In recent years, with the accelerated pace of global urbanization and a substantial increase in public health consciousness, sports facilities have far - reaching implications for the quality of life of residents. This study selects Tianhe District in Guangzhou as a case and explores the spatial distribution characteristics and influencing mechanisms of sports facilities through the application of GIS spatial analysis methods, the network dimension model, and the Geodetector. The findings are as follows: (a) The overall spatial pattern presents a distribution of “dense in the west and sparse in the east; dense in the south and sparse in the north”, featuring a multi - centered agglomeration pattern, mainly concentrated in areas such as Tianhe South Subdistrict and Linhe Subdistrict. (b) The facilities are arranged in a compact manner, with an average distance of less than 500 m, yet the distribution demonstrates a significant imbalance. (c) Human factors exert a significantly stronger influence on spatial differentiation compared to natural factors. Among them, transportation distance has the most substantial impact, while topography has the weakest influence. Moreover, the results of interaction detection reveal an enhanced interactive influence between pairwise factors on the spatial distribution of sports facilities. This research offers a scientific basis for optimizing urban sports facility planning and balancing resource allocation, significant for promoting high-quality urban public service development.
GeodetectorGISnetwork dimensionspatial distributionsports facilityThe author(s) declared that financial support was received for this work and/or its publication. This research was funded by the Guangdong Provincial Department of Education Characteristic Innovation Project, grant number 2024KTSCX092; Huizhou Science and Technology Project, grant number 2022CQ010026; the Guangdong Province Youth Innovation Talent Project, grant number 2018KQNCX245; and College Students’ Innovative Entrepreneurial Training Plan Program, grant number 151111323.section-at-acceptanceUrban ScienceIntroduction
With the continuous advancement of global urbanization, the development model of cities has shifted from large-scale physical expansion to high-quality improvement within cities. Urban planning increasingly emphasizes a people-oriented concept, and the supporting construction of urban facilities needs to be improved to be more convenient and friendly for urban residents. Sports facilities, as crucial public service infrastructures in cities, hold profound importance for residents’ quality of life, health levels, and urban sustainability.
In the field of sports facility-related research, the academic community has carried out extensive explorations on core issues such as their spatial distribution characteristics, accessibility levels, and quality differences, and has formed rich research results. From the perspective of research content and methods, existing studies mainly focus on three dimensions: spatial characteristic analysis, equity evaluation, and input-effect analysis. In terms of the analysis of spatial distribution and regional differences, Hoekman et al. (2016) conducted descriptive analyses on the existence and proximity of different types of sports facilities in the Netherlands, and tested the differential characteristics of sports facility distribution among quintile groups of regional poverty levels and urbanization levels through analysis of variance (ANOVA). In domestic studies, Li et al. (2025) systematically explored the spatial agglomeration characteristics, accessibility status, and social equity issues of public sports facilities in Shanghai at both street and grid scales by means of GIS technology and a variety of spatial analysis methods. As to equity-related research, the correlation between the spatial distribution of sports facilities and social inequality has been a key focus of academic attention, with numerous scholars conducting in-depth explorations around this direction (Shen et al., 2020; Zhang et al., 2022; Murray, 2009; Wang et al., 2023). In addition, in the field of investment and effect evaluation of sports facilities, some scholars have adopted socioeconomic models to conduct quantitative analysis and empirical testing on the investment costs of sports facilities and their multiple resulting effects (Hallmann et al., 2012; Hu et al., 2025; Rafoss and Troelsen, 2010). To summarize, existing studies have extensively employed geospatial analysis and statistics methods to carry out research related to the distribution of sports facilities, with the research focus mostly concentrated on social equity issues. In terms of exploring influencing factors, the main focus has been on variables such as accessibility and population, laying a solid foundation for research in this field. However, it should be noted that the spatial distribution of sports facilities is non-random, and the formation of its pattern is the result of the comprehensive effect of multiple factors. There remains room for expansion in current research: first, most existing studies on the distribution of sports facilities remain at the level of status description, making it difficult to fully and in-depth analyze the internal logic of its spatial pattern; second, in the research on influencing factors, existing achievements mostly focus on the discussion of single-dimensional factors (such as social factors) and fail to analyze the comprehensive impact and mechanism of action of various factors on the spatial distribution of sports facilities from a multi-dimensional perspective.
Based on these, this study takes Tianhe District, Guangzhou City as the case study area, integrates multi-source geospatial data, and employs spatial econometric methods to carry out targeted research. Specifically, the study will first quantitatively evaluate and analyze the spatial pattern characteristics and regional differences of sports facilities from four dimensions: spatial agglomeration, spatial correlation, spatial density distribution, and distribution balance. And then taking sports facility density as the dependent variable, and selecting topography, vegetation, road transport, population density, economic conditions, and land use type as independent variables, a geographical detector model will be constructed based on a uniform grid scale to identify the key factors affecting the spatial distribution of sports facilities and the intensity of their interaction effects. The research results aim to provide scientific basis and practical suggestions for solving the problem of spatial imbalance and optimizing the allocation of sports facilities, thereby promoting the equitable development and sustainable improvement of urban sports services.
Materials and methodsSite and data sources
As an economic hub in Guangzhou, Tianhe District is located in the core of the city. It covers an area of 96.33 square kilometers and has a population of 2.24 million. Its geographical location is shown in Figures 1a,b. The district benefits from transportation advantages, a dynamic sports culture, and rich supporting facilities. The area is equipped with comprehensive “one-stop venue with multiple projects” complexes, encompassing athletic fields, swimming pools, and other facilities. Its large-scale infrastructure provides citizens with a diverse range of fitness and recreational choices. Notably, in 2022, it became the first pilot area in Guangdong Province to successfully participate in the “15-Minute Community Fitness Circle” initiative. In addition, with the rapid development of social economy and the acceleration of life rhythm, people’s demand for a high-quality healthy sports lifestyle has become increasingly prominent (Zasimova, 2022; Rafoss and Troelsen, 2010). As a representative of high-economic-density, high-population-density urban core areas in the Pearl River Delta, exploring the spatial distribution of sports facilities in Tianhe District can provide a typical reference for the precise allocation of sports resources in developed urban core areas, and lay a comparative foundation for subsequent research on sports facility planning in other types of urban areas.
(a) Location of Guangzhou city in China. (b) Location of Tianhe District in Guangzhou; (c) Types and quantities (quantities in parentheses) of sports facility POIs in Tianhe District. Note: Coordinate systems are based on CGCS 2000 (China Geodetic Coordinate System 2000). Subfigure (a) adopts a custom Azimuthal Equidistant Projection (central meridian: 105°E, reference latitude: 35°N); Subfigures (b) and (c) adopt the 3-degree Gauss-Kruger Projection (central meridian: 114°E).
Map series showing Tianhe District in China. Panel (a) highlights China, indicating the location of Guangzhou City. Panel (b) zooms into Guangdong Province, with Tianhe District marked in green. Panel (c) details sports facility points of interest in Tianhe, with categories such as sports venues and leisure services, overlaid on a regional map.
The data utilized in this study includes sports facility data, comprising a total of 1,944 facilities, as indicated in Figure 1c. The data on influencing factors and other related information are listed in Table 1. The sports facility POI data was crawled from electronic maps (Baidu Maps and Amap) in 2025, ensuring high timeliness. This study focuses on the current spatial distribution patterns and their underlying mechanisms of sports facilities, rather than temporal dynamics. To isolate and examine these spatial relationships without the confounding effects of temporal changes, so static temporal datasets (population, land use, terrain, etc.) are adopted.
Data sources for sports facilities.
Index
Data Name
Data sources
1
Sports facility data, Bus stop data, Subway station data
Baidu Maps and Amap
2
Residential community POI data (1,977 entries)
Anjuke
3
Population raster data at 100-m resolution
https://doi.org/10.6084/m9.figshare.27323106.v1
4
Nighttime light data at 500-m resolution
National Earth System Science Data Center
5
Road network data
OpenStreetMap
6
Land use and vegetation cover data at 10-m resolution
National Qinghai-Tibet Plateau Science Data Center Platform
7
Terrain (DEM)
Geospatial Data Cloud
Methods
Specifically, the nearest neighbor index can conveniently determine the agglomeration characteristics of the overall distribution of facilities; kernel density analysis can globally depict the density differences of facility spatial distribution; Moran’s I can analyze the spatial correlation characteristics of facilities; the Lorenz curve can identify the distribution differences of facilities among different regions or groups; the network dimension model can deeply explore the equilibrium differences of facility attributes in both information and spatial dimensions; and the geographical detector can break through the distribution assumption limitations of traditional regression models, enabling the identification of multiple influencing factors and the analysis of their interaction effects. This study combines the above methods, with the core purpose of constructing a progressive research framework of “spatial pattern exploration—internal mechanism analysis” (Figure 2). This framework breaks through the limitation that a single method can only cover part of the research links, ensuring the systematicness and accuracy of the research from the description of spatial characteristics to the analysis of internal mechanisms.
The technical flowchart of the research.
Flowchart depicting a data analysis process. It starts with "Data Sources" including sports facilities, bus/subway stations, community points of interest, road network, and various raster data types. The next step is "Data Preprocessing," involving coordinate transformation, projection, and study area clipping. "Spatial Pattern Research" includes aggregation, correlation, density, and balance analyses. "Factor Classification" separates factors into natural (topography, land use, vegetation) and human (transportation, economy, population, location). "Factor Detection" involves geodetector methods, focusing on single-factor and two-factor interaction detection.Nearest neighbor index
The nearest neighbor index is a statistical measure that relies on precise geographic coordinate data. It is calculated by averaging the distances between each point feature and its nearest neighboring point features (Mitchel, 2025). This index assesses the spatial proximity of sports facilities within their layout, revealing whether their spatial distribution exhibits clustering characteristics. It effectively reflects the spatial distribution patterns of point features. This study employs the Average Nearest Neighbor tool to calculate the nearest neighbor index R for the overall distribution of sports facilities in Tianhe District at the subdistrict level. The index is calculated as Equation 1:R=r1rE=2r1Dwhere R denotes the nearest neighbor index; r1 denotes the actual distance between each point and its nearest neighbor; rE denotes the theoretical nearest neighbor distance; and D denotes point density. When R = 1, the point feature distribution type is random; when R > 1, the point feature distribution type is uniform; when R < 1, the point distribution is clustered.
Kernel density analysis
Kernel density analysis is a technique used to assess the distribution density of sample points across a given region. It employs a “kernel function” to create a smooth density estimate around each sample point. This method examines the spatial distribution of sample points, such as sports facilities, to determine the density of features in their immediate vicinity (Mitchel, 2025). It is extensively used in spatial analysis to map point or linear features, producing density heatmaps that offer a clear visualization of spatial patterns. This study applies kernel density analysis to investigate the spatial distribution of sports facilities at the subdistrict level in Tianhe District. The calculation is as Equation 2:fx=1nh∑i=1nkx−xihnwhere hn is the search radius, and kx−xihn is the kernel function.
Global Moran’s I
Global Moran’s I assesses the spatial autocorrelation of observed values throughout the study area, with outcomes ranging from −1 to 1 (Getis, 2010). A value exceeding 0 indicates positive spatial correlation in the density of sports facilities among neighboring spatial units, suggesting a pattern where “high values cluster with high values, and low values cluster with low values.” On the other hand, a value below 0 implies negative spatial correlation, characterized by an alternating pattern of high and low densities. When the value approaches 0, the distribution of sports facilities seems random, suggesting no significant correlation between adjacent spatial units. The analysis was conducted as Equation 3:I=NW∑i=1N∑j=1Nwijxi−x¯xj−x¯∑i=1Nxi−x¯2where x represents the observed values, w denotes the spatial weights, and W signifies the sum of all spatial weights.
Lorenz curve
The Lorenz curve is a statistical tool that measures the degree of imbalance in data distribution. It employs the cumulative sample proportion as the horizontal axis and the cumulative resource proportion as the vertical axis, reflecting distribution disparities through the shape of its curve. The greater the curve’s deviation from the diagonal line, the more pronounced the distribution imbalance. This study utilizes the Lorenz curve method to analyze the distribution balance of sports facilities. The calculation is as Equation 4:S=∑i=1nYi−50n+1100n−50n+1where S represents the imbalance index; n signifies the number of subdistricts; Yi denotes the cumulative percentage of the ith subdistrict in the ranking of sports facility quantities relative to the total sports facilities, sorted from largest to smallest. The value of S ranges between 0 and 1, with higher values indicating a greater imbalance in the distribution of sports facilities.
Network dimension model
The distribution characteristics of elements integrate the volume dimension and information dimension within the fractal dimension to analyze multidimensional features.
The volume dimension quantifies an evaluation metric in fractal geometry used to describe the complexity of complex shapes or data sets during coverage or measurement (Lacasa, 2013; Wen and Cheong, 2021). It examines the density and complexity of sports facilities in terms of their spatial distribution. By calculating the volume dimension at different scales, it reveals the spatial distribution characteristics of sports facilities in Tianhe District across scales, verifies whether they exhibit fractal characteristics, and further assesses their complexity. A higher volume dimension indicates greater spatial complexity of sports facility distribution, which may reflect more intricate details and irregularities.
The information dimension is a key fractal metric reflecting the informational characteristics, complexity, and internal imbalance of spatial point sets (Lacasa, 2013; Wen and Cheong, 2021). This study uses it to analyze spatial imbalance and information density, thereby assessing complexity. A smaller value indicates high information repetition or redundancy—manifested in similar facility types/quantities and limited diversity across some areas, alongside uneven distribution (dense in some regions, sparse in others). In contrast, a larger value denotes greater diversity and complexity in sports facilities’ spatial layout, which correlates with urban planning, population distribution, and transportation conditions in Tianhe District.
This study adopts a network dimension model to divide Tianhe District into grids of different quantities, investigating the balance characteristics of sports facility distribution at the district level and calculating the volume dimension and information dimension. The number of grids occupied by sports facilities (N(r)) varies with grid size (r). Assuming a scale-free spatial distribution of sports facilities, it is obtained as Equation 5:Nr∝r−a
Define the capacity dimension value a=D0, let Nij denote the number of sports facilities distributed across a grid at a given scale r, and N denote the total number of sports facilities. The probability Pij=Nij/N represents the probability of this distribution, and the information content is given by Equation 6:Ir=−∑ik∑jkPijlnPij
Assuming the spatial distribution of sports facilities is fractal, the expression is as Equation 7:Ir=I0−D1lnrwhere r represents the grid size for equal-ratio division; i and j denote the row and column indices of the grid, respectively; I0 is a constant; D1 is the information dimension.
The geographical significance of the network dimension model is as follows: The network dimension value ranges from 0 to 2. A higher value indicates a more balanced facility distribution, while a lower value indicates greater concentration. When D = 0, sports facilities within the area are concentrated at a single point; When D = 1, sports facilities are uniformly concentrated along a single line; when D = 2, sports facilities are evenly distributed, indicating greater balance. When D0=D1, sports facilities exhibit a simple fractal pattern.
Geodetector
Geodetector is designed to detect spatial variability and uncover its underlying driving forces. It comprises two core functions: factor detection and interaction detection (Wang et al., 2016). The primary objective of Geodetector is to quantify the extent to which a specific factor X accounts for the spatial variation of attribute Y. First, single-factor detection is used to identify influencing factors and assess their significance. Second, interaction detection is conducted to perform an analysis of the interactions between different factors.
In this study, the research area is divided into grid cells of uniform size. Slope, slope aspect, and population density are selected as independent variables, while the kernel density value of sports facilities is designated as the dependent variable. By computing the q-value, the explanatory power of each independent variable is quantitatively evaluated. A higher q-value implies a stronger influence of the corresponding factor on the spatial distribution of sports facilities.
Compared with traditional spatial analysis methods, this model’s notable advantage lies in its nonparametric estimation framework, which effectively mitigates the prevalent interference effects of multicollinearity in traditional regression models. The equation is as Equation 8:q=1−1Nσ2∑i=1LNiσi2where q denotes the influence degree of the influencing factor; L denotes its stratification; N and Ni denote the total number of units in the research area and the number of units in the ith stratum, respectively; and σ2 and σi2 denote the variance of unit density values across the research area and within the ith stratum, respectively.
ResultsSpatial aggregation
The “Average Nearest Neighbor” tool in ArcGIS was used to calculate the average observed distance for public sports facilities in Tianhe District, which was determined to be 70.55 m. The expected average distance is 131.82 m, resulting in a nearest neighbor index of 0.5352 and a Z-score of −39.3975. In terms of distance, the average is below 500 m, indicating a relatively compact distribution of sports facilities in Tianhe District. Residents within this coverage area have high accessibility. With a nearest neighbor index less than 1, it is evident that the sports facilities in Tianhe District display a significant spatial clustering pattern.
Spatial correlation
A spatial autocorrelation study was conducted on the distribution of sports facility points within a 50 × 50-m grid scale in Tianhe District, using the Geoda tool. The grid scale was delineated as the basic geographic spatial unit. The results indicate that the regional global Moran’s I index is 0.439, which is greater than 0, with a Z-score of 25.9 and a P-score of 0.0000. This signifies a significant positive spatial autocorrelation and clustering characteristics in the spatial distribution, as depicted in Figure 3.
(a) Moran’s I scatter Plot; (b) LISA cluster map.
Scatter plot labeled "Moran's I: 0.439" with counts and lagged counts. Accompanying map shows spatial clusters: gray for "Not significant" (1,018), red for "High-high" (154), blue for "Low-low" (279), light blue for "Low-high" (45), and pink for "High-low" (10).Spatial density
The results of the kernel density analysis, as depicted in Figure 4, reveal that the spatial distribution of sports facilities in Tianhe District is characterized by significant spatial heterogeneity. Overall, the distribution pattern is distinguished by a higher density in the western and southern areas, contrasted with lower densities in the eastern and northern zones. The density of facilities increases progressively from the north to the south, indicating a clear “core-periphery” gradient disparity. Facilities are predominantly concentrated in the central and southern regions of Tianhe District, spreading outwards from the core area and displaying marked polycentric clustering and edge-decay phenomena. The vicinity of the Tianhe Sports Center constitutes the largest high-density cluster, acting as the primary aggregation zone for sports facilities in Tianhe District. A secondary cluster has emerged around the Guangdong Olympic Sports Center, whereas sports facilities in the remaining areas are relatively few and far between, showing a low-density, dispersed distribution pattern.
Distribution of sports facility density in Tianhe District. Note: The coordinate system used herein is consistent with that of Figure 1c.
Map illustrating the density of sports facilities using a color gradient. Blue indicates low density, while red signifies high density. A scale bar in kilometers is included.Spatial balanceSpatial fractal characteristics
As shown in Table 2, sports facilities within the study area demonstrate significant scale-free intervals at the measured scale. Their spatial structure adheres to self-similarity patterns, suggesting fractal properties in the spatial distribution of sports facilities. As illustrated in Figure 5, the volume dimension value is 1.712, while the information dimension value is 0.7623, notably lower than the volume dimension value. This suggests that the spatial distribution density of sports resources in Tianhe District varies significantly and possesses a complex hierarchical structure. During self-organizing evolution, these resources exhibit polycentric clustering characteristics, potentially influenced by geographical conditions, the orientation of the transportation network, and disparities in population distribution.
Calculation results for the dimensionality of the sports facility grid.
k
lnk
N(r)
lnN(r)
I(r)
lnI(r)
2
0.6931
4
1.3863
0.9509
−0.0503
3
1.0986
8
2.0794
1.6813
0.5195
4
1.3863
14
2.6391
2.2066
0.7915
5
1.6094
20
2.9957
2.5778
0.9469
6
1.7918
26
3.2581
2.8305
1.0405
7
1.9459
35
3.5553
3.1334
1.1421
8
2.0794
43
3.7612
3.0927
1.1291
9
2.1972
54
3.9890
3.4219
1.2302
10
2.3026
63
4.1431
3.5263
1.2603
(a) Distribution of sports facilities in Tianhe District at various scales; (b) Double-log scatter plot of network dimensions for sports resource distribution in Tianhe District.
(a) Three scatter plots of green dots divided into grids with increasing sections labeled k=2, k=3, and k=10. (b) Graph showing Volume and Information dimensions with ln(k) on the x-axis and values up to 4.5 on the y-axis. Linear trends are indicated, with equations y = 1.712x + 0.2166 and y = 0.7623x + 0.3894.Spatial distribution balance
The imbalance index was calculated using the number of sports facility points per subdistrict, and corresponding Lorenz curves were plotted (Figure 6). Key conclusions are as follows: The imbalance index S of sports facility distribution across Tianhe District’s subdistricts is 0.3229 (<1), indicating a moderate uneven distribution. The Lorenz curve shows a distinct upward convex trend. Eleven subdistricts (Tangxia, Xiancun, Tianhe South, Shipai, Linhe, Wushan, Huangcun, Liede, Tianyuan, Longdong, and Qianjin) account for 74.84% of the district’s total sports facilities. In contrast, Zhuji, Shadong, and Shahe contribute only 3.40%. The subdistricts of Tangxia, Xiancun, Tianhe South, Shipai, and Linhe have a higher proportion and are the primary contributors to this imbalance. Overall, the distribution of sports facilities is somewhat uneven; while some subdistricts lack sufficient resources, the imbalance is not severe.
Lorenz curve depicting the spatial distribution of sports facilities in Tianhe District.
Lorenz curve graph comparing cumulative percentages of two distributions. The red Lorenz curve, representing income inequality, deviates from the green equality line. Villages like Shipai and Shadong are noted on the x-axis, indicating points of economic disparity.Influencing factors detectionInfluencing factors
This study defines the density of sports facility locations as the dependent variable. Based on existing references, a geographic detector analysis was conducted with natural conditions (Liu et al., 2022; Zhang and Qian, 2024; Zhang et al., 2022), road transportation (Bissonnette et al., 2012; Huang and Humphreys, 2014), population density (Carr et al., 2011; Zhang et al., 2022), spatial location (Zasimova, 2022), and economic foundation (Murray, 2009) as independent variables (Table 3).
Factors influencing the spatial distribution of sports facilities.
Factor type
Influencing factors
Indicators
Natural conditions
Topographic conditions
x1: Elevation
x2: Slope
x3: Slope aspect
Vegetation coverage
x4: Vegetation coverage
Urban land use
x5: Land use type
Spatial location
Distance to Residential Areas
x6: Distance to nearest residential area
Transportation factors
Distance to bus stops and subway stations
x7: Distance to nearest transportation facilities
Road network density
x8: Road network density
Economic foundation
GDP value
x9: Average Nighttime Light Data
Population conditions
Population conditions (Zasimova, 2022)
x10: Population density
Single-factor detection
Utilizing the Geodetector, this study performed a quantitative analysis to assess the factors affecting the spatial distribution of sports facilities at the grid level in Tianhe District. The findings, as Figure 7 indicated, reveal that factors x2 (Slope, q = 0.0035) and x3 (Slope aspect, q = 0.0100) have non-significant p-values (p ≥ 0.05), suggesting that slope and slope aspect have relatively weak and statistically insignificant impacts on the spatial layout of sports facilities. Conversely, all other factors have p-values less than 0.05, signifying their significant influence on the spatial arrangement of public sports facilities. In summary, the indicators are ranked by their influence in descending order as follows: Distance to nearest transportation facilities > Distance to nearest residential area > GDP > Population density > Vegetation coverage > Road network density > Land use type > Elevation > Slope aspect > Slope.
Single-factor analysis results of factors influencing the spatial layout of sports facilities in Tianhe District.
Bar graph displaying the influence of ten factors on an outcome, with the highest values for distance to nearest transportation areas (0.4155), distance to nearest residential area (0.3409), and GDP (0.2639). Other factors include population density, vegetation coverage, elevation, slope, slope aspect, land use type, and road network density with varying lower values.
The top four factors are primarily human-related, while the bottom four are natural factors, clearly indicating that human-related factors exert a greater influence than natural ones. At the micro level, regarding human factors, the distances to the nearest transportation facilities (x7) and residential areas (x6) exert the most significant influences, followed by GDP (x9), population density (x10), road network density (x8), and land use type (x5). In order of significance, transportation facilities have the most pronounced impact, followed by the economic base, population distribution, and land use.
Two-factor interaction detection
Building upon single-factor detection, multi-factor interaction detection was conducted to determine the types and strengths of interactions. As shown in Table 4, two-factor interactions were stronger than single-factor effects. When each influencing factor was combined with others, the q-values for both variables markedly enhanced their explanatory power for the dependent variable, demonstrating that bivariate enhancement is the dominant interaction type. The particularly evident interactions involving: x1 (Elevation) with all factors except x2 (Slope); x4 (Vegetation coverage) with x5-x10; x5 (Land use type) with x6-x10; x6 (Distance to nearest residential area) with x7-x10; x7 (Distance to nearest transportation facilities) with x8-x10; x8 (Road network density) with x9-x10; x9 (GDP) with x10 (Population density). The remaining interactions show nonlinear enhancement. Among these, x6 and x7 demonstrate powerful interactive effects with other factors. The most significant interaction is between x7 and x9 (q = 0.4898), followed by x7 and x8 (q = 0.4726). The x2-x7 interaction (q = 0.4408) is significantly greater than the sum of their individual effects (x2: 0.0035, x7: 0.4155). In summary, the spatial distribution of sports facilities is jointly constrained by multiple factors, including transportation conditions, population density, and economic foundation. The significant synergistic effects among factors reveal the complex, nonlinear nature of the mechanisms shaping the layout of urban sports facilities.
Interactive analysis results of factors influencing the spatial layout of sports facilities.
x1
x2
x3
x4
x5
x6
x7
x8
x9
x10
x1
0.0745
x2
0.1109
0.0035
x3
0.0875
0.0397
0.0100
x4
0.1818
0.2120
0.2020
0.1744
x5
0.1311
0.1056
0.1089
0.2017
0.0852
x6
0.3768
0.3617
0.3729
0.4194
0.3606
0.3409
x7
0.4449
0.4408
0.4476
0.4741
0.4329
0.4476
0.4155
x8
0.1803
0.1894
0.2015
0.2847
0.1929
0.4323
0.4726
0.1321
x9
0.2691
0.3181
0.2936
0.3042
0.2920
0.4587
0.4898
0.3374
0.2639
x10
0.2448
0.2230
0.2263
0.2830
0.1955
0.3972
0.4582
0.2905
0.3551
0.1745
Regular font indicates single-factor effects; bold font denotes enhanced interaction between two factors; italic font indicates nonlinear enhanced interaction.
Discussion
The study indicates that the spatial distribution of sports facilities in Tianhe District is significantly affected by proximity to transportation facilities, with an explanatory power (q = 0.4155). The spatial distribution of urban sports facilities is deeply influenced by transportation infrastructure, showing a pattern of facility clustering around transportation hubs. Subway stations and bus stops are central to the spatial layout selection of sports facilities. The nearer the distance to the closest subway station and bus stop, the more concentrated the distribution of sports facilities, which aligns with the “TOD model” (Transit-Oriented Development). Moreover, road network density significantly affects the distribution of sports facilities, with an explanatory power of 0.1321. This confirms the view that the accessibility of sports facilities affects people’s participation in sports, which is also consistent with the findings of existing studies emphasizing the positive correlation between transportation accessibility and sports facility utilization efficiency (Hoekman et al., 2016; Zhang et al., 2022; Zasimova, 2022). It further confirms the crucial role of transportation factors in the spatial distribution of sports facilities, emphasizing their decisive impact on facility location decisions.
The location factor (q = 0.3409) indicates that proximity to residential areas is another key determinant of sports facility distribution. The spatial distribution of sports facilities is closely tied to their spatial proximity to residential areas: the closer a facility is to residential areas, the higher its density. This embodies the “proximity service” principle in the allocation of public facilities. Rational planning of service radii ensures facility accessibility, reflecting the fundamental characteristic of urban sports facility layouts in meeting residents’ daily needs. This is consistent with the orientation of relevant national policies in China. Since 2021, multiple departments including the Ministry of Commerce have jointly promoted the construction of the “15-min convenient living circle” and launched pilot projects nationwide. Targeting community residents, this project aims to meet residents’ basic daily needs, with accessible public service facilities (including sports facilities) as an important support. In addition, the “Talent Plan for Building a Sports Power (2023–2035)” emphasizes continuously developing people-centered sports undertakings, promoting in-depth integration of national fitness and national health, and advancing towards the goals of building a sports power and a healthy China. In line with the above policies, it is evident that in the planning and construction of sports facilities, more consideration should be given to the convenience of residents’ travel to sports. On this basis, scientific and rational optimization of facility layout and resource allocation is required to ensure that sports resources can better serve the fitness needs of the general public. This also further verifies the practical significance of the research conclusion that “proximity to residential areas is a key determinant of sports facility distribution” for guiding the practice of sports facility planning in line with national policy orientations (Zhang et al., 2022; Wang et al., 2023).
Additionally, economic foundations and population distribution strongly influence the distribution of sports facilities, ranking only behind transportation conditions and spatial location. The explanatory power of economic foundations is 0.2639, indicating a robust positive correlation between regional economic development levels and sports facility distribution—higher GDP correlates with more concentrated sports facilities. Population distribution has an explanatory power of 0.1745, indicating a higher density of facilities in areas with greater population concentration. This reflects how the Tianhe District, as one of Guangzhou’s core urban areas, is significantly influenced by economic factors in sports facility distribution. Concurrently, facility construction consistently prioritizes population demand—densely populated zones typically exhibit higher demand for sports facilities, leading to increased distribution density. This embodies the fundamental principles of demand orientation and equitable access to urban resources. Land use exerts a relatively weak influence. However, the perspective proposed by Zasimova. (2022)—that constructing sports infrastructure near workplaces may have a stronger potential impact on increasing physical activity participation among working adults compared to investing in such facilities near residential areas—is worth considering. Building sports facilities around commercial land (e.g., office buildings and other workspaces) might be more conducive to promoting physical activity engagement.
In terms of natural factors, Vegetation coverage (q: 0.1744) has a strong influence. Sports facilities are predominantly located in areas with vegetation coverage ranging from 0 to 0.56. Regions with lower vegetation density and sparse surface vegetation tend to have fewer sports facilities. This pattern mirrorsoptimization of green ecological patterns in the Tianhe District, which are gradually leading to a more rational distribution of sports facilities. To build a refined ‘15-min living circle’, it requires not only upgraded infrastructure but also a green environment that complements it (Zhang and Qian, 2024). However, topography plays a minor role in the spatial distribution of these facilities. Most sports facility sites are situated within an elevation range of 1–63 m. The majority of these facilities are built on flat or gently sloping terrain with a slope gradient of 0–6°. A smaller number are found on slopes with a gradient of 6–12°, and very few exist on slopes exceeding 12°. Facilities on northeast, south, and southwest-facing slopes are more prevalent, whereas those on north, west, and northwest-facing slopes are less common. Overall, the distribution of facilities is relatively even across all orientations. This distribution may be attributed to the north-to-south topographical gradient in the Tianhe District, with lower elevations in the south and higher elevations in the north. This gradient results in three distinct landforms: low mountains and hills, terraces, and alluvial plains, with plains being the most prevalent.
This study takes Tianhe District, the high-economic and high-population density core area of Guangzhou, as the research case, and its conclusions are applicable to developed urban core areas with similar socio-economic development levels, high population agglomeration and mature urban supporting facilities. However, the applicability of the research conclusions to other types of urban areas needs to be cautiously verified, especially for less developed urban districts with relatively weak economic foundation and incomplete sports infrastructure, as well as suburban areas with low population density and scattered residential distribution. The spatial distribution characteristics and influencing factors of sports facilities in these areas are likely to be quite different from those in Tianhe District: for example, the construction of sports facilities in less developed areas may be more constrained by economic investment, while suburban areas may focus more on the layout of large-scale outdoor sports facilities adapted to low population density, rather than the intensive layout of community-based sports facilities in urban cores (Zhang et al., 2022). Different from areas with complex terrain, Tianhe District is an urban core area dominated by plains, resulting in minimal constraints on the spatial layout of sports facilities in the core urban area, which also limits the direct applicability of the research conclusions to mountainous cities or hilly urban areas with complex terrain conditions. Subsequent research can expand the research scope to multiple types of urban areas (including less developed urban districts, suburban areas and mountainous urban areas) to carry out comparative studies, and explore the differentiated spatial layout strategies of sports facilities adapted to different urban development types, so as to further improve the generalizability of the research results and provide more targeted references for the national differentiated planning and construction of urban sports facilities.
In addition, this study primarily focuses on the supply-side spatial distribution of sports facilities, without incorporating demand-side data such as residents’ sports participation behavior, facility usage frequency, and preference for different types of sports venues. This limits the ability to fully assess the matching degree between supply and demand of sports facilities, and to reveal the user-centric mechanisms that influence facility layout. Future research could integrate multi-source demand-side data (e.g., questionnaire surveys, mobile phone signaling data, or facility booking records) to conduct a more comprehensive analysis of the supply-demand relationship, which would provide more targeted insights for optimizing the spatial allocation of sports facilities to better meet residents’ needs. Furthermore, future research could build on this foundation by incorporating longitudinal dynamic data to explore how sports facility patterns evolve in response to urban development and population shifts over time.
Conclusion
This study investigates the spatial distribution characteristics of sports facilities in Tianhe District, employing spatial autocorrelation, nearest neighbor analysis, the imbalance index, the network dimension model, and kernel density analysis. It also uses the Geodetector to quantitatively assess the influence mechanisms and interactions of multidimensional factors, such as transportation conditions, population distribution, and economic levels, on the spatial distribution of sports facilities.
Spatial Distribution Pattern: In general, a spatial pattern characterized by “high density in the west and low density in the east; high density in the south and low density in the north” is observed. The density of facilities gradually increases from north to south.
Spatial Aggregation Characteristics: The results indicate a compact distribution of sports facilities with distinct clustering patterns, forming a multi-centered agglomeration model that exhibits clear spatial radiation effects in the Tianhe District. At the subdistrict level, the distribution of sports facilities shows some imbalance, with a primary concentration in the Tangxia, Xiancun, Tianhe South, Shipai, and Linhe Subdistricts.
Influential Factor Detection: The results indicate that human factors have a significantly more decisive influence than natural factors. Among these, transportation accessibility plays a dominant role in the spatial distribution of sports facilities, while spatial location and economic foundation also exert notable impacts. Furthermore, the spatial layout of sports facilities is constrained by the interaction of multiple dimensional factors, jointly influenced by transportation conditions, population density, economic levels, and other relevant elements. Moreover, concurrently, significant nonlinear synergistic effects exist among the influencing factors, rather than a simple linear additive relationship.
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI was used translate Chinese into English and polish the language expression, but not to generate the any content of the article.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Edited by:Chen Ren, Southeast University, China
Reviewed by:Jianzhou Gong, Guangzhou University, China
Pengfei Tai, Qufu Normal University, China
ReferencesBissonnetteL.WilsonK.BellS.ShahT. I. (2012). Neighbourhoods and potential access to health care: the role of spatial and aspatial factors. Health & Place, Infect. Insecurities18 (4), 841–853. 10.1016/j.healthplace.2012.03.00722503565CarrL. J.DunsigerS. I.MarcusB. H. (2011). “Validation of walk score for estimating access to walkable amenities,” in Original articles. 10.1136/bjsm.2009.069609GetisA. (2010). “Spatial autocorrelation,” in Handbook of applied spatial analysis: software tools, methods and applications. Fischer and arthur getis. Editor ManfredM. (Springer Berlin Heidelberg). 10.1007/978-3-642-03647-7_14HallmannK.WickerP.BreuerC.SchönherrL. (2012). Understanding the importance of sport infrastructure for participation in different sports – findings from multi-level modeling. Eur. Sport Manag. Q.12 (5), 525–544. 10.1080/16184742.2012.687756HoekmanR.BreedveldK.KraaykampG. (2016). A landscape of sport facilities in the Netherlands. Int. J. Sport Policy Polit.8 (2), 305–320. 10.1080/19406940.2015.1099556HuN.ShiG.CaiJ. (2025). The relationship between demographic factors and the geographical distribution of regional sports consumption. GeoJournal90 (3), 1–14. 10.1007/s10708-025-11397-3HuangH.HumphreysB. R. (2014). New sports facilities and residential housing markets. J. Regional Sci.54 (4), 629–663. 10.1111/jors.12120LacasaL.Gómez-GardeñesJ. (2013). Correlation dimension of complex networks. Phys. Rev. Lett.110 (16), 168703. 10.1103/PhysRevLett.110.16870323679650LiH.LiL.YuanLiJiQ.ZhaoJ.GeZ. (2025). A research on spatial configuration characteristics and social performance evaluation of public sports facilities in shanghai based on Geographic Information System (GIS). PLOS One20 (5), e0310585. 10.1371/journal.pone.031058540354497LiuB.TianYuGuoM.TranD.AbdulfattahA. Q. A.XuD. (2022). Evaluating the disparity between supply and demand of Park green space using a multi-dimensional spatial equity evaluation framework. Cities121 (February), 103484. 10.1016/j.cities.2021.103484MitchelA. (2025). The ESRI guide to GIS analysis, volume 2: spartial measurements and statistics. Available online at: https://agris.fao.org/search/en/providers/122621/records/64724fa453aa8c896305d3af (Accessed September 22, 2025).MurrayD. (2009). Reflections on public funding for professional sports facilities. J. Philosophy Sport36 (1), 22–39. 10.1080/00948705.2009.9714743RafossK.TroelsenJ. (2010). Sports facilities for all? The financing, distribution and use of sports facilities in Scandinavian countries. Sport Soc.13 (4), 643–656. 10.1080/17430431003616399ShenJ.ChengJ.HuangW.ZengF. (2020). An exploration of spatial and social inequalities of urban sports facilities in Nanning city, China. Sustainability12 (11), 4353. 10.3390/su12114353WangJ.-F.ZhangT.-L.FuB.-J. (2016). A measure of spatial stratified heterogeneity. Ecol. Indic.67 (August), 250–256. 10.1016/j.ecolind.2016.02.052WangJ.LiJ.ChengJ. (2023). Spatial disparity of sports infrastructure development and urbanization determinants in China: evidence from the sixth national sports venues census. Appl. Spatial Analysis Policy17 (2), 573–598. 10.1007/s12061-023-09557-4WenT.CheongK. H. (2021). The fractal dimension of complex networks: a review. Inf. Fusion73 (September), 87–102. 10.1016/j.inffus.2021.02.001ZasimovaL. (2022). Sports facilities’ location and participation in sports among working adults. Eur. Sport Manag. Q.22 (6), 812–832. 10.1080/16184742.2020.1828968ZhangF.QianH. (2024). A comprehensive review of the environmental benefits of urban green spaces. Environ. Res.252 (July), 118837. 10.1016/j.envres.2024.11883738570129ZhangB.DongY.KelobonyeK.ZhouR. Z.XuZ. (2022). Delineating walking catchment of the existing and proposed public sports facilities with open-source data: a case study of Nanjing. Appl. Spatial Analysis Policy16 (2), 729–749. 10.1007/s12061-022-09499-3