基于地理大数据和多源信息融合的区域未来人口精细化空间分布模拟研究——以珠江三角洲为例

首先采用队列因素法和CA-Markov模型对区域未来人口规模和土地利用格局进行模拟预测，并结合POI地理大数据，利用多源信息融合法构建区域未来人口精细化空间分布模拟模型，以珠江三角洲城市群2030年各区县精细化的人口空间分布预测进行实证分析。结果表明：① 采用队列因素法进行珠江三角洲各区县人口规模预测的相对误差大部分在5%以下，基于CA-Markov模型土地利用模拟的Kappa系数达到0.97；② 珠江三角洲城市群精细化的人口空间分布模拟数据与实际人口数据的拟合趋势线R²达到了0.90，模拟效果优于Worldpop数据集，体现了POI地理大数据与多源信息融合在精细化人口空间分布模拟上的优势；③ 珠江三角洲未来人口呈现由中心向外围扩散和递减的空间分布格局，空间差异显著且较为稳定，70%的人口集中在广州、深圳、东莞和佛山等核心城市。

Simulation of Fine Spatial Distribution of Regional Future Population Based on Geographical Big Data and Multisource Fused Method: A Case of the Pearl River Delta

The fine spatial distribution of regional population in future plays an important role in constructing relevant future plans for country and regions. More previous researches focused on the spatial distribution of current regional population, while little deals with the fine spatial distribution of future population. In this article, taking the fine simulation of spatial distribution of population in 2030 for counties in the Pearl River Delta (PDR) as an example, we firstly simulated future population and land use type using the methods of cohort-component and CA-Markov model, respectively, according to the census data and current land use maps. And combined with the big data of Point of Interest (POI), a model that simulates the fine spatial distribution of regional population in future was constructed by the multisource fused method. The results are as follows: Firstly, the absolute relative error is mainly less than 5% when predicting demographic data of PDR with the method of cohort-component. The Kappa coefficient of land use simulation in PDR by CA-Markov model is as high as 0.97. Secondly, the R² for fitting line between the simulated spatial distribution of population data and the actual population data reaches 0.90, which is better than Worldpop data set and demonstrates the advantage of big data of POI in the fine spatial distribution simulation of population. Thirdly, the spatial distribution of population in PDR decreases from the center to the periphery, performing the significant spatial difference. Meanwhile, 70% of the population concentrates in the core cities of Guangdong-Hong Kong-Macao Greater Bay, the core cities mainly include Guangzhou, Shenzhen, Dongguan and Foshan. The spatial distribution of population in PRD was significantly different and relatively stable, and showed a trend of expansion from the center to the periphery, while the overall expansion was small. This study puts forward a multisource fused method for the fine spatial distribution of regional future population, and provides scientific basis and reference for the region to make relevant development plans in the future.