Exploring future landscape changes for polycentric urbanization using cellular automata calibrated with radiation model

Abstract

The use of cellular automata (CA) is essential for exploring future urban growth scenarios in spatial planning. However, modeling polycentric urbanization processes with CA is still challenging due to the presence of spatial spillover effects arising from spatial interactions between different regions. This study proposes a hybrid framework that addresses the spatial spillover effect that emerges from multi-centers by coupling a radiation model (RM) and Markov chain (MC) with CA to simulate polycentric urbanization processes. The simulation capabilities of the RM-MC-CA framework were evaluated and validated by simulating Guangzhou’s actual urban growth from 2000 to 2020, and the future urban growth scenarios of 2035 and 2050 were simulated with this coupled model. Results showed this framework provides a spatio-temporal diffusion process consistent with the cooperative mechanism of urbanization from monocentric to polycentric. In terms of simulation accuracy, the proposed RM-MC-CA framework demonstrated the most promising performance compared to MC-CA, GM-MC-CA, and PLUS. Compared to classical MC-CA, the framework improved the Kappa, FOM, and Precision metrics by 0.54%, 3.93%, and 2.38%, respectively. These results indicated that incorporating spatial spillover processes into a CA model can enhance its ability to simulate polycentric patterns that promote high-quality urban development.