Dasymetric Estimation of Census Population Density: A Geostatistical Approach

17th International Conference on Geoinformatics, 12-14 August 2009

Zongyi He, Liwei Liu, Aihua Hu, Lu Xu

“The population density in choropleth maps is the average density calculated based on the polygon statistical units. It does not show the difference inside the polygon area and cannot reflect the actual population distribution. In order to obtain a real population distribution, it is necessary to spatially refine the statistical areal population data. Using population distribution model or surface interpolation can realize the spatialization of urban population. But these two methods are isolated in research, not closely related to each other. This paper studied the dasymetric mapping of urban population based on geostatistics with the combined use of these two methods. First, the variation function of geostatistics was introduced into the analysis of population spatial distribution model. The population distribution in a space presented a definite structure feature. The variation function curves were fitted through the population density at the sampling points with different intervals h in space, so as to quantitatively describe the change of the population spatial distribution. With some cases, the paper showed how the parameters of variation function could be utilized to analyze the spatial mode of population distribution. Then, a method for dasymetric mapping of urban population was put forward based on indicator Kriging’s interpolation. The theoretical model of the variation function could reflect the degree of spatial relativity of urban population distribution, and the Indicator Kriging can be used to carry out interpolation with sample weight coefficients derived by the theoretical model of the variation function. This was an overall modeling with partial interpolation approach, which could effectively control the influence of particular values, so as to improve the accuracy of urban population estimation. Population statistic data used in the case was acquired from the fifth census in Zhengzhou, China. Considering the large volume of the data, statistic uni- t in the study is confined to the street office level. The study area is the metropolitan area in City of Zhengzhou. Spatial database was built using ArcGIS.The case studied here indicated that the Indicator Kriging performs well in the interpolation of population data.”