地球信息科学学报 ›› 2014, Vol. 16 ›› Issue (2): 199-206.doi: 10.3724/SP.J.1047.2014.00199

• 地球信息科学理论与方法 • 上一篇    下一篇

加权人口密度连续分布模拟模型

刘梦鑫, 邬群勇, 卢毅敏   

  1. 福建省空间信息工程研究中心, 福州 350001
  • 收稿日期:2013-05-28 修回日期:2013-09-12 出版日期:2014-03-10 发布日期:2014-03-10
  • 通讯作者: 邬群勇(1973- ),男,博士,副研究员,研究方向为计算机应用与GIS。E-mail:qywu@fzu.edu.cn E-mail:qywu@fzu.edu.cn
  • 作者简介:刘梦鑫(1989- ),女,福建福州人,硕士生,研究方向为空间信息网络共享与服务。E-mail:invisible_wind@foxmail.com
  • 基金资助:

    国家“863”课题项目(2012AA022007);福建省自然科学基金项目(2011J01268)。

The Weighed Population Density Continuous Distribution Simulation Model

LIU Mengxin, WU Qunyong, LU Yimin   

  1. Fujian Spatial Information Research Centre, Fuzhou 350001, China
  • Received:2013-05-28 Revised:2013-09-12 Online:2014-03-10 Published:2014-03-10
  • Contact: 10.3724/SP.J.1047.2014.00199 E-mail:qywu@fzu.edu.cn

摘要:

人口分布是人文科学的重要研究对象。如何在数据不足的情况下获得较为符合实际的人口分布数据是业界的一大难题。为了解决这个问题,本文依照人口分布的规律,以人口统计数据、高程、海岸线、城市中心、河流和行政区划为影响因子,通过距离计算、重分类、加权计算等方法,建立了加权人口密度连续分布模拟模型,并进行了改进。基于2011年的福建省县级人口统计数据,本文在ArcGIS平台上模拟了该省的人口密度空间分布,模拟结果表明:各县人口总数误差与平均人口误差在40%以下的百分比分别为89%与84%,突出了福建省在城市中心、沿海、沿江,以及平原地区的人口聚集规律;各个县级行政区域内的数据集中,相对独立;县与县之间的区分边界数据又相互联系,变化连续,符合人口分布的实际情况。该模型相比于其他的人口模型易于理解,结构简单,可应用于各省的人口密度分布模拟,能有效地解决人口数据不足的问题。

关键词: 人口密度, 模拟模型, 统计数据, 连续分类法

Abstract:

Human population distribution is an important parameter to science research, but the security of population data leads to the dilemma that many researchers cannot acquire the population distribution figures, which is of significance in their work. To solve this problem, this paper concludes the rules of population distribution pattern and determines the statistics data, administration divisions, evaluation, and the distance to coast, city center and river, as the factors of this pattern. Based on this pattern, an algorithm is built to simulate the spatial distribution of population by the distance calculating, reclassifying and weighted calculating. The algorithm is transformed into work flow model which simulated the population spatial distribution of Fujian in 2011 and is improved in the way how the data was classified and how the edge between counties was manipulated. After extracting slope information from the DEM and calculating the distance to county centers, city centers, rivers and the coast, the model classified the slope DEM and distance data into different types by the 1/4 Standard Deviation (to decide how many categories should be classified) and Quantile (to decide the range of every classifications). Then, a general effect cost raster was worked out. Following the summary of the maximum and minimum of the cost in each county, the premier allocation of population was reckoned upon the normalized effect cost raster. However, the edge between counties and the discontinuous between raster cells lead to the improvement of classify method and the interpolation of statistics data. This is improved by the continuous classify method, the calculation of effect cost and weighted population density. The simulated result shows population gathers around cities, along the coast and rivers, and on plain areas. What is more, the data inside counties emphasis its center effects and the data at the edge is connected with adjacent counties in a continuous way, which is more likely to comply with the population distribution in reality. It is about 89% and 84% of the total number of the counties whose errors of the total population and average population density in each county are below 40%. Compared to the average population density of each county, the simulation method that considers terrain is more suitable for researches in other fields and is more understandable.

Key words: statistics, population density, simulation model, continuous classify method