Measuring urban dynamics in port-cities using an agent-based cellular automata land-use and transport interaction model

Abstract

The urban environment is a complex system where general urban morphology patterns emerge from the local interactions of many agents. Examination on these interactions are crucial to enable successful transport and urban planning. While cellular automata (CA) models are suitable for modelling this self-organisation behaviour, measuring individual interactions from empirical data is still complicated due to their auto-correlation and to path-dependence in urban systems evolution. Due to this, manual methods are more often used in the calibration of urban land-use and transport interaction (LUTI) models based on CA, limiting the results to a few explored settlements. This paper therefore presents a method for applying the gradient-descent algorithm with momentum to calibrate an agent-based LUTI-CA model by representing the model as an optimisation problem. This process allowed efficient calibration for 46 port-related settlements and 10 non-port settlements across Great Britain, enabling cross-sectional analysis of whether observed relationships are unique to a study area due to its path-dependence, or if they have similarities with results from other study areas.

We found unique characteristics of urban dynamics separating port from non-port settlements in smaller urban systems, while the distinctions between the two groups were less prominent in larger settlements. However, there were variations in urban dynamics between larger settlements with respect to the effects of their services and manufacturing activities on other land-uses. The automated calibration of agent-based urban CA models in this paper enabled the quantification of these urban dynamics in multiple settlements, providing the ability to look beyond the uniqueness of individual case studies to find general patterns of interaction. This paper therefore provides improvement in the calibration of LUTI models based on CA and contributes to the better understanding of the dynamics between port and urban systems.