基于影响因子动态更新的都市圈国土-人口协同模拟

都市圈作为典型的复杂巨系统,其国土、人口、交通等要素呈现出共同演化、协同发展的特征。现有基于元胞自动机的空间模拟方法通常分别进行国土或人口模拟,忽略了国土、人口之间的协同演化,从而限制了模拟性能。本文提出一种影响因子动态更新策略用以表征国土、人口两种要素间的相互作用,拓展经典元胞自动机模型构建协同模拟算法,实现了都市圈国土与人口空间分布态势的精准模拟。本文以深莞惠都市圈为例,对提出算法进行了实验验证。实验结果表明,本文提出协同模拟算法在深莞惠都市圈的国土模拟品质因素为0.274、人口模拟平均绝对百分比误差23.55%,分别优于传统基于随机森林的元胞自动机算法0.24和29.33%;相较于传统模型,在国土模拟中本模型在对于建设用地的模拟准确度提升了约3%,在人口模拟中本模型在对人口高密度的区域模拟误差降低了约6%。本文进一步预测了深莞惠都市圈2030年国土和人口发展空间态势。研究结果可为都市圈重大基础设施选址和发展情景推演提供技术支撑。

Synergistic Simulation of Land and Population in Metropolitan Area With Dynamic Influence Factors

Metropolitan area is a typical complex system with co-evolution and coordinated development of land, population, and transportation. It has become an urgent issue to guide the development path of metropolitan area scientifically by coordinating the territorial space and human resources in metropolitan area. Cellular automata is a discrete grid dynamic model. By constructing simple transformation rules, combined with strong spatial modeling ability and computing ability, it has become an effective means to study complex urban systems. However, existing cellular automata-driven spatial simulation methods usually simulate land or population separately. Their performance are limited as they ignore the co-evolution between land and population. This paper proposes the dynamic influence factors to characterize the interaction between land and population. The core idea is to take the simulation results of land and population as the influence factors of new cellular automata, and express the complex driving processes of land change and population change by updating the cell transformation probability. The spatial simulation model is repeatedly trained to describe the synergistic evolution between land and population in the process of development. The classical cellular automata model is then extended to construct a synergistic simulation framework which realizes the accurate spatial simulation of land and population in the metropolitan area. Taking the Shenzhen-Dongguan-Hui metropolitan area as an example, the experiment was conducted to verify the proposed method. The results show that the synergistic simulation algorithm can accurately simulate the land and population development processes. The simulation results of the proposed algorithm show a Figure of Merit of 0.274 for land and Mean Absolute Percentage Error of 23.55% for population, which outperfoms the traditional cellular automata algorithm coupled with random forest (Figure of Merit = 0.24, and Mean Absolute Percentage Error = 29.33%). Compared with traditional models, the simulation accuracy of this model for construction land is improved by about 3% in land simulation, and the simulation error of this model for high-density population area is reduced by about 6% in population simulation. We further predict the spatial scenario of the land and population development of the Shenzhen-Dongguan-Hui Metropolitan Area in 2030. The results show that the expansion of construction land mainly occurs in the undeveloped areas with good infrastructure (such as Huizhou area), the population further gathers in the high-density areas, and the growth is mainly concentrated in the core urban areas of the metropolitan area. These results provide technical support for locating important infrastructure facilities and development scenarios simulation in the metropolitan area.