摘要
人口空间分布数据是开展灾害风险管理的基础数据,而科学、准确的人口空间化方法是实现人口空间分配的重要途径.鉴于城镇与乡村居住地在灯光数据和自然条件影响程度上的差异,本文提出一种综合利用夜间灯光数据和多地理因子数据的人口空间化方法,并以辽宁省2010年人口数据空间化为例进行方法验证.研究表明:1)夜间灯光数据在灯光值较高的城镇地区与人口密度关系显著,而地理因子加权的方法则更适用于受自然条件约束多的乡村区域;2)辽宁省人口密度在0~65 572人·km^(-2)之间,在空间上,中部平原地区人口密度较高,而两侧山地地区人口密度相对较低,与实际情况相符;3)辽宁省抽样乡镇人口平均误差为15.3%,精度验证结果表明,本文提出的人口空间化方法能够在乡镇尺度上具有较高的精度.
Population distribution is a significant factor in pollution control and urban planning.Disaster frequency is increasing with global warming.In disaster control,human loss is the most fatal loss for a country.Therefore,precise population distribution data is necessarily required.However,traditional census data are usually at county levels or town levels at best.They are also incompatible with other geographic data and can hardly facilitate further spatial analysis.Population spatialization which scatters population into grid cells is an effective way of solving these problems.This experiment presents a new method of population spatialization by utilizing DMSP/OLS night-time light data and mixing other geographic data which are of great influence to population.Night-time light data are found to have stronger connection with population in higher DN areas.The method of mixing geographic data like land use type and DEM is more efficient in areas extremely restrained by natural conditions.Different parts of the province are therefore treated with different methods.Considering that population density in urban areas is usually higher and less restricted by natural conditions,it matches the former method.Population in rural areas suits the latter method.Plain in the middle part of the province is of higher population density than the east and west parts;population is highly concentrated in urban areas.The population density ranks from 0~65 572/km2 and the highest density place locates in the capital city Shenyang.All of the results match with actual real situation.A total of 443 towns are selected,actual and calculated populations are compared.The average relative error is only 15.3%,much better than the result of LandScan 1km spatialization data with an average relative error 40.7%,therefore our method is comparatively more accurate.
出处
《北京师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2015年第S1期57-61,共5页
Journal of Beijing Normal University(Natural Science)
基金
国家高分辨率对地观测重大专项基金资助项目(民用部分)
关键词
人口空间化
灯光数据
多地理因子数据
辽宁省
population spatialization
night-time light data
multi-graphic factors
Liaoning Province
