by proxy using transition rules that pass the presence of these entities as state information between cells based on heuristics designed to mimic lane changing, collision avoidance, stopping, queuing, and junction navigation.