基于2005年和2015年人口数据,运用收缩指数、地统计分析和地理探测器等方法,研究不同地理尺度下黄土高原地区人口收缩现象与地域分异格局,揭示黄土高原地区人口收缩驱动力差异及机理。结果表明：（1） 黄土高原地区县级与地级行政单元下人口收缩比例均较高,单一地理尺度识别人口收缩存在偏差。（2） 黄土高原地区的人口收缩现象在区域-省-地级市尺度均符合“核心-外围”模式,核心区域（城市）人口集聚,边缘地区（县）人口流失。（3） 经济发展水平差异和城乡发展不平衡成为黄土高原地区人口集聚与流失的主要诱因。地级市收缩型主要受城镇化和工业化水平驱动,为城市综合发展能力不足而导致的人口收缩;地级市局部收缩型处在社会-经济高速发展时期,内部要素分配不均导致内部局部收缩。

The execution of the strategy of west China development accelerated social and economic development in the Loess Plateau region. Unbalanced regional development has induced population migration, which has triggered a new phenomenon of population shrinkage in the western region. A population shrinkage index, geostatistical analysis, and geographical detectors based on resident population data in 2005 and 2015 were used to study the population shrinkage and regional differentiation in the Loess Plateau at prefecture-level cities and county scales. Based on these geographic scales, 45 prefecture-level cities in the Loess Plateau were classified by population change, revealing differences and mechanisms of population shrinkage driving forces in the Loess Plateau. We explore the driving factors of population shrinkage in the Loess Plateau from a socio-economic perspective, analyze the universality and specificity of population shrinkage, and enrich the research on mesoscale urban shrinkage in less developed regions in China. The results are as follows: (1) Population shrinkage in the Loess Plateau region is universal, and the proportion of population shrinkage is higher under county- and prefecture-level administrative units. The shrinkage rate of county-level administrative units was 34.49%, and that of prefecture-level administrative units was 22.22%. There is a bias in identifying a shrinking population at a single geographic scale, and the proportion of prefecture-level cities with at least one shrinking county-level unit accounts for 68.89% of all prefecture-level cities. At the prefecture-level city level, all provincial capital cities are shrinking, and general prefecture-level cities have high shrinkage rates. At the county level, municipal districts, county-level cities, and autonomous counties have low shrinkage rates. The number of general counties with a shrinking population is 90, accounting for 82.57% of shrinking county-level units. (2) Population shrinkage conforms to the core-periphery model at the regional-provincial-prefecture-level scale. Core area (cities) populations are concentrated, and marginal areas (counties) are losing population. The Guanzhong Plain and Ningxia Plain have become core areas of population concentration, and the Liupan Mountain District has become a concentrated population loss area; the provincial capital cities have stronger population concentration capabilities, and the ordinary prefecture-level cities have population shrinkage. Specific to the prefecture-level cities, the population will be concentrated in the municipal districts, and the population of peripheral counties will shrink. (3) Differences in economic development levels and imbalances in urban and rural development have become the main incentives for population agglomeration and loss in the Loess Plateau. The contraction of prefecture-level cities is driven by urbanization and industrialization, leading to insufficient comprehensive development capacity; prefecture-level cities are in a period of rapid socio-economic development, and the uneven distribution of internal factors leads to internal partial shrinkage. Coordinated regional development for resource allocation to different types of prefecture-level cities and ensuring a relatively fair society are issues that governments and planners should reflect on and pay attention to.