University of Twente Student Theses


Dasymetric estimation of population: A case study of the city of Enschede, The Netherlands

Sharif, Mohammad (2010) Dasymetric estimation of population: A case study of the city of Enschede, The Netherlands.

[img] PDF
Abstract:This research is aimed at disaggregating census population data into the finest possible scale. Census data are usually issued aggregated, because of preventing the disclosure of information of people and also controlling the volume of data. In order to utilize this type of data for specific objectives such as disaster management, urban planning, and disease spread etc., aggregated census data need to become highly detailed and spatially localized by disaggregating them into the smallest unit of measurement. Dasymetric methods, which are a type of areal interpolation techniques for transforming data from one set of spatial unit to another, are employed in this research. They use ancillary data in the process of disaggregating census data which help the internal distribution of populated regions to be inferred. Specifically, the chosen methods are Binary and 3-class dasymetric methods, where the former is quoted as a simple and easy to implement in GIS, while the later is severally cited as a robust method for disaggregating census tracts due to the use of a variety of ancillary classes (e.g. non-urban, low-density and high-density). These methods have been applied to different large study areas with various population density values. It is desired to examine the performance of dasymetric methods for a small and dense region; hence, the city of Enschede with 143 area and average population of 1091 ( ) is chosen. The finest administrative units to be considered as the source and target zones are realized as district and neighbourhood levels, respectively. The above methods are implemented on the census population data by making use of various types of ancillary datasets: first, the buildings‟ footprint land cover maps are utilized in order to assess the accuracy of 2D ancillary information, finding the right scale of these data and to justify previous studies regarding the performance of binary and 3-class dasymetric methods; second, the buildings‟ floor area map is used due to the evaluation of added value(s) of 3D data in the course of population estimation and finally, the postcode 6-digit information as an alternative to the area factor in current dasymetric methods. The assessments of statistical results show the better performance of binary dasymetric method to the 3-class dasymetric method (23% vs. 45% error value) which contradicts previous studies regarding the high usability of 3-class variables. This matter can be referred to the assumption for classifying the study region to three class variables. Moreover, the evaluation of the ancillary datasets indicates the competency of TOP10 dataset in comparison to the other congener areal datasets. By utilizing 3-Dimentional data, which are the buildings heights and their corresponding floor areas, the performance of dasymetric methods have improved where the binary method generated 18% error value and 3-class method produces 23% uncertainty for disaggregating population. Despite previous efforts for finding a proper type of ancillary data, a new version of ancillary dataset –postcode 6-digit units– outweighs (with 17% coefficient variation) those datasets which make use of areal units in the course of population disintegration. Hence, as an answer to the main objective of this research, the census population data from a small study area with high population density can be disaggregated from a large region to various sub-regions by implementing binary dasymetric and utilizing either 3D TOP10 areal map or postcode 6-digit units as ancillary data with an approximate coefficient variation of 17% per person. Keywords: Areal interpolation, dasymetric method, disaggregation census population data
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page