Home > Research > Publications & Outputs > A Comparison of Small-Area Population Estimatio...

Associated organisational unit

Links

Text available via DOI:

View graph of relations

A Comparison of Small-Area Population Estimation Techniques Using Built-Area and Height Data, Riyadh, Saudi Arabia

Research output: Contribution to journalJournal article

Published
Close
<mark>Journal publication date</mark>05/2016
<mark>Journal</mark>IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Issue number5
Volume9
Number of pages11
Pages (from-to)1959-1969
Publication StatusPublished
Early online date17/12/14
<mark>Original language</mark>English

Abstract

Small-area population estimation is important for many applications. This paper explores the usefulness of Landsat ETM + data, remotely sensed height data, census population, and dwelling unit data to provide small-area population estimates. Riyadh, Saudi Arabia was selected as a suitable area to test a set of methods for population downscaling. Two broad approaches were applied: 1) statistical modeling and 2) areal interpolation. With regard to statistical modeling, regression through the origin was used to model the relationship between density of dwelling units and built area proportion at the block level and the coefficients were used to downscale the density of dwelling units to the parcel level. Areal interpolation with ancillary data (dasymetric mapping) used the block and parcel levels as the source and target zones, respectively. The population distribution was then estimated based on the average population per dwelling unit. Eight models were developed and tested. A conventional regression model, using only built area as a covariate, was used as a benchmark and compared with the more sophisticated models. Remotely sensed height data were used to: 1) create number of floors; 2) classify the built area into different categories; and 3) increase the user's accuracy of the built area. It was found that remotely sensed height data were useful to explain the variation in the dependent variable across the selected study area. Dasymetric mapping was applied in order to provide a comparison, while acknowledging that the method uses population data not available in the regression approach.