Skip to main navigationSkip to main content
The University of Southampton
Geography and Environmental Science

Spatial Population Analysis and Modelling theme

Our work focuses on a range of innovative models for representation and analysis of spatial population distributions, including methodological development and application of automated zone design, spatiotemporal population modelling and synthetic estimation.

Automated zone design concerns the creation of geographical boundary systems optimised for specific purposes, such as the publication of census data. Prof Martin began working with the Office for National Statistics (ONS) on the concept of output areas for the 2001 census, providing a prototype software tool which was subsequently developed and implemented by ONS. The output areas were an entirely new set of small geographical units covering England and Wales, designed to have standardized statistical characteristics. The design algorithm was subsequently re-applied at larger scales to produce two levels of super output areas and is now known as the Neighbourhood Statistics geography hierarchy and reflected in the ONS geography policy. The original work has been taken forward together with Dr Cockings, Andrew Harfoot and Duncan Hornby, including research applications of automated zone design and development of census output areas and new workplace zones for 2011 output areas. The team’s AZTool software is freely available for download and has been increasingly used by research groups and national statistical organizations internationally such as Australian Bureau of Statistics and Statistics New Zealand. Current work includes application of automated zone design to data collection geographies and development necessary for these systems to continue to be used by ONS as the basis for 2021 Census output areas and workplace zones.

Spatiotemporal population modelling is concerned with the challenge of estimating spatial population distributions that are appropriate for specific times and dates. Conventional population mapping uses data sources such as censuses which generally record all population at a place of residence, with some additional information on places of work. In reality, at any one point in time the population is spread across a diverse and continually changing range of locations, with the result that a conventional residence-based population distribution is of little value for applications such as emergency planning or vulnerability mapping. As an enormously greater range of population data sources become available, it is increasingly possible to consider the generation of time-specific population models and this challenge is the core focus of the work of a group comprising Prof Martin, Dr Cockings, Hugh Darrah and postgraduate students Alan Smith, Becky Martin and Rebecca King. An initial ESRC-funded Population 24/7 project developed the SurfaceBuilder 24/7 software and has been taken forward using new and emerging forms of data under the Pop247NRT project.

The group’s global mapping work, combined in WorldPop and synthetic population estimation are strongly supportive of our population health research. Synthetic multilevel estimation concerns the application of model-based techniques to data from surveys or alternative sources, together with information on covariates, in order to produce estimates of unmeasured characteristics of the populations of small areas which take account of locally varying hierarchical relationships. Under the supervision of Prof Graham Moon, this work has been taken forward by Pierre Dutey-Magni and as part of an ESRC-funded project “Estimating Census Health Geographies” with Dr Twigg and Dr Taylor.

Reflecting the impact of our work, members of the research theme hold key leadership and advisory positions. Prof David Martin is currently co-director of the ESRC’s UK Data Service and National Centre for Research Methods. Both Profs Martin and Moon were co-investigators in the Administrative Data Research Centre for England (2013-2018).

Train station
Population in time and space
Share this research area Share this on Facebook Share this on Twitter Share this on Weibo
Privacy Settings