Globalization has significantly reshaped global population dynamics, particularly in rapidly urbanizing developing countries like China and India. China's immense population and aging demographics necessitate a thorough understanding of its population trends to mitigate potential challenges. While urbanization has led to population concentration in urban areas, a deeper understanding of population redistribution at the county level is crucial. Previous research has highlighted the "dense in the southeast and sparse in the northwest" pattern, initially identified by the Heihe-Tengchong line, but lacked comprehensive, nationwide county-level analysis across all seven national censuses. This study addresses this gap by analyzing population distribution changes at the county level in China, providing a detailed picture of spatial and temporal patterns, and offering valuable insights for future policy decisions.

Existing studies on China's population distribution have focused on specific periods or regions, lacking a comprehensive, nationwide county-level analysis spanning all seven national censuses. Scholars have examined global and national population distribution patterns using various methods and indicators like population density, concentration, imbalance, and potential. However, studies employing a multi-scale approach to analyze population dynamics beyond urban areas remain limited. Population forecasting models, including probabilistic models, Leslie matrix models, and ARIMA models, have been used but often lack the granular spatial detail of this study. This research builds upon this existing body of work by offering a detailed spatial and temporal analysis at the county level for a comprehensive understanding of population distribution patterns.

This study utilized data from China's seven national population censuses (1953-2020), county-level administrative division maps from the Chinese Academy of Sciences, and other supplementary data. Key methods included: 1. **Population Density Calculation:** Population density (AD) was calculated as AD_i = P_i⁄A_i, where P_i is the population and A_i is the land area of county i. 2. **Spatial Autocorrelation Analysis:** Global and local Moran's I indices were calculated using GeoDa software and ArcGIS to assess the spatial autocorrelation of population changes across counties. Global Moran's I assesses overall spatial correlation, while local Moran's I identifies clusters of high-high (population growth), low-low (population decline), and other spatial relationships. 3. **Gini Coefficient Calculation:** The population-land Gini coefficient (G_R) was used to measure the inequality in the distribution of population across land areas. 4. **ARIMA Modeling:** An ARIMA (p, d, q) model was employed to forecast China's total population from 2023 to 2035. The model's parameters (p, d, q) were determined through analysis of autocorrelation and partial autocorrelation functions, followed by model fitting and validation.

The study revealed several key findings: 1. **Growth and Regional Disparities:** County populations showed significant growth and regional differences over the study period. The average population size increased from 228,000 to 498,000, but growth rates decelerated over time. The largest growth was between the second and third censuses (41.33%), with the smallest between the sixth and seventh (5.64%). 2. **Spatial Distribution Patterns:** The "dense in the southeast, sparse in the northwest" pattern persisted, although the proportion of the population in the southeastern half decreased slightly while that in the northwestern half increased slightly. Counties with populations >500,000 were mainly located in the southeast, while those with <100,000 were primarily in the northwest. The number of large-population counties increased, while the number of small-population counties decreased. 3. **Population Change Divergence:** More than 40% of counties exhibited population growth between censuses, with the highest percentage (96.43%) between the second and third censuses. However, this percentage decreased over time, with the lowest (46.64%) between the sixth and seventh censuses. Significant population decline was observed in northeastern and border areas. 4. **Spatial Autocorrelation:** Spatial autocorrelation analysis showed that population change exhibited significant spatial clustering, with "high-high" clusters around provincial capitals and "low-low" clusters in economically underdeveloped areas. This underscores spatial agglomeration trends. 5. **Population Evolution Types:** Analysis of population density and growth rate categorized county-level population change into different types: low-density rapid growth, medium-density rapid growth, medium-density stable growth, and medium-density negative growth. The shift from rapid to negative growth highlights an inflection point in China's population dynamics. 6. **Gini Coefficient of Population Distribution:** The population-land Gini coefficient consistently exceeded 0.7, indicating extreme inequality in population distribution. This inequality intensified over time, reflecting increased population centralization. Approximately 24% of China's land supports 86% of its population, highlighting a significant spatial imbalance. 7. **Population Projection:** The ARIMA (2,2,2) model projected a decrease in China's total population from 1410.05 million in 2023 to 1343.68 million by 2035.

The findings highlight the need for a rational understanding of population distribution changes in China. The shift towards negative population growth, coupled with spatial inequalities, poses significant challenges for sustainable socio-economic development. The research findings align with the observed concentration of economic activity and administrative resources in specific regions, creating a spatial mismatch between population distribution and economic opportunities. The concentration of population in a few areas necessitates adjustments to development strategies, particularly concerning resource allocation and infrastructure development in less developed regions. The decreasing population in some areas raises concerns about potential security implications in border regions, highlighting the need for tailored policies to attract and retain people.

This study provides a comprehensive analysis of spatial and temporal population changes in China at the county level. The findings reveal significant regional disparities, a shift towards population decline, and increasing spatial concentration. The projected population decrease emphasizes the need for proactive policy interventions to address the challenges associated with negative population growth and regional imbalances. Future research should focus on analyzing the factors driving these changes, investigate population dynamics at smaller spatial scales (township level), and explore the policy implications of these findings more deeply.

This study has several limitations. First, it does not explicitly model the drivers of population change. Second, it lacks a detailed subregional analysis (east, west, etc.) of population dispersion. Third, the analysis is limited to the county level, and a more granular analysis at the township level would provide a richer understanding of population distribution patterns. Finally, the accuracy of population projections depends on the assumptions underlying the ARIMA model.