Sensing Spatiotemporal Patterns in Urban Areas: Analytics and Visualizations Using the Integrated Multimedia City Data Platform

Having the ability to detect emerging patterns in cities is crucial for efficient management of urban resources. Patterns that are useful in identifying and addressing future resource needs include spatial changes in urban form and structure as well as temporal changes in human concentrations and activity patterns during the course of a day. Other patterns of interest are characteristics of local populations in dynamically changing neighbourhoods and social and functional spaces. In this paper, we use the Integrated Multimedia City Data (iMCD) platform, which brings together multiple strands of structured and unstructured data, to examine such trends in the Greater Glasgow region. We present an approach to, first, understand spatial and time-dependent changes that capture the flow of resources needed to meet demands of residents and businesses at different times and locations, and second, generate hypotheses regarding urban engagement, activity patterns and travel behaviour. We use social media data, GPS trajectories, and background data from the UK Population Census for this purpose. The approach identifies the 'roughness' in activity patterns across the urban space that are indicative of different concentrations of social and functional activities. When the time dimension is added to the mix, we are able to uncover time-varying transitions from one type of use pattern to another in different parts of the region. Such transitions, particularly in mixed-use areas, allow early detection of points of excess urban metabolism, with implications for traffic congestion, waste production, energy and other resource consumption patterns. Finally, the ability to detect what citizens talk about socially may provide a way to understand whether or not the language patterns detected in different parts of the city reflect underlying uses and concerns. A preliminary step to evaluate this idea is explored by extracting context-awareness and semantic enrichment to socially-generated data.