Urban and rural have long been treated as a switch: a settlement is one or the other, and everything between the two labels disappears. A study published in Frontiers of Urban and Rural Planning argues that this binary habit has become a scientific liability, and it proposes a replacement. Researchers Juliana Melo Linhares Rangel, Apiano Ferreira Morais and Marcelo Alves Ramos introduce the Urbanicity Gradient Index, or UGI, a continuous 0-to-100 measure designed to place any locality on a single spectrum running from isolated rural community to global megacity.
The core innovation is how the index decides what matters. Instead of letting researchers assign importance to each variable by judgment, the UGI uses Principal Component Analysis to derive weights from empirical urbanization patterns in the data itself. The index integrates four dimensions: population size, population density, distance to the nearest urban center, and infrastructure development across seven domains spanning economic, health, transportation, communication, sanitation, social and educational facilities. Population and density are recorded as continuous numbers, while every infrastructure item is a simple binary presence-or-absence indicator.
To build and test the index, the team assembled data from 100 localities covering the full urbanization spectrum: 27 rural and peri-urban communities, 19 small towns, 23 medium-sized cities and 31 large cities, including the 20 most populous cities on Earth. Data collection methods varied by locality type. For rural communities in Brazil, where official infrastructure records were often missing or unreliable, local partners carried out systematic assessments through direct observation and community surveys. Small and medium-sized cities in the state of Ceará were profiled using standardized municipal data from IPECE, while large cities drew on IBGE statistics and demographic analysis.
The infrastructure component is deliberately coarse. All 34 infrastructure variables are coded as present or absent rather than measured in quantity or quality, a choice grounded in urban allometry, the finding that infrastructure elements scale in correlated ways as settlements grow. A university’s presence, for example, typically travels with advanced telecommunications and specialized healthcare even when those are not directly counted. The trade-off is real: a village with one under-resourced health post scores identically to one with a large hospital on the clinic variable. But the simplification means the index can be populated through direct observation, basic surveys or simplified administrative records, making it usable where detailed service data simply do not exist.
The mathematics behind the scoring is layered. Population size, the dominant dimension worth up to 40 points, is modeled with a sigmoid function whose S-shaped curve reflects an empirical observation: very small settlements barely differ in urban character, very large cities approach maximum urbanicity, and the steep middle transition is where categorical systems lose the most information. Population density adds up to 10 points through an exponential function with diminishing returns. Distance to the nearest municipal seat contributes up to 10 points through linear decay, with the 50-kilometer threshold borrowed from Rural-Urban Commuting Area codes as the boundary of daily urban influence. Infrastructure fills the remaining 40 points, weighted by each variable’s empirical contribution across the first two principal components.
When applied to the 100 localities, the index separated settlements cleanly around its 50-point threshold. Rural and peri-urban communities scored between 4.5 and 50.0, urban areas between 50.1 and 100.0, with minimal overlap. Megacities such as Tokyo, Delhi and Shanghai approached the maximum, isolated rural communities fell below 25, and small towns and transitional areas occupied the intermediate range. Used as a binary classifier at the 50-point cutoff, the UGI achieved perfect agreement with traditional urban-rural categories, with a Cohen’s kappa of 1.00 and all 27 rural and 73 urban localities correctly classified. Factor analysis returned five factors accounting for 100 percent of variance in theoretically interpretable patterns.
The most demanding test was robustness. The team ran 5,000 independent simulations: five subsample sizes from 50 to 90 localities, each evaluated across 1,000 runs in which the entire pipeline, from standardization through PCA weight derivation, was rebuilt from a random subsample and then applied to score all 100 localities. Binary classification held perfectly in every condition, covering 500,000 individual classifications. Continuous scores were even more stable: mean absolute error fell from 0.21 UGI points at 50 localities to 0.07 at 90, correlations held at 1.000 throughout, and every locality at every sample size landed within 5 points of its full-sample reference value.
Sensitivity analyses probed the index’s arbitrary-looking constants. Testing distance thresholds from 10 to 100 kilometers in 5-kilometer steps, 17 of 19 configurations met stringent stability criteria, with the 50-kilometer baseline sitting comfortably inside a wide corridor of defensible alternatives. A separate analysis of the sigmoid steepness parameter, sweeping values from 0.5 to 5.0, found the empirical optimum at 1.92, close to the chosen value of 2, with binary classification agreement remaining at 98 to 100 percent even as continuous scores varied. The authors attribute much of this stability to the index’s deliberate asymmetry: roughly half the scale is anchored to population figures that do not change with recalibration, while infrastructure and distance act as refinement layers.
Is the index secretly just a population proxy? The authors tested this directly by computing a restricted version without infrastructure and distance. Despite a high overall correlation of 0.952, the full index changed the rank of 96 percent of localities, with a mean shift of 7 positions and a maximum of 27. Notably, the reordering extended to the top of the distribution: megacities such as Delhi and Mumbai were separated by infrastructure differences, and Porto Velho, a Brazilian city of 460,000 people with comprehensive services, scored 96.4 and ranked alongside global megacities, while larger but poorly served towns fell below it.
The authors are candid about the limits. The validation sample is concentrated in Brazil, so the UGI should be understood as a methodological framework rather than a portable scale: weights and thresholds must be recalibrated locally using a dataset spanning the full rural-urban spectrum. The binary infrastructure coding sacrifices information on service quality and coverage intensity, and assigning one score to an entire locality obscures intra-urban variation in large cities. Longitudinal use carries its own hazard, since urbanicity scales can lose measurement invariance over periods as long as 15 years, meaning score changes may reflect instrument drift rather than genuine urbanization. What the index offers now is resolution: a settlement scoring 35 and one scoring 48 are both rural on paper, but the continuous scale can track the incremental infrastructure and demographic shifts that categorical systems register only as a sudden reclassification.
Subject of Research: Development and validation of a continuous Urbanicity Gradient Index measuring settlement urbanization from rural to metropolitan on a 0-to-100 scale
Article Title: Beyond binary urban-rural classifications: a continuous urbanicity gradient index
Article References: Rangel, J. M. L., Morais, A. F., & Ramos, M. A. (2026). Beyond binary urban-rural classifications: a continuous urbanicity gradient index. Frontiers of Urban and Rural Planning, 4(1), Article 18. https://doi.org/10.1007/s44243-026-00089-2
Image Credits: AI Generated
DOI: 10.1007/s44243-026-00089-2
Keywords: urbanicity, urban-rural continuum, Principal Component Analysis, index development, urbanization measurement, infrastructure indicators, population density, Brazil, robustness testing, settlement classification, SDG 11, geospatial analysis
News Source: Denise Maddox. (October 29, 2026). New continuous index scores urbanicity from 0 to 100. Scienmag.
