Sustaining Glasgow's Urban Networks: the Link Communities of Complex Urban Systems

As cities grow in population size and became more crowded (UN DESA, 2018), the main future challenges around the world will remain to be accommodating the growing urban population while drastically reducing environmental pressure. Contemporary urban agglomerations (large or small) constantly impose burden on the natural environment by conveying ecosystem services to close and distant places, through coupled human nature [infrastructure] systems (CHANS). Tobler’s first law in geography (1970) that states that “everything is related to everything else, but near things are more related than distant things” is now challenged by globalization.

When this law was first established, the hypothesis referred to geological processes (Campbell and Shin, 2012, p.194) that were predominantly observed in pre-globalized economy, where freight was costly and mainly localized (Zhang et al., 2018). With the recent advances and modernisation made in transport technologies, most of them in the sea and air transportation (Zhang et al., 2018) and the growth of cities in population, natural resources and bi-products now travel great distances to infiltrate cities (Neuman, 2006) and satisfy human demands. Technical modernisation and the global hyperconnectivity of human interactions and trading, in the last thirty years alone resulted with staggering 94 per cent growth of resource extraction and consumption (Giljum et al., 2015). Local geographies (Kennedy, Cuddihy and Engel-Yan, 2007) will remain affected by global urbanisation (Giljum et al., 2015), and as a corollary, the operational inefficiencies of their local infrastructure networks, will contribute even more to the issues of environmental unsustainability on a global scale.

Another challenge for future city-regions is the equity of public infrastructure services and policy creation that promote the same (Neuman and Hull, 2009). Public infrastructure services refer to services provisioned by networked infrastructure, which are subject to both public obligation and market rules. Therefore, their accessibility to all citizens needs to be safeguarded. The disparity of growth between networked infrastructure and socio-economic dynamics affects the sustainable assimilation and equal access to infrastructure in various districts in cities, rendering it as a privilege. Yet, the empirical evidence of whether the place of residence acts as a disadvantage to public service access and use, remains rather scarce (Clifton et al., 2016). The European Union recognized (EU, 2011) the issue of equality in accessibility (i.e. equity) critical for territorial cohesion and sustainable development across districts, municipalities and regions with diverse economic performance.

Territorial cohesion, formally incorporated into the Treaty of Lisbon, now steers the policy frameworks of territorial development within the Union. Subsequently, the European Union developed a policy paradigm guided by equal access (Clifton et al., 2016) to public infrastructure services, considering their accessibility as instrumental aspect in achieving territorial cohesion across and within its member states.

A corollary of increasing the equity to public infrastructure services among growing global population is the potential increase in environmental pressure they can impose, especially if this pressure is not decentralised and surges at unsustainable rate (Neuman, 2006). This danger varies across countries and continents, and is directly linked to the increase of urban population due to; [1] improved quality of life and increased life expectancy and/or [2] urban in-migration of rural population and/or [3] global political or economic immigration.

These three rising urban trends demand new approaches to reimagine planning and design practices that foster infrastructure equity, whilst delivering environmental justice. Therefore, this research explores in depth the nature of growth of networked infrastructure (Graham and Marvin, 2001) as a complex system and its disparity from the socio-economic growth (or decline) of Glasgow and Clyde Valley city-region. The results of this research gain new understanding in the potential of using emerging tools from network science for developing optimization strategy that supports more cecentralized, efficient, fair and (as an outcome) sustainable enlargement of urban infrastructure, to accommodate new and empower current residents of the city.

Applying the novel link clustering community detection algorithm (Ahn et al., 2010) in this thesis I have presented the potential for better understanding the complexity behind the urban system of networked infrastructure, through discovering their overlapping communities. As I will show in the literature review (Chapter 2), the long standing tradition of centralised planning practice relying on zoning and infiltrating infrastructure, left us with urban settlements which are failing to respond to the environmental pressure and the socio-economic inequalities. Building on the myriad of knowledge from planners, geographers, sociologists and computer scientists, I developed a new element (i.e. link communities) within the theory of urban studies that defines cities as complex systems. After, I applied a method borrowed from the study of complex networks to unpack their basic elements. Knowing the link (i.e. functional, or overlapping) communities of metropolitan Glasgow enabled me to evaluate the current level of communities interconnectedness and reveal the gaps as well as the potentials for improving the studied system’s performance.

The complex urban system in metropolitan Glasgow was represented by its networked infrastructure, which essentially was a system of distinct sub-systems, one of them mapped by a physical and the other one by a social graph. The conceptual framework for this methodological approach was formalised from the extensively reviewed literature and methods utilising network science tools to detect community structure in complex networks. The literature review led to constructing a hypothesis claiming that the efficiency of the physical network’s topology is achieved through optimizing the number of nodes with high betweenness centrality, while the efficiency of the logical network’s topology is achieved by optimizing the number of links with high edge betweenness.

The conclusion from the literature review presented through the discourse on to the primal problem in 7.4.1, led to modelling the two network topologies as separate graphs. The bipartite graph of their primal syntax was mirrored to be symmetrical and converted to dual. From the dual syntax I measured the complete accessibility (i.e. betweenness centrality) of the entire area and not only of the streets.

Betweenness centrality of a node measures the number of shortest paths that pass through the node connecting pairs of nodes. The betweenness centrality is same as the integration of streets in space syntax, where the streets are analysed in their dual syntax representation. Street integration is the number of intersections the street shares with other streets and a high value means high accessibility.

Edges with high betweenness are shared between strong communities. Based on the theoretical underpinnings of the network’s modularity and community structure analysed herein, it can be concluded that a complex network that is both robust and efficient (and in urban planning terminology ‘sustainable’) is consisted of numerous strong communities connected with each other by optimal number of links with high edge betweenness. To get this insight, the study detected the edge cut-set and vertex cut-set of the complex network. The outcome was a statistical model developed in the open source software R (Ihaka and Gentleman, 1996). The model empirical detects the network’s overlapping communities, determining the current sustainability of its physical and logical topologies.

Initially, an assumption was that the number of communities within the infrastructure (physical) network layer were different from the one in the logical. They were detected using the Louvain method that performs graph partitioning on the hierarchical streets structure. Further, the number of communities in the relational network layer (i.e. accessibility to locations) was detected based on the OD accessibility matrix established from the functional dependency between the household locations and predefined points of interest. The communities from the graph of the ‘relational layer' were discovered with the single-link hierarchical clustering algorithm. The number of communities observed in the physical and the logical topologies of the eight shires significantly deviated.