Modelling urban growth: Integrating both physical and human dimensions in developing world context

Abstract

Listen

Translate article

Designing effective policy to ameliorate the adverse impacts of urbanisation and urban sprawl demands a comprehensive understanding of the development process. While the urban modelling literature have extensively explored the impact of physical factors such as current urban land cover, slope, elevation and accessibility to roads on new urban growth, little is known about how the decision of urban development actors such as private developers impact on urban growth. This study aims to redress this deficit in knowledge on urban growth through modelling both the physical factors and decisions of development actors in a developing world context. The case study context is the Jakarta Metropolitan Area (JMA), Indonesia, the world’s third largest megacity. The JMA represents an ideal case area given the predominance of private land developers on the provision of its urban areas. The study sets three objectives: 1) to understand the urban growth patterns in JMA over the past two decades and identify the leading factors impacting urban growth in the region; 2) to model the land selection behaviours of private land developers through a cellular automata and agent-based urban model; and 3) to simulate the heterogeneous characteristics of developer agents and how their location selections impact on urban development in JMA.The first objective investigates the extent and spatial patterns of urban land cover in JMA. Through a spectral land cover classification method applied to Landsat satellite imagery a proliferation of urban areas on the outskirts of Jakarta with growth rates of 50 square kilometres per annum were revealed. The emergence of urban areas along the east-west and southern corridors of the JMA is in large part the result of transforming from vegetation land cover. The characteristics of urban spatial patterns and their change over time were measured using landscape metrics wherein the results show that in JMA’s core, urban development evolved into a single connected urban land cover, while a more fragmented urban pattern was observed on the outskirts of the JMA’s core. Taken together the findings point to different urban development processes forming these patterns. The retrieved land cover map provides the base map in the urban model that constitutes the second and third objective.The second objective reveals how developers’ capital levels impact the emergence of urban areas. A combination of cellular automata (CA) and agent-based model (ABM) is constructed with CA representing the physical land factors and ABM representing the decision behaviours of developer agents under various capital investment scenarios. The ABM embeds microeconomic concept governed by the cost-profit assessment in a spatially explicit context. The modelling results show that developers have the capacity to influence urban development through a diverse set of capabilities. In the context of JMA, the most profitable locations for new urban development are within 10 to 20 kilometres from the Jakarta CBD, which could only be realized by developers with a minimum initial capital of Rp. 5 trillion (AUD$ 500 or US$ 375 million in 2016) and can also access the maximum lending at 75 percent scale.The third objective investigates the consequences of developers’ heterogeneous characteristics on the development of urban areas through simulating various scenarios in the urban land market. Three types of developer agents were considered based on their capital possession levels, categorised as large, medium and small scale. Their land development strategies including expected level of profit margins, land searching territory and perceived land values on urban areas were simulated individually in the CA-ABM framework before allowing all agents to operate concomitantly within a single land market. The results reveal that different types of developers lead to different locational choices and hence resulting in different urban forms. Developers with larger scale of capital possession tend to target at larger and more profitable areas closer to toll roads and to the CBD; their selections of urban locales are more predictable than that of the medium and small scale developers, which form smaller size urban development that are more fragmented and less predictable spatially.In summary, this study provides both conceptual and methodological contributions particularly in the provision of new urban areas from the perspective of private urban developers. The models supply a proof-of-theory about developers’ underlying reasoning for selecting urban land and help to explain why particular patterns of urban form emerges on certain locations. Through the models’ dynamic visualization and disaggregated analytical unit (i.e. a spatial resolution of 300 meters that represent the physical land factors and individual units to denote developers), the model elucidates the complex interactions between different types of developers and the resulting urban areas. Taken together, the results offer more enriched information for the decision makers to formulate the spatial planning that better accommodates the provision of urban areas desired by the private land developers.