Comparative Modeling of Neighborhood Configuration Effects on Parcel‐Level Urban Growth Simulation Using Vector‐Based Cellular Automata

ABSTRACT Neighborhood configuration is a key determinant of vector‐based cellular automata (VCA) model behavior, particularly for accurate parcel‐level urban growth simulations. Buffer‐based and topology‐based configurations differ in spatial metrics and anisotropy, resulting in distinct spatial domains and local interaction patterns that can significantly affect simulation outcomes. However, the comparative performance and suitability of these configurations remain underexplored. This study evaluates these two methodological approaches by combining different scope definitions and neighborhood rule algorithms within a VCA framework for simulating urban growth in five subregions of Jiangyin City, China. Key findings include: (1) While buffer‐based configurations achieve similar optimal accuracy to topology‐based ones, they exhibit greater fluctuations in Figure‐of‐Merit (FoM) trends as neighborhood size increases. (2) Neighborhood rules under both configurations exert a stronger influence on simulation accuracy, with their impact on transition probability distributions following a consistent pattern across districts. Generally, the polynomial decay function outperforms other approaches in achieving superior accuracy. (3) Neighborhood scope definitions play a more pivotal role in shaping urban expansion patterns. Buffer‐based configurations are more sensitive to infilling expansion, promoting compact urban growth and suiting areas focused on internal land‐use development. In contrast, topology‐based configurations effectively capture edge expansion, producing dispersed urban forms and fitting regions with significant external expansion.