PROGRESS IN GEOGRAPHY ›› 2016, Vol. 35 ›› Issue (11): 1317-1328.

• Orginal Article •

### A method for demographic data spatialization based on residential space attributes

Nan DONG1,2(), Xiaohuan YANG1,2,*(), Hongyan CAI1

1. 1. State Key Laboratory of Resources and Environmental Information System, Institute of GeographicSciences and Natural Resources Research, CAS, Beijing 100101, China
2. University of Chinese Academy of Sciences, Beijing 100049, China
• Online:2016-11-25 Published:2016-11-25
• Contact: Xiaohuan YANG E-mail:dongnan67@126.com;yangxh@igsnrr.ac.cn
• Supported by:
National Natural Science Foundation of China, No.41271173;National Science and Technology Support Program, No.2012BAI32B06

Abstract:

Fine spatial scale population distribution has increasingly become a research hotspot yet a difficult question in the field of population geography. It has practical values in application and scientific significance for relevant research, such as disaster risk and impact assessment, resource allocation, and construction of smart cities. Residential building scale is considered an important part of fine spatial scales for population distribution. Research on the spatialization of population data at this scale has increasingly attracted academic attention. In this study, a population distribution vector data set at the residential building scale was established for six residential committees in Xuanzhou District, Anhui Province in 2015 based on residential space attribute data. Data used in the study include residential building patch area, percentage of housing area within residential building patches, building floor number, and public area rate. The method takes residential space attributes as variables for spatializing population data and treats residential building patches as population distribution location in geographical space with town boundary and town-level demographic data as controls. The spatialization method used in this study reveals detailed information about population distribution in urban areas. The results show that: (1) The population distribution data, obtained by using residential space attributes, are proved to be of high accuracy and reliability. The mean absolute relative error for 29 communities (villages) is less than 7%. The absolute relative error of 25 out of 29 communities (villages) is less than 10%. The proportion of patches whose estimated number of people is in reasonable range is higher than 74% in a total of 1102 residential building patches. The proportion of patches whose relative error is in slightly underestimated area (-10%, 0) and overestimated area (0, 10%) is higher than 9%. (2) Building volume , defined by residential building patch area and building floor number, is a key factor to estimate accurately the number of people within a residential building. The percentage of housing area can further improve model accuracy. Public area rate plays an important role to increase estimated number of people in underestimated area and decrease that in overestimated area, but is too weak to adjust the estimated number of people to reasonable range. In conclusion, spatialization based on residential space attributes can be an important method for population spatialization research at the residential building scale.