An agent- and GIS-based virtual city creator: A case study of Beijing, China

Abstract

Many agent-based integrated urban models have been developed to investigate urban issues, considering the dynamics and feedbacks in complex urban systems. The lack of disaggregate data, however, has become one of the main barriers to the application of these models, though a number of data synthesis methods have been applied. To generate a complete dataset that contains full disaggregate input data for model initialization, this paper develops a virtual city creator as a key component of an agent-based land-use and transport model, SelfSim. The creator is a set of disaggregate data synthesis methods, including a genetic algorithm (GA)-based population synthesizer, a transport facility synthesizer, an activity facility synthesizer and a daily plan generator, which use the household travel survey data as the main input. Finally, the capital of China, Beijing, was used as a case study. The creator was applied to generate an agent- and Geographic Information System (GIS)-based virtual Beijing containing individuals, households, transport and activity facilities, as well as their attributes and linkages.

Experimenting with Agent-Based Model Simulation Tools

“…GIS-based systems use multiple spatial data models for representing and storing information about phenomena with spatial location and extent [21,22]. In the ABM context, GIS data are particularly crucial, especially for the extensive use of simulation in transportation, urban mobility [30,31], epidemic, and pandemic models [32,33] that incorporate spatial and network topologies to model, for instance, people's realistic activity [34]. Additionally, ABM should transparently handle multiple types of agents and fields within the same simulation without the developer's involvement.…”