The effect of land use categories resolution on geography cellular automata model

Cellular automata model is a useful tool to understand the urbanization process. But nontrivial questions have to be answered about the choice of model parameters such as cell state resolution. The objective of this study is to evaluate the sensitivity of cellular automata to cell state resolution when using CA in a geographic context. The study chooses CA-Markov model to simulate urban expansion in the Huadu district (Guangzhou city, china). Two cell state resolutions are designed (urban/non-urban land use categories, multiple land use categories). Then the sensitivity of cell state resolution is undertaken by evaluating the resulting simulation accuracy of the model and the morphology of urban area. Results show that cell state resolution have a significant impact on simulation result in terms of both urban area and spatial pattern. The model outputs using two different cell state resolutions have great difference in term of both the resulting simulation accuracy of the model and the morphology of urban area. The urban area and patch numbers of model output produced by multiple land use dataset are higher than the model output with urban/non-urban land use dataset. While the urban growth rate and mean annual increment urban area are lower than the latter. Therefore, we shouldn't ignore the effect of modifying land use categories resolution on model output when using CA model to simulate urban expansion.