Historically, cellular automata (CA) is a discrete dynamical mathematical structure defined on spatial grid. Research on
cellular automata system (CAS) has focused on rule sets and initial condition and has not discussed its adjacency. Thus,
the main focus of our study is the effect of adjacency on CA behavior. This paper is to compare rectangular grids with
hexagonal grids on their characteristics, strengths and weaknesses. They have great influence on modeling effects and
other applications including the role of nearest neighborhood in experimental design. Our researches present that
rectangular and hexagonal grids have different characteristics. They are adapted to distinct aspects, and the regular
rectangular or square grid is used more often than the hexagonal grid. But their relative merits have not been widely
discussed. The rectangular grid is generally preferred because of its symmetry, especially in orthogonal co-ordinate
system and the frequent use of raster from Geographic Information System (GIS). However, in terms of complex terrain,
uncertain and multidirectional region, we have preferred hexagonal grids and methods to facilitate and simplify the
problem. Hexagonal grids can overcome directional warp and have some unique characteristics. For example, hexagonal
grids have a simpler and more symmetric nearest neighborhood, which avoids the ambiguities of the rectangular grids.
Movement paths or connectivity, the most compact arrangement of pixels, make hexagonal appear great dominance in
the process of modeling and analysis. The selection of an appropriate grid should be based on the requirements and
objectives of the application. We use rectangular and hexagonal grids respectively for developing city model. At the
same time we make use of remote sensing images and acquire 2002 and 2005 land state of Wuhan. On the base of city
land state in 2002, we make use of CA to simulate reasonable form of city in 2005. Hereby, these results provide a proof
of concept for hexagonal which has great dominance.
|