Identification and classification of activity centers based on passenger flows data.

by Yu Zhao

Institution: KTH Royal Institute of Technology
Year: 2014
Keywords: Engineering and Technology; Teknik och teknologier
Record ID: 1366677
Full text PDF: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-149467


In the past decades, the spatial structure of metropolitan areas has progressively changed towards a more polycentric structure. Many researchers have studied this polycentric structure in the context of North American and European metropolitans by identifying sub-centers, mainly using two methods which are analyzing the employment or population density or mobility data. In spite of huge effort in identifying sub-centers, fewer studies characterize the identified sub-centers and classify them based on their patterns and features simultaneously. And this research will identify sub-centers and then classify them as well. Following the introduction of polycentricity and a review of previous methodologies for identification and classification of sub-centers, this study introduces two different algorithms: flow-based and distance-based for identifying sub-centers based on passenger flows data at public transport stations. In addition, the study presents the classification process of identified clusters based on time-dependent passenger flows data. Temporal profiles of each cluster are created and used to describe their characteristics, and then classification is conducted based on hierarchical clustering analysis. As a case study, the emergence of polycentric structure in Stockholm County is analyzed using public transport passenger flows at each station including metros, commuter trains, buses and light rails. After comparing results of the two proposed algorithms, the distance-based is chosen for Stockholm case. The identification algorithm yields 17 clusters. These 17 clusters are then classified using three different indicators based on flow data by time intervals. As a result, we have three classification results. Finally, the classification results are analyzed and synthesized by considering the urban environment of clusters and their roles in transport network, providing a comprehensive interpretation of resulted clusters. Clusters are classified into seven more general classes of center, business, residential, hub or combinations of them. The result suggests that each cluster is associated with distinctive functions and they are all active, unlike ‘’sleeping towns’’, however, clusters in the inner city are still able to generate and attract more flows and flows are still more concentrated in the central part, indicating the aim to release pressure from central part by polycentric structure hasn’t been fully achieved yet.