An activity space-based gravity model for intracity human mobility flows

Abstract

Modeling and estimating urban human mobility flows is significant to multidisciplinary applications such as transportation engineering, urban planning and epidemic prevention. In recent decades, a wide range of trip distribution models for urban human mobility flows have been developed, among which the gravity model is the most classic and widely used. However, like most of the existing trip distribution models, the original gravity model adopts population as the proxy of urban region's attractiveness, which could not fully reflect the urban region's attractiveness as well as its temporal dynamics. Moreover, the original gravity model depends on the selection of proper distance decay functions, which require context-specific adjustable parameters derived from empirical mobility data. Taking advantage of the notion of activity space, we innovatively develop a proxy of urban region's attractiveness, and further propose a new activity space-based gravity (ASG) model. The case study results show that the ASG model significantly outperforms the original gravity model in estimating the overall human mobility flows and the hourly human mobility flows between urban regions. This study contributes to the state-of-the-art understanding of urban human mobility and human mobility flow modeling. The findings also provide insights into transportation management, infrastructure development and urban planning.