Investigating the Serviceability of Urban Green Spaces from a Spatial Perspective: A Comparative Study Across 368 Cities on the Chinese Mainland

Abstract

Assessing urban green spaces’ (UGSs) serviceability is crucial for ecosystems and well-being, but traditional approaches focus only on the quantity of UGSs while neglecting their spatial configuration or depend on region-specific data sources, significantly limiting their applicability for comprehensive assessments and comparisons of UGSs. To address this problem, we proposed a novel triangular indicator framework for evaluating UGS serviceability from a spatial perspective using public geospatial data. This framework integrated three independent indicators to capture the ecological value and residents’ utilization of UGSs and one composite indicator for the proportion of high-quality UGS services. Our approach was applied across 368 cities in mainland China, and significant geographical differences in UGS provision and usage equality were identified. Cities with similar UGS service characteristics were grouped using clustering, providing tailored improvement suggestions. Lastly, a regression analysis was conducted to compare the proposed indicator system with traditional metrics in relation to economic, demographic, and environmental satisfaction data, highlighting the advantages of our approach and its complementary role alongside traditional ones. This study offers a new method for large-scale UGS evaluation, aiding policymakers in refining UGS distribution, improving environmental equality, and formulating effective planning strategies to promote sustainable urban development.

1. Introduction

Urban green spaces (UGSs) constitute a critical infrastructure component in urban systems, which yield diverse benefits for the urban environment and residents’ well-being, including reducing air pollution and noise, balancing temperature, and enhancing residents’ mental health [1,2,3]. With the rapid urbanization process occurring worldwide, the influx of increasing populations in urban areas has rendered green space a scarce resource [4]. This phenomenon is even more severe in China. According to the 17th National Census of China, the urban population made up 63.89% of the total in China by 2021, and the per capita park area stood at 14.87 square meters, which is 18.8% lower than the global average of 18.32 square meters [5]. Hence, how to better plan UGSs and enhance their service level to meet the immense demands of urban populations has become a critical urban issue in China.

UGS services refer to the public goods and well-being obtained by residents from the green space system [6,7]. Evaluating the service capabilities of UGSs plays a critical role in planning green areas within cities [8]. To date, large-scale assessments of UGS services often rely on conventional indicators such as per capita green space area and the proportion of land covered by green spaces [9,10,11]. For example, the World Health Organization (WHO) considers green space coverage as a key indicator for assessing UGSs’ impact on public health and human well-being [12]. In the Chinese government’s “National Garden City Evaluation Standards”, per capita green space area and green coverage rate are the primary indicators [13]. These traditional indicators have also promoted numerous studies on the derivative effects of UGSs. Yuan et al. [14] explored the impact of green coverage rate on residents’ subjective well-being; Mihara et al. [15] found a significant correlation between green space coverage rate and residents’ pessimistic emotions and attention concentration; and Chen and Hu [16] employed ‘per capita green space area’ to measure residents’ accessibility to UGSs. However, while these traditional metrics provide a macroscopic view of the quantity of urban greenery, they do not fully capture the intrinsic service value and functional benefits that UGSs offer to residents [11]. Even with the same amount of green coverage, a more rational spatial layout of green areas and accessible urban transportation systems can effectively enhance the service level of UGSs [17].

In light of the limitations of traditional “quantity” indicators, considerable efforts have been made to develop more refined metrics to describe the “quality” of UGSs. Related studies can be generally divided into two categories. The first category delves into the landscape of UGSs and employs a variety of indices, such as connectivity, shape, and contagion index, to evaluate the quality of UGSs [18,19,20]. Despite these indices capturing the landscape characteristics of UGSs from various perspectives, they overlook the humanistic perspective of UGS services, that is, to what extent these UGSs are utilized in residents’ daily lives, which means they struggle to comprehensively reflect the quality of service that UGSs provide for residents [21,22].

Another batch of research puts particular emphasis on the actual utilization of UGSs by the population. Among these, a significant body of international literature explores the “usability” of green spaces, focusing on whether UGSs are truly accessible, safe, and comfortable for daily use. These studies often incorporate factors such as physical accessibility, infrastructure, and user perception [23,24,25,26]. However, their assessments often rely on fine-grained data sources, such as field observations or questionnaire-based surveys [26], which constrain their scalability and hinder their applicability in large-scale spatial analyses. Some scholars have sought alternative metrics to assess the actual utilization of UGSs, among which green space exposure is one of the most commonly used measures. This concept of “exposure” may have different interpretations in various contexts, but it primarily represents residents’ potential utilization of UGSs based on their quantity and the distance that residents need to travel to access them [27,28]. For example, Song et al. [28] used the population-weighted average greenery coverage to measure residents’ exposure to UGSs. Green space exposure can effectively capture the service level of UGSs from a human-centric perspective. However, in current studies on green space exposure, the supply of UGSs is primarily quantified by their area [29,30], while their spatial configurations are often overlooked, especially in large-scale investigations. Although some studies have sought to combine green space exposure with UGSs’ spatial configurations for evaluation purposes, these efforts are often impeded by the reliance on fine-grained surveys, such as field investigations with photos of UGS scenes [31], which are difficult to acquire and limit the applicability of their indicators in large-scale assessments.

To address the abovementioned problems, we propose a new triangular framework to evaluate the serviceability of UGSs. The framework defines three independent metrics, namely Connectedness, Availability, and Visuality, to characterize UGSs’ serviceability by their morphological features and spatial relationships with buildings and road networks. Then, by identifying UGSs that simultaneously meet the scopes of the above indicators, a composite indicator Superiority is further defined to reflect the proportion of high-quality UGSs within a city. We apply this framework to 368 cities across mainland China to comprehensively analyze the geographical patterns of UGS serviceability. Additionally, using open population datasets, we explore disparities in exposure to different types of green spaces under this framework and examine associated inequality issues. Finally, we examine the correlations between our indicators and various socio-economic variables, validating the rationality of our approach and highlighting its greater practical implications compared to traditional ‘quantity’ metrics of UGSs. The study will answer the following three research questions: (1) How would a practical indicator system be developed for large-scale UGS serviceability assessment? (2) What are the geographical patterns of UGS serviceability in mainland China? (3) How do regional differences in UGS serviceability inform planning strategies?

The rest of this paper is organized as follows. Section 2 introduces the study area and datasets; Section 3 and Section 4 elaborate on the methodology and experimental results; and Section 5 discusses and concludes the study.

2. Study Area and Data

2.1. Study Area

As shown in Figure 1, we selected 31 province-level divisions encompassing 368 cities in mainland China as the study area. In the context of China’s administrative system, province-level divisions are the highest-level subnational administrative units, which represent all first-level administrative regions in mainland China [32]. Due to differences in data sources and regional development policies, Hong Kong, Macau, and Taiwan were not included in this study. In addition to city-level analysis, the whole study area was divided into four major economic regions, including western, eastern, central, and northeastern, based on the official classification by the National Bureau of Statistics [33]. This regional division reflects broad differences in geographic features, economic development, and urbanization patterns, and was used in subsequent analyses to reveal how UGS serviceability and spatial inequality vary across China’s major economic regions.

2.2. Dataset

In this study, we focused on UGS services, as well as the inequality issues they raise among the Chinese population, in the year 2019. Remote sensing imagery was used to extract UGSs, while building and road network data were employed for related spatial analysis when computing serviceability indicators. To this end, we collected multiple public datasets to facilitate the investigation. The details of these datasets are outlined in

Table 1. Description of main data used in this study.

Data Type	Source	Spatial Resolution
Satellite imagery	Landsat-8	30 m
Building rooftop area	Sentinel-2 imagery	2.5 m
Road network	OpenStreetMap	N/A
Gridded population	WorldPop Project	100 m

.

2.2.1. Satellite Imagery

The Landsat 8 Operational Land Imager (OLI) Surface Reflectance Tier 1 product, which is extensively applied in land cover research, was acquired to delineate the spatial distribution of UGSs in this study. These images contain five visible and near-infrared bands, one near-infrared (NIR) band, and two shortwave infrared 1 (SWIR1) bands, with a spatial resolution of 30 m. All remote sensing images were acquired and processed on Google Earth Engine (GEE, https://earthengine.google.com/ (accessed on 6 November 2023)) platform.

2.2.2. Building Rooftop Area Data

The China Building Rooftop Area (CBRA) dataset was utilized to locate inter-city buildings and identify the UGSs surrounding them. This allowed us to explore the availability of UGSs around residential and workplace areas. CBRA is the first full-coverage building rooftop area dataset in China, which is produced based on Sentinel-2 images at a spatial resolution of 2.5 m. The OA and F1 scores of CBRA exceed 82% and 62%, respectively, which ensures the reliability of related analysis [34].

2.2.3. OSM Road Network

We retrieved road network data from OpenStreetMap (OSM, http://download.geofabrik.de/ (accessed on 9 October 2023)) to identify UGSs adjacent to road segments, which will be used to measure the service capacity of UGSs in residents’ travel activities. OSM is the world’s largest open vector dataset and is widely used in studies related to Chinese cities [35]. OSM’s road network comprises fundamental spatial information, including latitude, longitude, road types, and road names. Given that this study focuses on the service performance of UGSs during people’s daily commutes, we excluded motorways from the road network to prevent potential overestimation in subsequent analysis.

2.2.4. Population Data

Population data were retrieved from the WorldPop project to explore the spatial inequality of UGS services in this study. WorldPop is one of the most commonly used global population datasets, offering population information at a resolution of up to 100 m with a relative error of approximately 6% in urban settings [36]. To reduce computational costs, we leveraged the $1 \mathrm{km}\times 1 \mathrm{km}$ population grid for China from the WorldPop dataset to conduct our exploratory analysis [37].

3. Methods

Figure 2 presents the workflow of this study, which contains three main steps. Firstly, we extracted the pixels belonging to UGSs from satellite imagery based on the official statistics of green coverage reported by the Chinese government. Secondly, we adopted clustering analysis methods to explore the spatial patterns of city-level UGS services and the inequality issues associated with different types of UGSs. Finally, we conducted a regression analysis to explore the correlations between our indicators and various socio-economic metrics, which aimed to evaluate the effectiveness of our indicator framework and its practical implications compared to traditional “quantity” metrics (i.e., green coverage rate).

3.1. Extraction of Green Space

The Normalized Difference Vegetation Index (NDVI) is an effective remote sensing index for distinguishing UGSs at a large scale [38]. Mathematically, the NDVI can be calculated as the ratio of near-infrared (NIR) to red (R) reflectance, as depicted in Equation (1) [39], as follows:

$$\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}={\displaystyle \frac{\left(\mathrm{N}\mathrm{I}\mathrm{R}-\mathrm{R}\right)}{\left(\mathrm{N}\mathrm{I}\mathrm{R}+\mathrm{R}\right)}},$$

(1)

To delineate the UGS range within the study area, we generated the mean NDVI map for the study area based on the Landsat 8 imageries in the year 2019. Furthermore, we incorporated green coverage statistics supplied by the Chinese government to establish an NDVI threshold. This threshold was set to ensure that the area of pixels surpassing it aligned with the officially documented UGSs. Finally, we reclassified the NDVI map based on this threshold to obtain the UGSs as the data basis for further analysis.

3.2. Indicator Framework for UGS Serviceability Assessment

To evaluate UGS serviceability at the city level, this study introduced three key indicators, namely, Connectedness, Visuality, and Availability, along with an integrative metric, Superiority, which generally reflects the integration of the other three. Meanwhile, we classified UGSs based on the characteristics covered by different indicators, ultimately identifying four categories of UGSs, that is, Connected, Roadside, Neighborhood, and Superior UGSs. The proposed indicators, as well as their calculation methods, are summarized in

Table 2. Description of the proposed indicator framework for UGS serviceability.

Indicator	UGS Category	Calculation Method
Connectedness	Connected	The ratio of connected green space pixels over 10 hectares to the total area of green space pixels.
Visuality	Roadside	The ratio of the green space pixel area in the road buffer to the total area of the urban road buffer.
Availability	Neighbourhood	The ratio of the green space pixel area in the building buffer to the total area of the urban building buffer.
Superiority	Superior	The ratio of the area of green space satisfying the above three conditions to the total area of green space pixels.

. The details for each indicator are introduced in the following subsections.

3.2.1. Connectedness

Many studies have emphasized the importance of green space connectivity. It is generally believed that large, connected green spaces have a stronger appeal to residents, providing a comfortable and relaxing environment while also presenting significant ecological value [40,41]. In light of this, we defined the indicator Connectedness to incorporate the concept of connectivity into the assessment of UGS services.

Mathematically, Connectedness is calculated as the ratio of connected green space area to the total green space area, which can be written as follows:

$$\mathrm{C}\mathrm{o}={\displaystyle \frac{{\mathrm{Area}}_{\mathrm{connected}\_\mathrm{ugs}}}{{\mathrm{Area}}_{\mathrm{ugs}}}},$$

(2)

where ${\mathrm{Area}}_{\mathrm{ugs}}$ represents the pixel area of UGSs within a city and ${\mathrm{Area}}_{\mathrm{connected}\_\mathrm{ugs}}$ represents the green space area that meets the connectivity criteria. Specifically, according to the official green space planning standard of China, which defines 10 hectares as the criterion to distinguish large UGSs [42], any connected green space area (based on an eight-connectivity rule) exceeding 10 hectares will be included in the calculation of ${\mathrm{Area}}_{\mathrm{connected}\_\mathrm{ugs}}$.

3.2.2. Visuality

Roads are the most essential public spaces for residents’ activities. The extent of greenery along roads directly reflects the level of urban green space services provided in residents’ daily lives [43,44]. In this study, we defined the indicator Visuality to reflect the visual service of green spaces along roads during people’s commuting. Particularly, Visuality is calculated as the proportion of green space within a certain buffer of the road network, which can be expressed as follows:

$$\mathrm{V}\mathrm{i}={\displaystyle \frac{{\mathrm{Area}}_{\mathrm{road}\_\mathrm{ugs}}}{{\mathrm{Area}}_{\mathrm{roadbuffer}}}},$$

(3)

where ${\mathrm{Area}}_{\mathrm{road}\_\mathrm{ugs}}$ represents the area of UGS pixels within the road buffer, ${\mathrm{Area}}_{\mathrm{roadbuffer}}$, based on the visual range of individuals. Given that the visible distance in a pedestrian’s view has typically been set between 20 and 100 m in previous studies [45,46], this study adopted 50 m as the visuality range to buffer streets and calculate the Visuality indicator.

3.2.3. Availability

Apart from roads, buildings are important places for work and daily life. Previous studies have indicated that neighborhoods with a higher availability of accessible green spaces can significantly enhance residents’ overall satisfaction [47]. Therefore, the greenery level around buildings can represent the availability of UGSs during residents’ relatively static activities. Hence, this study defined Availability as a measure of greenery service at the neighborhood scale. Specially, Availability is calculated as the proportion of green space within a certain buffer zone around buildings, which can be expressed as follows:

$$\mathrm{A}\mathrm{v}={\displaystyle \frac{{\mathrm{Area}}_{\mathrm{building}\_\mathrm{ugs}}}{{\mathrm{Area}}_{\mathrm{buildingbuffer}}}},$$

(4)

where Av represents the area proportion of UGS pixels within the building buffer, ${\mathrm{Area}}_{\mathrm{buildingbuffer}}$. Similar to the approach for calculating Visuality, we applied a 50 m buffer around buildings to define the relevant area for analysis.

3.2.4. Superiority

The previous three independent indicators, respectively, describe the service levels of UGSs from the perspectives of morphology, visual benefits, and accessibility. We further defined green spaces that meet all three of the aforementioned indicators as superior green spaces. We believe that these spaces provide both ecological services and a high utilization efficiency for residents. Hence, the composite indicator Superiority is defined as follows:

$$\mathrm{S}\mathrm{u}={\displaystyle \frac{\mathrm{A}\mathrm{r}\mathrm{e}{\mathrm{a}}_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{\_}\mathrm{u}\mathrm{g}\mathrm{s}\cap \mathrm{r}\mathrm{o}\mathrm{a}\mathrm{d}\mathrm{\_}\mathrm{u}\mathrm{g}\mathrm{s}\cap \mathrm{b}\mathrm{u}\mathrm{i}\mathrm{l}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{\_}\mathrm{u}\mathrm{g}\mathrm{s}}}{\mathrm{A}\mathrm{r}\mathrm{e}{\mathrm{a}}_{\mathrm{u}\mathrm{g}\mathrm{s}}}},$$

(5)

where $\mathrm{A}\mathrm{r}\mathrm{e}{\mathrm{a}}_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{\_}\mathrm{u}\mathrm{g}\mathrm{s}\cap \mathrm{r}\mathrm{o}\mathrm{a}\mathrm{d}\mathrm{\_}\mathrm{u}\mathrm{g}\mathrm{s}\cap \mathrm{b}\mathrm{u}\mathrm{i}\mathrm{l}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{\_}\mathrm{u}\mathrm{g}\mathrm{s}}$ denotes the area of pixels that simultaneously meet the criteria for Connectedness, Visuality, and Availability.

3.3. Spatial Inequality in People’s Utilization of UGS Services

To investigate spatial inequality issues in residents’ use of different types of UGSs, we utilized 1 km resolution population grid data along with the distribution of the target UGSs within each grid to construct a city-scale Lorenz curve, where the cumulative proportion of the target UGSs is plotted against the cumulative proportion of population within the city. As shown in Figure 3, the Gini index was then calculated based on the area proportions of regions A and B in the Lorenz curve, $\mathrm{G}\mathrm{i}\mathrm{n}{\mathrm{i}}_{\mathrm{x}}={\displaystyle \frac{{\mathrm{A}}_{\mathrm{x}}}{{\mathrm{A}}_{\mathrm{x}}+{\mathrm{B}}_{\mathrm{x}}}}$, which was used to quantify spatial inequality of the serviceability, ranging from zero to one, where zero indicates perfect equality and one indicates the maximum inequality. In this study, a higher Gini index indicates a greater degree of spatial inequality in UGS serviceability, where $\mathrm{x} \in \{\mathrm{Co}, \mathrm{Vi}, \mathrm{Av}, \mathrm{Su}\}$, with $\mathrm{A}$ representing the area between the curve and the line of equality and $\mathrm{B}$ representing the total area below the Lorenz curve.

4. Results and Analysis

4.1. Evaluation of UGS Serviceability Under the Indicator Framework

4.1.1. Spatial Analysis on Different Indicators of UGSs

The evaluation results for Connectedness, Visuality, and Availability and their boxplots were grouped by economic regions and are shown in Figure 4. It can be observed that the three indicators all exhibited a clear north–south disparity in China. Upon a closer examination of individual cities, notable disparities emerged among different indicators and across various urban centers. As shown in the boxplot in Figure 4a, Connectedness was generally high across the entire study area, particularly in the central region, with values ranging from 0.5 to 1. This suggests that the connectivity of UGSs in mainland China is generally strong, which offers significant potential for ecological services [48]. The minimum value occurred in Jiuquan, with a value of zero. This reflects a broader pattern in the northwest of mainland China, where arid climates and desertification make it difficult to establish large-scale connected green spaces. In contrast, the maximum value of 0.99 was observed in Shennongjia, demonstrating its extremely high ecological potential for green space. It is worth noting that some economically developed cities, such as Tianjin and Shanghai, exhibited a relatively poor performance. In this regard, these cities need to place greater emphasis on improving green connectivity and enhancing their ecological potential.

The absolute values of Visuality and Availability were relatively lower than those of Connectedness, which indicates that the green infrastructure in most cities still requires improvement. Moreover, from a geographical perspective, the northern regions of mainland China exhibited significantly lower indicator values. In particular, the northeastern region showed near-zero Availability scores according to Figure 4c, which indicates a severe shortage of accessible green infrastructure in residents’ daily lives. It is worth noting that, in regions such as the Greater Khingan Rang and Baishan, the level of Visuality was high, while the corresponding Availability was relatively low. This suggests that while the public green space services in these regions are relatively sufficient, the Availability of accessible green spaces within residential areas still needs improvement.

The results of Superiority are shown in Figure 5. Compared to the other indicators, the value of Superiority was significantly low, indicating that most UGSs cannot meet the above conditions simultaneously. This also highlights a fundamental contradiction in urban green space development. On one hand, urbanization often leads to a reduction in large green areas due to land redevelopment and infrastructure expansion. On the other hand, street and neighborhood greenery, which directly serve residents, tend to be fragmented due to high construction and maintenance costs. Therefore, effectively integrating and utilizing naturally formed large-scale green spaces amid urbanization remains a key challenge for most cities in China.

As shown in Figure 6, by ranking all indicators at the provincial level (including five autonomous regions and four municipalities) based on Superiority as the primary key, we obtained some interesting findings that may offer valuable insights for UGS development. (1) Shanghai, which ranked first in Superiority, performed averagely in Connectedness, Visuality, and Availability. A similar situation can be observed in Jiangsu. This indicates that despite having limited UGS resources, Shanghai and Jiangsu, as typically economically developed regions in China, can effectively integrate green space resources through planning efforts to form multifunctional green infrastructure that supports both ecological services and public use. (2) Provinces such as Guizhou, Guangxi, and Yunnan had relatively lower Superiority, although they performed well in the other three indicators. This is in stark contrast to the situation in Shanghai and Suzhou—while these regions possess good green resources, they lack integrated green space facilities, such as medium-to-large community parks, and the overall layout of urban greenery still requires further improvement. (3) Some southern regions, such as Chongqing Municipality and the provinces of Sichuan, Hainan, and Fujian, ranked high across all indicators, which not only confirms their ecological potential, but also highlights the emphasis on greening along roads and their human-oriented design of UGSs in residential areas. (4) Some northern regions, such as the provinces of Qinghai and Inner Mongolia, ranked low across all four indicators. While this can be partly attributed to climatic constraints, it also reflects the underdeveloped state of UGS planning and infrastructure in these regions.

4.1.2. Urban Cluster Characteristics Based on UGS Serviceability Classification

To identify common patterns of UGS serviceability across cities, we applied hierarchical agglomerative clustering based on our indicators: Connectedness, Availability, Visuality, and Superiority. All indicators were normalized prior to clustering to ensure comparability. We used Ward’s linkage method [49], which minimizes within-cluster variance and is equivalent to using Euclidean distance as the dissimilarity measure. The resulting dendrogram (Figure 7) served as a reference for determining the optimal number of clusters. Based on interpretability and overall structure, we identified four distinct clusters that represent different urban UGS serviceability profiles across mainland China.

The distribution and box plots of all indicators for each cluster are depicted in Figure 8. It can be observed that (1) Cluster 1 exhibited generally high values across all indicators, particularly in Connectedness. Hence, we termed this cluster the “resource-abundant” type. Most of the southwestern regions, southeastern coastal cities, and Greater Khingan were grouped into the first cluster. These areas have abundant green space resources, with a relatively high proportion within urban areas. The UGSs in these areas not only have strong ecological potential, but are also well-structured and efficiently integrated into the urban landscape. (2) The second cluster included several cities from the northwest of China, as well as a few from the northeast. All indicators for this cluster were extremely low, which means that these areas have the weakest UGS service capacity in China. Therefore, we termed this cluster the “resource-constrained” type. (3) Compared to Cluster 1 and 4, Cluster 3 exhibited a moderate performance across most indicators while presenting a consistently high Superiority score. We termed this cluster the “well-planned” type. Many economically developed cities in the central and southern regions, such as Shanghai, as well as provincial capitals like Chengdu, Changsha, Wuhan, and Nanjing, were classified into Cluster 3. This reflects the state of UGS services in most economically developed cities in China—despite having relatively limited green space resources and large, complex built environments, strategic planning efforts have been made to create large contiguous green spaces, such as city parks, within urban areas. As a result, both ecological value and residents’ utilization of UGSs are significantly enhanced. (4) Cluster 4 presented a moderate level of Connectedness but performed poorly in other indicators. We called this cluster the “underdeveloped” type. Most northern cities fell into the fourth cluster. These areas have a relatively rich ecological potential, but the construction of public UGSs within urban areas needs to be significantly strengthened.

4.2. Inequality Characteristics Associated with UGS Services

4.2.1. Analysis of Gini Coefficients for Proposed Indicators in Cities

To investigate the disparities in people’s utilization of different types of UGSs, we calculated the Gini coefficients across all the proposed indicators for each city based on the approach described in Section 3.3. Notably, some cities were excluded from this analysis because they did not have UGSs that met the requirements of particular indicators.

The Gini coefficient results for all indicators are shown in Figure 9. Several observations can be made from this figure.

Based on the distribution of Gini coefficients, we observed significant differences in the inequality of UGS services related to different types. The utilization rate of ‘neighborhood UGSs’, corresponding to the Availability measure, generally received lower Gini values. This suggests that China has made active efforts to improve green spaces in residential areas [50], ensuring relatively equitable access to green resources for residents within their neighborhoods. Similarly, the utilization rate of ‘roadside UGSs’, corresponding to the Visuality measure, also received relatively low Gini values. Thus, roadside greening was well-planned, and residents’ visual access to greenery was relatively equitable.

However, ‘superior UGSs’ exhibited higher Gini values, indicating that, in China, only a small portion of the population can access ‘superior UGSs’. While the quantity of UGSs was emphasized, the quality of these spaces was often overlooked, resulting in highly unequal services for ‘superior UGSs’. We also noted significant differences in the Gini values of the utilization of ‘connected UGSs’ across regions, reflecting large disparities in the planning of ‘connected UGSs’ between cities. This was particularly evident in the western regions, where some urban residents struggled to enjoy equitable access to ‘connected UGSs’.

We observed significant regional differences in UGS service inequality. As shown in Figure 9a, cities in central and eastern China exhibited an unequal utilization of ‘connected UGSs’, despite having abundant green resources. However, ‘roadside UGSs’ and ‘neighborhood UGSs’ showed a lower spatial inequality, indicating a more equitable distribution within residents’ activity areas, likely due to their long development histories, stable economies, and balanced urban planning [51]. Therefore, expanding and improving green spaces in human activity zones has become crucial.

In the northeast, green space resources were scarce and inequality in utilization was high. In the western region, there was a significant difference in inequality between the northwest and southwest. Cities in the southwest, such as Chongqing, had a more balanced resource distribution, while cities in the northwest, particularly those in Xinjiang, Tibet, and Qinghai, faced both limited resources and high inequality in utilization.

4.2.2. Hierarchical Clustering of UGS Utilization Inequality Across Cities

To further analyze the differences in green space utilization inequality across cities, we performed hierarchical clustering on the Gini coefficient results. Based on the dendrogram in Figure 10, three clusters were identified, with their spatial distribution and box plots for all proposed indicators shown in Figure 11. It can be observed that (1) the first cluster included most southern cities and the Greater Khingan region, corresponding to the “resource-abundant” type discussed in the previous section. These areas not only had abundant green space resources, but also ensured spatially equitable access for residents. Therefore, we designated this cluster as the “resource-equal” type. (2) The second-largest cluster included most northern cities, with extremely high Gini values across all indicators. This indicated that this region had the highest UGS utilization inequality in the country. Therefore, we designated this cluster as the “resource-unequal” type. (3) The third cluster included most cities in central and eastern China, as well as special cities like Beijing. This cluster exhibited high Gini values for Superiority and Connectedness, while the Gini values for Visuality and Availability were relatively low. We designated this cluster as the “unequal high-quality resources” type. This reflects the UGS utilization inequality characteristics in many high-income cities in China, where limited ‘neighborhood UGSs’ and ‘roadside UGSs’ are utilized as equitably as possible, but truly high-quality UGSs, which mean ‘connected UGSs’ and ‘superior UGSs’, are enjoyed by only a few. The distance between urban workers and high-quality UGSs remains substantial [52].

4.3. Validation of the Rationality of Our Indicator Framework

A well-designed urban greening index should be able to reflect the differences among cities under varying socio-economic conditions, which will help policymakers and researchers to understand the impacts of greening initiatives to optimize resource distribution. Therefore, to justify the rationality of the proposed indicator framework and its advantages over traditional approaches, we further conducted an analysis of the correlations between different indicator systems and several selected socio-economic indicators. By treating green space indicators as independent variables and each selected socio-economic variable as a dependent variable, we employed the Ordinary Least Squares (OLS) method to establish regression models between them at the city level. All independent variables passed the Variance Inflation Factor (VIF) test. The degree of correlation between these two sets of variables was then quantified based on the goodness of fit (i.e., R-squared value) of the regression model. In particular, we selected green coverage rate (GCR, which represents the proportion of green space to the total land area of the city) as the representative of traditional indicators for comparison.

From the perspective of economic development, we selected two key metrics, per capita GDP (Gross Domestic Product) and population density, both extracted from the authoritative “China Statistical Yearbook 2019” [33].

Table 3. OLS results in relation to per capita GDP with our indicators and GCR.

Independent Variable	Model 1	Model 2
Connectedness	−0.068
Visuality	0.054
Availability	−0.191
Superiority	0.527
GCR		−0.051
Constant	0.139	0.133
R-squared	0.432	0.021

presents the regression results in relation to per capita GDP. Our indicator framework achieved a significantly higher R-squared value of 0.432 for per capita GDP compared to the traditional GCR indicator. Specifically, Superiority showed a positive correlation with per capita GDP, suggesting that high-quality UGSs might contribute to economic development [53]. Conversely, Availability was negatively correlated with per capita GDP. This can be explained by the fact that, in some economically developed cities, extensive and compact built-up areas lead to a relatively smaller proportion of green spaces surrounding buildings [54].

Table 4. OLS models that predict population density using our indicator system and traditional GCR indicator.

Independent Variable	Model 1	Model 2
Connectedness	0.014
Visuality	−0.095
Availability	−0.041
Superiority	0.607
GCR		0.137
Constant	0.097	0.011
R-squared	0.468	0.001

demonstrates the regression results between different indicators and population density. Similar to the results for per capita GDP, our indicator system exhibited a significantly higher R-squared value of 0.468 with population density compared to a traditional GCR with a value of 0.001. Particularly, the Superiority indicator exhibited a robust positive correlation with population density. This finding reinforces the notion that, in some densely populated and economically developed areas, although rapid urbanization may scarify a certain degree of UGS coverage, the rational planning and construction of green spaces have significantly improved the service quality of their greening infrastructure. In that sense, the Superiority indicator might not only reflect the state of urban economic development, but also offer a certain attraction potential for the influx of population.

To further explore the relationships between different indicators and residents’ subjective perceptions of the urban environment, we collected data about residents’ satisfaction with the ecological environment in the Zhejiang and Jiangsu provinces, in light of data availability and the representativeness of these two regions in China’s urbanization. The data were obtained from a standardized survey conducted by the Provincial Department of Ecology and Environment, providing public satisfaction scores with the ecological environment under a uniform standard for each city [55].

As shown in

Table 5. OLS models predict public satisfaction with the ecological environment using our indicator system and traditional GCR indicator.

Independent Variable	Model 1	Model 2	Model 3
Connectedness	−0.518		−0.042
Visuality	−0.159		1.119
Availability	0.218		0.155
Superiority	−0.119		−0.236
GCR		−0.516	−1.432
Constant	0.869	0.689	0.813
R-squared	0.299	0.291	0.423

, our indicator system (Model 1) and GCR (Model 2) showed similar correlations to the dependent variable. However, when all metrics were combined as independent variables (Model 3), the R-squared value increased significantly to 0.423. This indicates that both the actual service capacity of UGSs and the area of UGSs influence residents’ satisfaction with the urban environment. Additionally, in Model 3, only the Visuality and Availability indicators, which are tightly related to urban built-up areas, showed a positive correlation with public satisfaction with the ecological environment, while other indicators, which are conceptually related to the natural environment, exhibited a negative correlation. This suggests that residents are more concerned with the level of greening in their immediate living areas rather than the ecological service capacity of green spaces. Concurrently, the negative correlation observed between GDP and resident satisfaction in Models 2 and 3 indicates that merely increasing the overall greening level of a city does not necessarily lead to an enhancement in resident satisfaction; instead, the key to improving satisfaction lies in the convenience and usability of green spaces within the context of residents’ daily activities, demonstrating the regression results between different indicators and population density.

5. Discussion and Conclusions

This paper proposed a novel triangular indicator system and conducted a detailed evaluation of the service capacity of UGSs in 368 cities across mainland China from the perspectives of ecological function and actual usage by residents. Our research highlighted the differences and potential patterns in the geographical distribution of UGS service capacity and inequality issues in mainland China, and accordingly provided targeted suggestions for improving UGS services in different regions. At the same time, we used regression analysis to compare the correlations between different green space indicators and selected socio-economic indicators, verifying the rationality of the proposed indicators and their relative advantages and complementary roles compared to traditional indicators.

Some interesting findings were yielded in this study. First, it was observed that, for most cities, the Connectedness of UGSs ranged from 0.5 to 1, whereas both Visuality and Availability typically scored below 0.5. This indicates that while the UGSs in China demonstrate a relatively high capacity for ecological services, there is still considerable room for improvement in the construction of green infrastructure in residential living areas and urban public spaces. Furthermore, our analysis revealed that, in certain economically advanced regions, such as Shanghai, despite moderate scores on the three individual indicators, they generally excelled in terms of indicator Superiority. This means that these economically developed areas might have placed a greater emphasis on the efficient use of limited UGSs during their urban development process, and it also indicates that a high GCR does not necessarily promise a high serviceability of UGSs. Notably, a representative case is Greater Khingan in Heilongjiang Province, which lies to the west of the Hu Huanyong Line—a traditional boundary often associated with a sparse population and underdeveloped infrastructure [56]. However, in our analysis, Greater Khingan was classified into the “resource-equal” cluster, with consistently low Gini coefficients across all serviceability indicators. This outcome can be attributed to the region’s rich ecological resources and low population density, which, together, result in a relatively even spatial distribution of UGS services. This case highlights the strength of our framework in identifying service equality patterns that do not necessarily conform to conventional geographic or economic classifications, thereby offering a more nuanced understanding of UGS distribution at the national scale. Lastly, the study highlights severe inequality issues in people’s access to high-quality UGSs (i.e., UGSs within the scope of Superiority), which, once again, emphasizes the importance of the scientific and rational planning of UGSs.

Compared to some existing evaluation approaches, our method has the following advantages. (1) The utilization of public data. Traditional studies on the serviceability of green spaces often rely on local survey data or detailed information on green space distribution and attributes. This research, however, is based on publicly available remote sensing imagery and geospatial data, with statistical data derived from government public reports. This enables our approach to be effectively applied to large-scale UGS evaluation tasks and facilitates a comparative analysis of evaluation results across different regions. (2) Ease of implementation. The calculation of our indicators involves only some common spatial analyses, which are highly operable and efficient in computation. This is beneficial for the long-term and large-scale monitoring and assessment of green space service capabilities. (3) Strong interpretability. Based on the results of the regression analysis with socio-economic metrics, our indicators showed a significant correlation with these metrics. Through these indicators, policymakers and researchers can gain a better understanding of how greening measures impact the social and economic development of cities.

There are also some limitations in this study. Firstly, the data used in this study were from the year 2019, and this might not fully reflect the temporal changes in UGS services given the rapid urbanization in China. In future work, incorporating data from recent years could enhance our understanding of the dynamic changes in UGS services. Secondly, subjective parameters, such as the buffer size in computing Availability and Visuality, were used in our approach. While the rationality for the selection of these parameters is discussed, their impact on the evaluation outcomes requires further validation. Thirdly, although the current indicators can reflect the overall service level of UGSs at the macro level, there is a lack of discussion on the individual differences in UGSs at a fine scale. For example, well-developed public facilities can significantly enhance the attractiveness of UGSs, thereby improving their service capacity to residents [57,58]. Lastly, this study did not incorporate topographic factors such as elevation or slope, which could affect both the accessibility and usability of green spaces, particularly in cities with complex terrain [59]. Future studies could integrate topographical data to improve the comprehensiveness of UGS serviceability assessments, especially in the context of geographically diverse regions like China. Moreover, more indicators could be added based on the current framework to enhance its effectiveness for specific evaluation needs.