Urban Population Distribution Models and Service Accessibility Estimation

Mitchel Langford, Gary Higgs, Jonathan Radcliffe, Sean White

Research output: Contribution to journalArticlepeer-review


This paper examines the influence of alternative population distribution models on GIS-based spatial accessibility analyses using the two-step Floating Catchment Analysis technique. Two population models were tested: the de facto standard of even-distribution within census tracts and a dasymetric-based approach. The latter builds on previous research through the use of a novel methodology that integrates raster map data with a recently built mailing information database in order to enhance the precision with which residential areas are identified. Analysis was conducted for a case study area (Cardiff, South Wales) in order to examine variations in accessibility to a number of public services in the city. The dasymetric model showed a general tendency to report lower accessibility scores, but detailed patterns depended on local factors and, to some extent, on modelling assumptions and methodological issues. A paired T-Test analysis demonstrated that significant differences in outcomes were dependant on the population model adopted. Accessibility-based measures are increasingly being incorporated into deprivation indicators and the paper concludes by suggesting that, if such analysis is to inform urban planning, local service provision and the spatial allocation of financial resources, greater attention needs to be given to the method of population representation utilised in such models.
Original languageEnglish
Pages (from-to)66 - 80
Number of pages14
JournalComputers Environment and Urban Systems
Issue number1
Publication statusPublished - 1 Jan 2008


  • population estimation models
  • dasymetric techniques
  • accessibility measures
  • floating catchment analysis
  • GIS


Dive into the research topics of 'Urban Population Distribution Models and Service Accessibility Estimation'. Together they form a unique fingerprint.

Cite this