Simulating the co-emergence of urban spatial structure and commute patterns in an African metropolis: a geospatial agent-based model

Urban spatial structure and mobility patterns co-evolve. A fundamental process that underpins the emergent structure of cities and commuting patterns is location choice with respect to housing and employment. Consequently, location choice models constitute an important component of land use and transport interaction (LUTI) models and urban growth models. This paper documents the development and application of the Metropolitan Location and Mobility Patterns Simulator (METLOMP-SIM). METLOMP-SIM is an integrated geospatial and agent-based model that simulates how urban structure and travel patterns co-emerge, as a function of the location decisions of heterogeneous households and individuals within a heterogeneous, spatially-explicit urban context. After specifying a generic conceptual model that identifies the model’s structure, entities and agents’ decision-making framework, METLOMP-SIM is implemented using the Kumasi Metropolis in Ghana, West Africa as case study. Overall, the implemented model demonstrates that the encoded micro-scale behaviour of household and individual agents are able to mimic some macro-scale residential and job location patterns, and patterns of home-work trip production and attraction that closely match patterns in the case study metropolis, based on the available observational data. METLOMP-SIM responds to the some of the core limitations around model realism in previous agent-based urban location choice modelling efforts. The current model represents property and job market dynamics in both their ‘formal’ and ‘informal’ manifestations, making it sensitive and potentially relevant to urban land use model development and applications in Sub-Saharan African cities, and possibly in other developing countries.