Tropical Geography ›› 2020, Vol. 40 ›› Issue (2): 346-356.doi: 10.13284/j.cnki.rddl.003220

Previous Articles     Next Articles

Accuracy Comparison of Four Gridded Population Datasets in Guangdong Province, China

Lin Danchun1, Tan Min2, Liu Kai1(), Liu Lin3, Zhu Yuanhui3   

  1. 1.School of Geography and Planning, Sun Yat-sen University, Guangdong Provincial Engineering Research Center for Public Security and Disaster, Guangdong Key Laboratory for Urbanization and Geo-Simulation, Guangzhou 510275, China
    2.Guangzhou Urban Planning & Design Studio, Guangzhou 510030
    3.Center of Geographic Information Analysis for Public Security, School of Geographic Sciences, Guangzhou University, Guangzhou 510006, China
  • Received:2019-06-17 Revised:2020-01-06 Online:2020-03-10 Published:2020-05-15
  • Contact: Liu Kai E-mail:liuk6@mail.sysu.edu.cn

Abstract:

The spatial distribution of population is foundational information for policy making, disaster prevention, economic development, environmental protection, and natural sciences or socioeconomic research. Analyzing different gridded population products is essential for learning their characteristics and proper application. Considering timeliness of data and convenience of access, this study compared the spatial consistency of WorldPop, GPW v4 (Gridded Population of World, version 4) and two types of Gridded Population of China datasets in Guangdong Province, China, a province with significant population density differences. The sixth census data in 2010 were divided into high, middle, and low density groups and used as reference data to validate these four datasets on numerical and spatial distribution of error. Moreover, the study used average absolute error, average relative error, root mean square error, correlation coefficient, four indicators, ratio error statistics, and the Taylor diagram for quantitative evaluation. Then, the spatial distribution of the ratio error and the accuracy of the four datasets in different population density regions were analyzed. Finally, an evaluation of error sources and applications from two aspects, including methods for spatializing census data and variables selection of spatial model, was performed. The findings of the study are as follows: 1) These four population datasets have similar spatial distribution trends, and the population is concentrated in the Pearl River Delta regions such as Guangzhou, Shenzhen, Dongguan, Foshan, Zhongshan and Chaoshan regions such as Shantou and Jieyang. The WorldPop dataset had the highest relative accuracy in the whole Guangdong Province, as well as in the high population density area, whereas the GPW v4 dataset performed better than WorldPop in the middle and low population density areas; however, GPW v4 had a shortcoming in depicting population distribution beneath the township level because of the areal-weighting method used. The two types of Gridded Population of China datasets had relatively lower accuracy than the other two datasets. 2) The accuracy of the four population datasets in Guangdong Province was limited mainly by the spatial method based on the regression algorithm, the weighting method, the selection of model variables, and also related to the geographical and social environment of Guangdong Province. 3) The WorldPop dataset is suitable for detailed research in areas with medium and high population density, GPW v4 is suitable for long-time, minimal research units larger than townships, and the first type of Gridded Population of China dataset is suitable for study that requires consideration of the distribution and spatial heterogeneity of population beneath the township level. The second type of Gridded Population of China dataset is suitable for long-time study that requires consideration of the details of the population distribution and changes of spatial pattern. This study provides important basic information for research and applications using gridded population datasets.

Key words: WorldPop, GPW v4, Gridded Population of China, accuracy assessment, Guangdong Province

CLC Number: 

  • C924.2