Understanding the dynamics of urban areas of interest through volunteered geographic information
Meixu Chen; Dani Arribas-Bel; Alex Singleton (2019). Journal of Geographical Systems, 21(1), 89-109. DOI: 10.1007/s10109-018-0284-3
Abstract
Obtaining insights about the dynamics of urban structure is crucial to the framing of the context within the smart city. This paper focuses on urban areas of interest (UAOI), a concept that provides functional definitions of a city’s spatial structure. Traditional sources of social data can rarely capture these aspects at scale while spatial information on the city alone does not capture how the population values different parts of the city and in different ways. Hence, we leverage volunteered geographic information (VGI) to overcome some of the limits of traditional sources in providing urban structural and functional insights. We use a special type of VGI—metadata from geotagged Flickr images—to identify UAOIs and exploit their temporal and spatial attributes. To do this, we propose a methodological strategy that combines hierarchical density-based spatial clustering for applications with noise and the ‘α-shape’ algorithm to quantify the dynamics of UAOIs in Inner London for a period 2013–2015 and develop an innovative visualisation of UAOI profiles from which UAOI dynamics can be explored. Our results expand and improve upon the previous literature on this topic and provide a useful reference for urban practitioners who might wish to include more timely information when making decisions.
Extended Summary
This research examines how social media data can reveal the changing popularity and usage patterns of different areas within cities over time. The study develops a new approach to identify and map ‘urban areas of interest’ (UAOIs) - parts of cities that attract significant human activity and attention - using location data from photographs uploaded to Flickr. Traditional methods for understanding urban dynamics often rely on expensive surveys or administrative data that cannot capture real-time changes in how people actually use and value different neighbourhoods. The research addresses this limitation by analysing 1.5 million geotagged Flickr photographs from Inner London between 2013 and 2015. The methodology combines advanced clustering algorithms (HDBSCAN) with boundary-drawing techniques (alpha shapes) to identify areas where many people consistently take photographs, indicating high levels of interest or activity. This approach represents a significant improvement over previous density-based methods by requiring fewer manual parameter adjustments and producing more robust results. The analysis reveals distinct spatial patterns, with stable UAOIs consistently appearing around major tourist attractions like Trafalgar Square, cultural venues such as museums on Museum Lane, business centres including the City of London and Canary Wharf, and transport hubs like King’s Cross and St. Pancras International. The research demonstrates considerable temporal variation in these areas, with some locations showing strong seasonal patterns while others emerge and disappear rapidly, potentially indicating special events or temporary attractions. For example, the study identifies a UAOI that appeared only during January and February 2013 in north Camden, likely corresponding to snowfall events at Hampstead Heath that attracted photographers. The paper introduces innovative spatiotemporal profiles that track how the percentage of each neighbourhood classified as a UAOI changes over time, revealing three distinct patterns: areas with consistent seasonal variation, locations with pronounced summer peaks, and neighbourhoods with irregular or event-driven popularity. These findings have significant implications for urban planning and policy. The methodology provides city planners and local authorities with near real-time insights into changing urban dynamics, enabling more responsive resource allocation for tourism, transport, and public services. Unlike traditional data sources that may take months or years to collect and analyse, social media data offers continuous monitoring of urban popularity patterns. The research demonstrates that volunteered geographic information can complement official statistics to provide a more nuanced understanding of how people actually experience and value different parts of the city, supporting evidence-based decision-making in smart city initiatives.
Key Findings
- Geotagged Flickr data successfully identifies stable urban areas of interest including tourist attractions, cultural venues, and business centres in London.
- Advanced HDBSCAN clustering algorithm outperforms traditional density-based methods for extracting urban areas from social media location data.
- Temporal analysis reveals three distinct patterns: consistent seasonal variation, pronounced summer peaks, and irregular event-driven popularity changes.
- Spatiotemporal profiles enable real-time monitoring of neighbourhood popularity, offering advantages over traditional urban data collection methods.
- The methodology provides urban planners with timely insights for resource allocation and evidence-based decision-making in smart city contexts.
Citation
@article{chen2019understanding,
author = {Meixu Chen; Dani Arribas-Bel; Alex Singleton},
title = {Understanding the dynamics of urban areas of interest through volunteered geographic information},
journal = {Journal of Geographical Systems},
year = {2019},
volume = {21(1)},
pages = {89-109},
doi = {10.1007/s10109-018-0284-3}
}