Urban road networks as physical systems: Data, theory, and computation

Road networks are spatial records that resulted from the confluence of many individual and group interactions shaped by the culture, politics, and socio-economic forces in cities. Just like any physical phenomena, science can be used in order to better understand their nature. In understanding systems that are interconnected, interdependent and non-linear, like cities and their road networks, computational tools and concepts from the paradigm of Complexity Science are required. This work aims to show that city road networks are complex systems that exhibit emergent statistical regularities that are robust regardless of geographical location, levels of planning, and temporal age. Despite the statistical regularities, however, properties that are unique to the cities in consideration can still be measured, which are tied to the historical development and future management. Here, the different roles of road networks as complex urban entities are explored and classified into three (3) thematic clusters of investigation: (i) Road networks as morphological descriptors (ii) Road networks as indicators of evolution; and (iii) Road networks as transport channels. For the first part of this dissertation, urban road networks are viewed as morphological descriptors of cities. Here, statistical distributions of the geometrical properties of road networks are provided for two representative datasets under different levels of planning: the cities that make up Metropolitan Manila demonstrate bottom-up self-organized growth, whereas Brasilia and the Australian Capital Territory centered on Canberra represent strict top-down planning. The distribution of segmented areas of the cities exhibits a dual power-law behavior, with the larger areas following the 1.9 scaling exponent seen in other cities, while the smaller areas have a lower exponent of 0.5, which is thought to be related to practical factors. While all cities are shown to promote the creation of straight road segments, planned cities are found to have an abundance of sinuous roads with sinuosity values approaching π. Finally, to capture the nontrivial statistical patterns observed, a simple model based on nearest-neighbor directed branching combined with sectional grid formations is proposed. Secondly, the unique quality of spatial permanence is explored by investigating the growth of the Manila road network for over a century of development. Throughout a city’s growth, there are quantities that (i) changed and (ii) did not change over time. These manifested in both spatial and temporal domains. Through georeferencing and digitization of hand-drawn historical maps, the status of Manila’s road network at various moments in its history is extracted. Visual and metrical assessments indicated significant well-planned moments punctuating the generally self-organized expansion, especially the more recent densification at reclamation areas that coincided with substantial economic growth. Although recent reclamation has greatly increased the statistics of the extremely short and peripheral nodes, the road network of Manila has statistical regularities that are likewise seen in other worldwide road network data sets. Finally, the groups of nodes with the highest closest centralities resemble the net- work’s historical structure, enabling for automated identification of the city’s historical centre. Studies like this one unsheathes important information from these long-term spatial records, which can subsequently be used to design appropriate policy responses to future urbanization. Finally, the work explores the role of the peripheral sections of the road network in the flow of traffic through the system. Road network studies, particularly in densely populated places, must take into consideration not only the topological (i.e. network structure) but also the real physical and geographical limits that impact the system’s efficiency. The collection of shortest pathways through low-betweenness centrality nodes present in the network’s periphery is particularly investigated. Travel between these nodes is characterized by very sinuous pathways, which is to be anticipated given that these nodes represent the network’s most inaccessible places. Shorter pathways are more likely to have a wide range of sinuosity values, whereas longer paths are more straight. To define the network’s most inaccessible regions, a classification of peripheral node inaccessibility is proposed based on topological (network centrality) and spatial (physical dimensions) attributes. Given the current conditions, studies like this one can provide important insights for the administration and enhancement of municipal transportation networks