Stochastic model-based methods for handling uncertainty in areal interpolation

Alistair Geddes, David A. Elston, Matthew Hodgson, Richard Birnie

    Research output: Contribution to journalArticlepeer-review

    1 Citation (Scopus)

    Abstract

    Handling of uncertainty in the estimation of values from source areas to target areas poses a challenge in areal interpolation research. Stochastic model-based methods offer a basis for incorporating such uncertainty, but to date they have not been widely adopted by the GIS community. In this article, we propose one use of such methods based in the problem of interpolating count data from a source set of zones (parishes) to a more widely used target zone geography (postcode sectors). The model developed also uses ancillary statistical count data for a third set of areas nested within both source and target zones. The interpolation procedure was implemented within a Bayesian statistical framework using Markov chain Monte Carlo methods, enabling us to take account of all sources of uncertainty included in the model. Distributions of estimated values at the target zone level are presented using both summary statistics and as individual realisations selected to illustrate the degree of uncertainty in the interpolation results. We aim to describe the use of such stochastic approaches in an accessible way and to highlight the need for quantifying estimation uncertainty arising in areal interpolation, especially given the implications arising when interpolated values are used in subsequent analyses of relationships.
    Original languageEnglish
    Pages (from-to)785-803
    Number of pages19
    JournalInternational Journal of Geographical Information Science
    Volume27
    Issue number4
    Early online date2 Nov 2012
    DOIs
    Publication statusPublished - 2013

    Keywords

    • areal interpolation
    • census data
    • small area geography
    • data models

    Fingerprint Dive into the research topics of 'Stochastic model-based methods for handling uncertainty in areal interpolation'. Together they form a unique fingerprint.

    Cite this