Exploration of spatial scale sensitivity in geographic cellular automata

Cellular automata (CA) are individual-based models in which states, time, and space are discrete. Spatiotemporal dynamics emerge from the simple and local interactions of the cells. When using CA in a geographic context, nontrivial questions have to be answered about the choice of spatial scale, namely cell size and neighbourhood configuration. However, the spatial scale decisions involved in the elaboration of geographic cellular automata (GCA) are often made arbitrarily or in relation to data availability. The objective of this study is to evaluate the sensitivity of GCA to spatial scale. A stochastic GCA was built to model land-cover change in the Maskoutains region (Quebec, Canada). The transition rules were empirically derived from two Landsat-TM (30 in resolution) images taken in 1999 and 2002 that have been resampled to four resolutions (100, 200, 500, 1000m). Six different neighbourhood configurations were considered (Moore, Von Neumann, and circular approximations of 2, 3, 4, and 5 cell radii). Simulations were performed for each of the thirty spatial scale scenarios. Results show that spatial scale has a considerable impact on simulation dynamics in terms of both land-cover area and spatial structure. The spatial scale domains present in the results reveal the nonlinear relationships that link the spatial scale components to the simulation results.