Modeling Connections in Urban Networks: Distance, Gravity, Radiation (the Case of the USA)

The network revolution has breathed new life into the study of spatial nodal structures. While data on the centers and their characteristics are often available, data on connections are still relatively rare, especially when it is necessary to consider network dynamics over a long period of time. This makes it relevant to consider various approaches to modeling connections between points with known attributes. Such approaches can be grouped into three: nearest neighbor, gravity, and radiation models. The first two have become quite widespread in human geography, spatial economics and regional studies, while radiation models were proposed just over a decade ago and remain little known. We understand the volume of connections in urban networks as the potential volume of diverse flows between them. We model the US urban network for 2010 in three variants. Then we focus on geographical and structural differences and provide some quantitative comparisons. The nearest neighbor method makes it easy to identify the main dividing lines in the urban network and show the main urbanization gradients, but does not allow one to evaluate the strength of connections and misses their multiplicity. The gravity model depends on the calibration of the distance decay parameter and prioritizes links between nearby large cities as well as upward links between medium-sized cities and their nearest large neighbor, emphasizing the hierarchy of urban networks. The radiation model gives more weight to links between nearby cities of similar sizes and shows a smaller range of link strengths, emphasizing the “flat” nature of the network ontology.

urban studies, USA, urban systems, network models, cardinal places