Markov–Cellular Automata in Geospatial Analysis

Abstract

Spatial simulation models are indispensable for modelling land use/cover changes (Wu and Webster 1998; Messina and Walsh 2001; Soares-Filho et al. 2002), deforestation and land degradation (Lambin 1994; Lambin 1997; Etter et al. 2006; Moreno et al. 2007), urban growth (Clarke et al. 1997; Couclelis 1989; Cheng and Masser 2004; Gar-On Yeh and Li 2009), climate change (Dale 1997) and hydrology (Matheussen et al. 2000). For land use/cover change studies, spatial simulation models are critical for understanding the driving forces of change, as well as to produce “what if” scenarios that can be used to gain insights into future land use/cover changes (Pijanowski et al. 2002; Eastman et al. 2005; Torrens 2006). Recently, the knowledge domain of spatial simulation modelling has advanced owing to the rapid developments in computer technology, coupled with the decrease in the cost of computer hardware. In addition, developments in geospatial, natural and social sciences concerning bottom-up, dynamic and flexible self-organising modelling systems, complemented by theories that emphasize the way in which decisions made locally give rise to global patterns, have enriched spatial simulation models (Tobler 1979; Wolfram 1984; Couclelis 1985; Engelen 1988; Wu and Webster 1998; Batty 1998). To date, numerous spatial simulation models have been developed and applied, particularly for land use/cover modelling (Clarke et al. 1997; Kaimowitz and Angelsen 1998; Messina and Walsh 2001; Soares-Filho et al. 2002; Walsh et al. 2006).