首页|多源人口空间化数据集精度评析与融合——以陕西省为例

多源人口空间化数据集精度评析与融合——以陕西省为例

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
以陕西省全域为研究区,以乡镇为单元评析了上述4种不同来源的人口空间化数据集在精度上的差异.在此基础上,分别通过基于方差加权模型的分区融合方式和基于多元线性回归(multiple linear regression,MLR)模型的整体融合方式构建两种模型,以实现源数据集的信息融合和精度提升.研究结果表明:在4种人口空间化数据集中,World‑Pop整体精度最高,GPWv4与WorldPop精度接近,而由于原始输入数据尺度较大,LandScan与CnPop在乡镇尺度的人口数量估值精度偏低;两种融合模型在提升人口数据精度方面都发挥了一定作用,适用性较好,且相对而言MLR模型的提升效果更好.
The Accuracy Evaluation and Fusion of Multi-source Population Spatialization Datasets in Shaanxi Province
Currently,large-scale population spatialization da-ta set products such as WorldPop,GPWv4,LandScan,CnPop are available for public download,which have different accuracy and population information from different regions. To explore their differences and the possibility of mutual fu-sion to improve the quality of population product,we select Shaanxi as the research area and towns as the research unit,to analyze the accuracy differences of the four population spatial-ization datasets from different sources. On this basis,we build a partitional fusion model based on variance weighting,and an overall fusion model based on multiple linear regres-sion (MLR),to fuse the information of the four datasets and improve the accuracy of the source dataset. The research re-sults show that:First,WorldPop has the highest overall accu-racy among the four population spatialization datasets,and the accuracy of GPWv4 is close to that of WorldPop. Due to the large scale of the original input data,LandScan and CnPop have large errors in population estimation at the township lev-el. Second,both fusion models play a role in improving the accuracy of population data with good applicability,and MLR model has a better fitting effect.

spatialized population datasetaccuracy assessmentfusionvariance weight modelmultiple linear regression model

李紫涵、吴田军、王洁、葛咏、高鹏、王江浩

展开 >

长安大学地球科学与资源学院,陕西西安,710064

长安大学理学院,陕西西安,710064

自然资源部大地测量数据处理中心,陕西西安,710054

中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室,北京,100101

展开 >

人口空间化数据集 精度评析 融合 方差加权模型 多元线性回归模型

国家杰出青年基金陕西省重点研发计划内蒙古自治区科技重大专项重庆市农业产业数字化地图项目长安大学中央高校基本科研业务费专项

417250062021NY-1702021SZD003621C00346300102120201

2024

10.14188/j.2095-6045.2022063

测绘地理信息
武汉大学

测绘地理信息

CSTPCD
影响因子：0.563
ISSN：1007-3817
年,卷(期)：2024.49(4)