Flow matters : using space of flows to understand urban dynamics in complex city

Abstract

As technology has become an indispensable part of city, human and land have been connected with unprecedented frequency and resolution. The connectiveness captured by urban flows fuels new perspectives of quantifying how people behave and how city work. The present thesis investigates the modelling and quantification of multi-dimensional urban flows through numerical and multilayer network approach, and demonstrates the extracted knowledge can be used for enhancing comprehension of urban dynamics through flow-based paradigm. Leveraging rich data captured from mobile devices, traditional public transport, shared mobility services, and volunteer geographic information, the framework of defining, extracting, integrating, and modelling urban flows in unified model is demonstrated. Bridging the latest theory and methods of network science, a shared mobility multiplex network and a temporal multiplex network are constructed, from which multilayer statistical feature and community structure results constitute to the limited knowledge on how new transport mode influence flow patterns of the traditional, and how multi-flow-induced urban structure may change over transport modes and time. Along with chapters, a set of explicit metrics are developed and discussed for quantifying flow patterns. Some of these, such as multilayer degree and multiplex PageRank, the latest methods developed by network scientist to tackle the drawbacks of single-layer network analysis, are adopted on the urban flow models in this thesis. Moreover, diversity of spatial interaction (DSI) is a new metric defined and developed for the first time on quantifying diversity from flow data. We construct DSI by integrating multiple aspects of activity diversity being separately studied before. Its effectiveness is validated and further explored by intersecting with land characteristics data, offering powerful insights on revealing the positive and negative implications of flows for urban places. For urban vitality evaluation, latest metrics such as ridership variations, are integrated with the proposed flow diversity to develop a comprehensive framework of urban diversity, based on which the multiscale spatially varying relationship between diversity metrics and vitality is inferred and discussed. The use of spatial coefficients for profiling unique urban context is presented at the end. As an interdisciplinary body of work conducted by a geography reseaecher, this thesis puts forward evidence on importance and effectiveness of multi-flows data analytics for fundamental questions in geography such as human-land relationship, and for latest topics in urban studies such as urban vitality and dynamic structure, substantiating the merits of numerical and multilayer network approach to urban flows in complex city.