Source-linked AI summary
Ht-Index for Quantifying the Fractal or Scaling Structure of Geographic Features
Bin Jiang, Junjun Yin
TL;DR
The paper addresses why geographic features are widely regarded as fractal despite some studies rejecting that property under a strict fractal-dimension definition. It proposes the ht-index, based on recurring patterns of far more small things than large ones, and uses case studies to show how it complements fractal dimension in characterizing geographic complexity.
Problem
Strict fractal-dimension definitions have led some geographic studies to claim that the fractal property is not universal.
Method
The paper defines the ht-index as one plus the number of recurrences of far more small things than large ones across scales and computes it using nested rank-size plots.
Results
The case studies show that ht-indices characterize complexity across nightlight imagery, terrain surfaces, and urban street networks, with terrain reaching 17 and urban streets ranging from 5 to 9.
Takeaways & Limitations
The ht-index complements fractal dimension by capturing hierarchical scaling levels and indicates greater complexity, heterogeneity, and diversity at higher values.
Takeaways & Limitations
Fractal dimension does not differentiate the complexity of individual developmental stages or iterations, and less-developed features may fit lognormal or exponential rather than power functions.
Abstract
from arXiv · showhide
Although geographic features, such as mountains and coastlines, are fractal, some studies have claimed that the fractal property is not universal. This claim, which is false, is mainly attributed to the strict definition of fractal dimension as a measure or index for characterizing the complexity of fractals. In this paper, we propose an alternative, the ht-index, to quantify the fractal or scaling structure of geographic features. A geographic feature has ht-index h if the pattern of far more small things than large ones recurs (h-1) times at different scales. The higher the ht-index, the more complex the geographic feature. We conduct three case studies to illustrate how the computed ht-indices capture the complexity of different geographic features. We further discuss how the ht-index is complementary to fractal dimension, and elaborate on a dynamic view behind the ht-index that enables better understanding of geographic forms and processes. Keywords: Scaling of geographic space, fractal dimension, Richardson plot, nested rank-size plots, and head/tail breaks
1. Introduction
Geographic features are irregular and scale-dependent, challenging Euclidean descriptions and motivating fractal geometry. The paper argues that claims against universal geographic fractality arise from fractal dimension’s strict definition and introduces the ht-index as a complementary measure.
- Motivation: Geographic features such as mountains and coastlines are irregular, rough, and not adequately described by Euclidean heights and lengths.Coastline length increases with finer measuring scales, making geographic length scale-dependent.
- Motivation: Fractal dimension characterizes irregularity or roughness through the degree of space-filling, but can be mistaken for a density measure.The paper distinguishes fractal dimension as fractal-based from density as a Euclidean concept.
- Scaling pattern: Fractal patterns commonly contain far more small things than large ones across geographic and social phenomena.Examples include cities and streets, although the supplied passage truncates additional examples.
- Paper objective: The paper rejects claims that fractality is non-universal and proposes the ht-index to capture scaling structure not captured by fractal dimension.Its purpose is to quantify geographic complexity rather than prove that geographic features are fractal.
- Paper objective: The paper reviews fractal dimension, defines and computes the ht-index using nested rank-size plots, and demonstrates it through three geographic case studies.The planned studies cover geographic features at country and city levels.
2. Fractal dimension and its limitations
Fractal dimension characterizes self-similar forms through scale–detail relationships, but its strict power-law requirement and form-focused scope limit its treatment of geographic scaling and development.
- Fractal dimension and scaling: Fractal dimension measures complexity through the logarithmic ratio D = log(N)/log(r), where r is scale change and N is the detail needed to cover a pattern.In a Richardson plot, the logarithmic distribution-line slope equals D.
- Fractal dimension and scaling: For the Koch curve, each iteration reduces scale by 1/3 while increasing detail from 4 to 16 to 64, yielding D = log(4)/log(3) = 1.26.This illustrates the pattern of far more small things than large ones.
- Scale and scaling: Fractals comprise scaling hierarchies across many scales; strict forms can have infinite ranges, whereas real features such as fern leaves have finite ranges.The Koch curve at iteration 3 spans scales from 1/27 to 1/3.
- Scale and scaling: The fractal concept of scale is tied to a hierarchy of scales, differing from geography’s cartographic, analysis, and phenomenon meanings.The paper distinguishes fractal scaling from conventional geographic uses of scale.
- Limitations: Fractal dimension’s power-law requirement can exclude geographic features whose scale–detail relationships are lognormal or exponential, while real data make these heavy-tailed forms difficult to distinguish reliably.The paper notes that least-squares fitting may misidentify lognormal or exponential relationships as power laws.
- Limitations: A relaxed scaling view treats lognormal and exponential relationships as possible for developing phenomena, with power laws associated with more fully developed stages.The fern leaf is presented as less fractal in early growth stages and more power-law-like later.
- Limitations: Seeing geographic scaling depends on perspective and scope: local street-network views may miss patterns visible across entire streets.The paper emphasizes that scale selection affects whether far more small things than large ones can be observed.
- Limitations: Fractal dimension captures fractal forms but does not distinguish complexity across growth or iteration stages that contain different numbers of scales.The ht-index is introduced to capture this stage-dependent structure.
3. Ht-index and nested rank-size plots
The ht-index counts how often the pattern of far more small things than large ones recurs across scales, using iterative ranking and head/tail breaks to reveal scaling hierarchies.
- Definition: The ht-index is one plus the number of recurrences of far more small things than large ones across different scales.A higher ht-index indicates a more complex geographic feature.
- Koch curve example: For the Koch curve, segment sizes 1/27, 1/9, and 1/3 occur in counts 64, 16, and 4, yielding an ht-index of 3.The distribution contains far more small segments than large ones.
- Computation: Head/tail breaks iteratively split the segments around means, separating minority heads of larger segments from majority tails of smaller segments.The first mean separates 20 segments above it from 64 below it; the second separates 4 from 16.
- Nested rank-size plots: Nested rank-size plots display the recurring heavy-tailed pattern by repeatedly plotting the head, making the scaling hierarchy visually intuitive.The Koch curve exhibits the far-more-small-than-large pattern twice, corresponding to ht-index 3.
- Derived scales: For the Koch curve, head/tail breaks automatically derive three scales, which are labeled levels rather than iterations.The derived levels look the same as the three iterations but are computed from the iteration-3 curve.
- Geographic scaling: The head need not always remain below 25 percent, but geographic scaling commonly retains an unbalanced partition between minority heads and majority tails.The head/tail imbalance is presented as a recurring signature of scaling in geographic space.
4. Case studies: Computing the ht-index of geographic features
Three case studies apply head/tail breaks to nightlight imagery, terrain heights, and urban street connectivity, showing that ht-indices characterize scaling complexity across geographic features.
- Nightlight imagery: Nightlight imagery contains far more dark than light pixels, and head/tail breaks organize its 11,766,012 pixels into hierarchical classes.Pixel lightness ranges from 0 to 63; the first mean is 7.5, with 26 percent above and 74 percent below it.
- Nightlight imagery: The nightlight imagery has ht-index 3 because two valid means produce three classes.The analysis treats nightlight as a proxy that captures the pattern of human settlements.
- Terrain surface: The United States DEM contains approximately 3 million pixels and shows a scaling hierarchy of low and high terrain values across places and scales.Heights range from -147 to 4161 meters, with more low locations than high ones.
- Terrain surface: The terrain surface has ht-index 17, substantially higher than the nightlight imagery, consistent with its greater heterogeneity and diversity.The paper presents this as evidence that ht-index may indicate complexity among geographic features or fractals.
- Urban streets: The urban-street case study measures street-network hierarchy through the recurring pattern of more less-connected streets than well-connected ones.It applies the ht-index to city morphology from the perspective of street connectivity.
- Cross-case conclusion: Across nightlights, terrain, and urban streets, the case studies show far more small things than large ones as a recurring fractal or scaling property.The paper concludes that ht-index can characterize complexity across these geographic features.
5. Implications of the ht-index
The ht-index complements fractal dimension by capturing hierarchical levels of scale heterogeneity, while fractal dimension captures the degree of heterogeneity. This distinction supports a dynamic view of geographic forms and processes.
- The ht-index captures the inherent hierarchy of geographic features, supporting statistical mapping, map generalization, and cognitive mapping.
- The ht-index expresses hierarchical levels for heterogeneous scales, while fractal dimension expresses the degree of heterogeneity.
- In adjusted Koch curves, fractal dimension increases with scaling ratio, whereas ht-index increases only when fine scales are added.
- The ht-index offers a measure of spatial heterogeneity, but its effectiveness requires further study.
- The ht-index supports a dynamic view in which adding fine structures or removing them makes patterns more fractal over time.
- The paper broadens fractal structure from strict power laws to power-law-like distributions when far more small things than large ones persists across most hierarchical levels.
6. Conclusion
The paper develops the ht-index as a simple alternative for quantifying geographic fractal or scaling structure through recurring patterns of far more small things than large ones. It complements rather than replaces fractal dimension by capturing hierarchical detail and complexity.
- The ht-index quantifies geographic fractal or scaling structure through the recurring pattern of far more small things than large ones.
- The ht-index captures the hierarchical levels of that recurring structure.
- The ht-index is an alternative to fractal dimension for capturing detailed aspects of fractals, not a replacement for fractal dimension.
- Higher ht-index values indicate geographic features that are more complex, heterogeneous, mature, and/or natural, and may contain more information.