Version 2 2019-12-01, 23:08Version 2 2019-12-01, 23:08
Version 1 2019-09-16, 07:27Version 1 2019-09-16, 07:27
conference contribution
posted on 2019-12-01, 23:08authored byYi Lu, Shawn Laffan
Cellular Automata (CA) have formed an important part of the geocomputational and spatial analysis toolbox for three decades. Some of the most important components of CA are the transition rules to determine the changes in cell states across iterations. These are applied at the micro-sale, but lead to emergent patterns at the macro-scale. An important consideration for transition rules is the spatial extent over which they will be valid. One does not expect the same rules to apply equally across an entire region, yet most CA implementations only support one set of transition rules that are applied everywhere. In this paper, a vector CA model with spatially partitioned transition rules is proposed to identify the expansion of urban residential areas across heterogeneous study area. Initial experiments using two sub-regions of Ipswich, Queensland, Australia, indicate that the spatially partitioned approach can improve the accuracy of vector CA.