Estimating Functional Urban and Regional Boundaries from Road Networks: The Case of South Korea

Abstract

How can we identify functional regions using alternative but more widely available data sources? For local economic and policy analysis, researchers often need to redefine urban or regional boundaries reflecting functional connectivity and complementarity, which do not necessarily respect administrative boundaries. Conventional methods employed for this purpose are based on detailed cross-municipality traffic and trade flows data, but application of the methods is highly constrained by data availability. In the context of developing countries, lacking quality subnational data sets makes it hard for the application, although rapid urbanization and the imminent need for policy intervention increase demand for urban analysis. This study is motivated to fill the gap, and we introduce an alternative method subject to low data requirement, taking South Korea as an example. Our case selection is for the validation of our analysis results (vs. the results from conventional methods), although South Korea is far from a developing country in any standard. The method we employ is a percolation model in physics, which has also been applied for urban analysis. The only data input required for our model is national level road network data, which is widely available through various official and open sources. In this model, we focus on road junctions for connectivity, and convert the network data into an extensive origin-destination table. Then, we test a wide range of cross-junction distance thresholds to analyze system-level hierarchies and connectivity. From this analysis, we ultimately aim to demonstrate road network data can substitute for subnational flows data in drawing functional urban and regional boundaries. Our results show that distance thresholds of 740 meter (~0.46 mile), 1.64 to 1.66 km (~1.02 miles), and 3.08 to 3.12 km (~1.93 miles) have particular importance in defining functional urban areas in Korea’s context. When a distance threshold of 740 meter is tested, the model detects major cities and core urban areas. Applying a threshold of 1.64 to 1.66 km identifies Korea’s major urban agglomerations or metropolitan regions, which are functionally well connected. Finally, a threshold of 3.08 to 3.12 km helps distinguish a few areas presenting exceptionally weak connectivity (and thus highly isolated) from other parts of the national urban system, which then can be considered with priority for publicly initiated transport infrastructure projects.