Abstract: The agglomeration effect of population, which explains why many people live near places where many other people also live, is one important interaction that influences human population. We examine the agglomeration effect by measuring the distribution of the logarithmic differences between populations living in two places separated by some distance. The shapes of the distributions of the logarithmic differences closely resemble each other without depending on the regions or the country in cases of small scale of separation distance. This result suggests a unified explanation to understand the population distributions of various regions.
Introduction
For decades, urban scientists have relied on Zipf’s Law to describe city size distributions. However, Zipf’s Law suffers from a "boundary problem": the mathematical results shift significantly depending on how administrative city limits are defined. This study seeks to bypass these boundaries by examining the fundamental human interaction that drives us to cluster together—the agglomeration effect. By analyzing population density across Japan and Europe, researchers aimed to determine if universal properties of population distribution exist independent of geography or culture.
Methodology
To avoid the pitfalls of administrative definitions, the researchers utilized a mesh-based spatial division. Maps were divided into identical square blocks of varying sizes, defined by a parameter $BS$ (block size in km).
Key technical steps included:
- Inhabitable Area Filtering: Calculating "inhabitable area" by excluding non-residential land like lakes and steep mountains, which is critical for topographically diverse regions like Japan.
- Logarithmic Differences: Borrowing techniques from financial time-series analysis, the study measured the logarithmic difference in population between neighboring squares to capture the relative intensity of clustering.
- Scaling Analysis: The researchers observed how these distributions evolved as the block size ($BS$) increased from 0.5 km to 10 km.
The Scale of Universality
The study's most significant finding is the existence of a "Scale of Universality." At small scales ($BS = 0.5$ km), the population distribution curves for all eight regions of Japan—ranging from the rural expanses of Hokkaido to the hyper-urban Kanto region—overlap almost perfectly. This suggests that the local "rules" governing how humans settle near one another are universal and independent of regional density.
Results and Discussion
The data suggests that population distribution is driven by two distinct forces:
- Internal Interactions: Social and economic ties between individuals dominate at small scales (under 1 km), creating universal patterns regardless of the broader environment.
- External Interactions: Factors such as topography, climate, and regional history dominate at larger scales. This explains why regional differences only become apparent as the grid squares increase in size (e.g., $BS = 10$ km).
Mathematically, the distribution of population differences transitions from a truncated normal distribution at small scales (indicating highly correlated neighbors) toward an exponential distribution at larger scales as the correlation between distant points weakens. Despite massive differences in average density—Kanto is approximately 11 times denser than Hokkaido—the local agglomeration behavior remains identical.
Conclusion
This research demonstrates that while cities may appear different globally, the underlying mathematical pattern of local human clustering is remarkably consistent. By identifying that universality emerges at scales below 1 km, the study provides a new quantitative index for the strength of the agglomeration effect. Future work will focus on developing generative models that use these local interaction rules to simulate how population configurations grow and evolve over time.
Summary updated