【摘要】：人口空間化是實現人口統(tǒng)計數據與其他環(huán)境資源空間數據融合分析的有效途徑。本文選取夜間燈光數據、道路網數據、水域分布數據、建成區(qū)數據、數字高程模型和地形坡度數據作為影響珠江三角洲人口分布的變量因子,利用隨機森林模型對珠江三角洲2010年人口數據進行了30 m格網空間化,并將模擬結果與三個公開數據集作精度對比,最后基于隨機森林模型的變量因子重要性分析珠江三角洲人口空間分布的影響因素。結果表明:本文模擬整體精度達到82.32%,均優(yōu)于World Pop數據集以及中國公里網格人口數據集,接近GPW數據集,而且在人口密度中等區(qū)域模擬精度最高;通過對變量因子重要性進行度量,發(fā)現夜間燈光強度是珠江三角洲人口分布的最重要指示性指標,到水域的距離、到建成區(qū)的距離和路網密度對珠江三角洲人口分布均具有重要作用。利用隨機森林模型結合多源信息能夠實現高空間分辨率的人口空間化,可為精細化城市管理提供重要數據源,也可為相關政策決策制定提供支持。
[Abstract]:The spatialization of population is an effective way to realize the fusion analysis of population statistics and other environmental resources spatial data. In this paper, night lighting data, road network data, water distribution data, built area data, digital elevation model (Dem) and topographic slope data are selected as variable factors affecting population distribution in the Pearl River Delta (PRD). The population data of Pearl River Delta (PRD) in 2010 are spatialized by random forest model, and the simulation results are compared with the accuracy of three open data sets. Finally, the influence factors of population spatial distribution in Pearl River Delta are analyzed based on the variable factor importance of stochastic forest model. The results show that the overall accuracy of the simulation is 82.32%, which is better than the World Pop data set and the Chinese kilometer grid population data set, and is close to the GPW data set. Moreover, the simulation precision is the highest in the medium area of population density. By measuring the importance of variable factors, it is found that nighttime light intensity is the most important indicator of population distribution in the Pearl River Delta, and the distance to the waters. The distance to the area and the density of the road network play an important role in the population distribution of the Pearl River Delta (PRD). The random forest model combined with multi-source information can realize the population spatialization with high spatial resolution, which can provide important data sources for fine urban management and support the decision-making of related policies.
【作者單位】： 中山大學地理科學與規(guī)劃學院廣東省城市化與地理環(huán)境空間模擬重點實驗室綜合地理信息研究中心;
【基金】：國家自然科學基金重點項目(41531178) 廣州市科技計劃項目(201510010081) 國家自然科學基金項目(41001291)~~
【分類號】：C924.1
，

本文編號：2456469