Cellular Automata Modelling Approach for Urban Growth

Abstract

The urban expansion is always an inevitable issue in our human history and has become more intensive during a past several decades with explosive population growth of the world. The urban expansion is sometimes praised as a result of economic development, but at the same time, it might induce serious problems such as traffic jams, soaring price of real estate, trash problems, and shortage of natural resources. Thus, it is one of serious concerns many countries are facing. Therefore, a lot of modeling techniques have been introduced, discussed and developed for this problem (Batty, 1971; Forrester, 1969, Makse et al., 1995). Among them, cellular automaton (CA) is one of most dominant techniques, because both of the urban expansion and the CA inherently and similarly include complexity in their behavior. CA was initially introduced by John von Neumann and Stanislaw Ulam as a simple model for biological process such as self-production (Burks, 1971). CA may express any non-ABSTRACT Cellular Automaton (CA) consists of a regular grid cells of which states change according to simple repetitive rules regulated by their contiguous and adjacent cells, which often expresses an unexpected complexity. Thus, CA is one of the major techniques to imitate and/or assess complex behaviors of natural systems. CA can be applied to physical and biological phenomena, such as turbulence in fluid, patterns of biological growth, and wildfire, and also some human-induced phenomena such as urban growth that is the main target of this review. In 1970s, cellular approach was initially adopted in geography, showing the clue to the urban growth application. To overcome the limitation or constraints the conventional standard cell-space models inherently include, alternative formulations were theoretically proposed in 1980s. And the pioneering work applied to realistic cities was conducted in 1990s. Subsequently, numerous models have been presented by relaxing original rules to express reality and by introducing some additional techniques such as geographical information system and system dynamics, thus far. This paper reviews 87 published cellular automata studies on urban growth simulations, urban land use change assessments, urban planning and related information from 18 countries, and examines the characteristics of each relaxation method. In addition, the scale problems are frequently discussed in the validation of the CA model is addressed.

Explaining a century of Swiss regional development by deep learning and SHAP values

“…While LUTI and agent-based models are usually applied at local scales with detailed datasets, CA models can be applied at regional scales and use more general datasets such as terrain, land-use maps, transport networks, or locations instead of actual drivers for urban/regional growth (Li and Gong, 2016). The basis for these models is the definition of grid cells that evolve according to certain transition rules based on the states of adjacent cells (Otgonbayar et al, 2018).…”