Martin, David, Cockings, Samantha and Leung, Samuel
Population 24/7: building time-specific population grid models.
European Forum for Geostatistics Conference, The Hague, Netherlands,
05 - 09 Oct 2009.
PDF (Paper presented at the European Forum for Geostatistics Conference 2009, The Hague, Netherlands, 5-7 October 2009)
Many areas of social science research and public policy rely on small area geographical representations of population. Studies of disease prevalence, crime rates, exposure to environmental hazards, transportation modelling and the more applied challenges of emergency planning, service delivery and resource allocation rely fundamentally on statistics relating to the distribution of population. Grid-based population models have considerable advantages for population representation, offering more meaningful representation of settlement and neighbourhood pattern, including the geography of unpopulated areas, and providing stability through time. As a result, gridded models have seen extensive use where population must be integrated with environmental phenomena
(Brainard et al., 2002; Mennis, 2003).
Current approaches to spatial population modelling, whether based on conventional small areas or regular grids, rely almost exclusively on residential locations for the geographical referencing of population, drawing heavily on census definitions of the ‘resident population’.
There are however, good conceptual and practical arguments for modelling population at different times, incorporating population movements from seasonal to diurnal timescales, so as to predict population exposure to a specific hazard, or potential customer numbers during a working day. This paper addresses these issues by presenting work in progress on a two-year project to develop 24-hour gridded population models of the UK. The project is based on an existing adaptive kernel density approach for building gridded population models (Martin, 1996), which is now being extended to become a spatiotemporal kernel density estimation method. We begin by briefly reviewing space-time population modelling methods, then move to discuss data sources and our modelling approach and conclude with some illustrative results from our initial work
Actions (login required)