Assessment of the Temporal and Spatial Changes and Equity of Green Spaces in Guangzhou Central City Since the 21st Century

Abstract

Green space (GS) equity is a crucial component of environmental justice. From the perspective of environmental justice, this study focuses on the equity of GS across sub-districts with varying GDP levels in Guangzhou, quantitatively assessing and comparing GS equity in areas with different development statuses. However, existing research still lacks sufficient exploration of the relationship between micro-scale socioeconomic indicators and GS equity. To address this gap, this study investigates the inequality of GS availability across neighborhoods during the rapid urbanization process in Guangzhou’s central urban area from 2000 to 2020. Key indicators for measuring GS availability—including GS area, per capita GS area, and NDVI—were selected and calculated for each sub-district in 2000 and 2020. This approach reveals spatial disparities in GS distribution between the two years. Subsequently, the Theil index and Gini index were employed to assess the degree of inequality in GS. Using GS area data and NDVI data, this study analyzes per capita GS area and NDVI values across sub-districts with different development levels in Guangzhou’s central urban area. Statistical methods such as the Theil index were then applied to evaluate the equity of these indicators. The findings indicate that between 2000 and 2020, Guangzhou experienced significant urbanization, a notable decline in total GS area, a marked improvement in NDVI values, and a substantial improvement in GS equity. There is a conflict between the supply of green resources and the demand for high-density economic/population centers. This research provides scientific evidence for urban planners and policymakers to promote the equitable distribution and sustainable development of GS.

1. Introduction

Since the 21st century, the rapid growth of the urban population and the swift pace of urbanization have significantly altered land use patterns. Green space (GS) has been continuously replaced by other land uses, leading to a shortage of GS available to urban residents [1,2]. This trend is particularly pronounced in developing countries, especially in China. The urbanization rate in China surged from 36.22% in 2000 to 63.89% in 2020, and it is projected to reach 80% by 2050. However, between 2000 and 2014, GS coverage continued to decline in both old and new urban areas of China’s 98 most densely populated cities [3,4,5,6,7]. As land use conflicts intensify and economic development pressures mount, intensifying demand for GS and urgent environmental equity concerns [8]. Research indicates that 68.3% of Chinese cities face high inequality in GS exposure [9]. The issue of inequality in GS due to urbanization poses a significant threat to environmental justice [10]. The United Nations Sustainable Development Goal 10 emphasizes that inequality threatens long-term socioeconomic development and has a negative impact on poverty reduction and public health [11].

GS is a broad and complex concept [12], usually referring to open spaces in urban areas that are mainly covered by natural and semi-natural vegetation [13,14]. These spaces cover a wide range of Land uses and land cover (LULC), including forests, grasslands, parks, sports facilities, and even agricultural land [15]. The benefits of GS are widely recognized as an essential infrastructure for cities [16].

⯁

Ecological dimension: GSs are critical to maintaining ecological sustainability within urban landscapes, contributing to biodiversity conservation and climate regulation [17]. As vital habitats for diverse flora and fauna species [18], GSs constitute essential ecological infrastructure.

⯁

Social dimension: GSs provide environments conducive to meaningful and positive social interactions, thereby driving social cohesion, equity, and enhanced human health and well-being [19,20].

⯁

Mental health dimension: Multiple psychological metrics demonstrate significant correlations with GS [21,22]. GSs enhance residents’ psychological well-being through mood stabilization and stress reduction [23]. When physical activities occur in green environments, their mental health benefits are substantially amplified [24,25].

⯁

Economic dimension: Nature–culture interactions significantly influence urban residents’ willingness to pay (WTP) for biodiversity conservation in parks [26]. GSs generate substantial economic spillover effects in housing markets [27].

Building on this integrated benefit framework, GS planning and design must prioritize equity in spatial distribution.

Environmental justice is based on the principle that everyone has the right to a healthy environment [28]. Significant progress has been made in the study of GS equity, focusing on urban parks, GS coverage, and vegetation condition [29,30,31,32,33,34,35]. There is general agreement that differences in GS equity tend to exist between different socioeconomic groups in different areas at the urban scale [36]. This inequality is particularly pronounced in countries at different levels of development, where affluent neighborhoods typically have higher diversity, resourcefulness, and vegetation cover [37,38,39], while disadvantaged populations tend to poorer GS characteristics [40,41,42,43,44]. Most previous studies of socioeconomic data have focused on demographics [13,45,46,47,48], ethnicity (race, national origin, religion) [49], economic status (income) [50,51,52,53,54,55], educational attainment [56], occupation, housing status (housing status, percentage of owner-occupied housing, house price) [57], transport ownership [58], etc., while there are fewer studies on Gross Domestic Product (GDP) as an indicator. Socioeconomic indicators represented by GDP influence the government’s allocation of land and financial resources to GSs and thus their spatial layout [59]. Yutian Lu et al. investigated the driving effect of GDP on the spatial equity of GS in 263 Chinese cities and found that GDP increases the equity of community GS and sub-district greening, while economic development and increased foreign population may deteriorate the level of equity in public GS accessibility [60]. Yi Chen et al. examined the spatial patterns of GS, land rent, and population in a select group of 20 Chinese cities, using the Gini index to assess inequality in per capita GS area and ecosystem services, with results showing a negative correlation between per capita ecosystem service inequality and city size (as measured by population and GDP) [61]. However, there is a lack of existing studies examining the relationship between socioeconomic indicators and GS equity at the micro level. Neighborhoods are the main location for residents’ daily activities, and the ecological service function of GS directly affects residents’ daily experience and health [62]. Second, neighborhood-scale studies can reveal the impact of micro-level socioeconomic factors on GS equity and help develop more targeted policy and planning measures to improve GS equity in specific areas [63]. By analyzing GS equity at the neighborhood scale, the interactions between GS and socioeconomic factors can be better understood, thus providing a scientific basis for sustainable urban development.

In study on the equity of GS layout, three indicators, GS area [64,65,66], GS area per capita [61], and Normalized Difference Vegetation Index (NDVI), were selected to comprehensively evaluate the equity of GS layout. Among them, GS area reflects the total amount of GS and is the basic indicator for measuring the scale of GS; GS area per capita focuses on the distribution of GS resources among residents from the distribution dimension and is able to reflect the differences in the accessibility of GS resources by residents in different regions [67]; and the NDVI is a widely used remote sensing indicator based on the spectral contrast between Near-Infrared (NIR, strongly reflected by vegetation) and red light (vegetation absorption) to measure vegetation health and density, with higher values indicating higher vegetation density and better health, thus assessing GS from the quality dimension [40,68,69,70,71]. These three indicators comprehensively characterize GS from the total, distributional, and quality dimensions, respectively, providing a multidimensional perspective for the evaluation of the equity of the spatial layout of GS.

The relationship between micro-scale socioeconomic indicators and GS equity remains understudied in existing research. In order to fill the gap of existing studies, this study focuses on the rapid urbanization process of Guangzhou’s central urban area from 2000 to 2020 and explores the inequality of urban GS availability in various neighborhoods, contributing to the existing literature. GS area, GS area per capita, and NDVI are selected as key indicators of GS availability, and the values are calculated for different sub-districts in 2000 and 2020, respectively. This study reveals the differences in the spatial distribution of GS between the two years and then assesses the inequality in the availability of GS by applying the Theil index and the Gini index. Guangzhou’s urbanization rate increased significantly during this period, and the area and distribution of GS also changed significantly. This study can provide a scientific basis for urban planners and decision-makers to promote the equitable distribution and sustainable development of GS.

General objective: To develop a sub-district-level assessment framework for GS equity, quantitatively analyzing the spatiotemporal evolution patterns and driving mechanisms of GS resource allocation in Guangzhou’s central urban areas between 2000 and 2020.

Our specific objectives are as follows:

• Dynamic spatial–temporal quantification: Characterize spatiotemporal variations in total GS area, per capita GS availability, and NDVI values, identifying causal factors underpinning these changes.
• Multidimensional equity evaluation: Deploy a dual-model framework (Gini coefficient + Theil index) to assess equity evolution across three dimensions: scale, accessibility, and quality.
• Socioeconomic mechanism decoding: Establish a GDP-GS interaction model at the micro-scale (sub-district units) to elucidate correlations between economic development and GS dynamics.
• Precision optimization strategies: Propose targeted pathways—including subsurface GS systems, rooftop greening technologies, and cross-administrative ecological compensation mechanisms—to alleviate resource deficits in high-density urban cores.

2. Materials and Methods

2.1. Study Area and Data Sources

Guangzhou is located in the Pearl River Delta, belonging to the southern subtropical maritime monsoon climate, which is warm and humid, is suitable for the growth of a variety of plants, and provides good natural conditions for the construction of the city. At the same time, Guangzhou, as the economic center of the south, has experienced rapid urbanization and economic development since the beginning of the 21st century, attracting a large number of foreigners; it has a high population density and has seen a corresponding increase in demand for GS. According to the Statistical Yearbook and the Statistical Bulletin of National Economic and Social Development, by the end of 2020, the number of Guangzhou’s household population was projected to grow from 7.01 million in 2000 to 9.85 million in 2020, and the GDP was projected to grow from CNY 238.307 billion in 2000 to CNY 2501.911 billion in 2020 [72]. However, similar to many other cities, there is a spatial mismatch between population and GS in Guangzhou [73]. Therefore, optimizing the spatial layout of GS and providing more convenient and equitable access to GS has become a key task in Guangzhou’s GS planning.

Since the core urban area of Guangzhou has concentrated on the typical characteristics of Guangzhou’s economy, society, and ecology in the process of urbanization, this study refers to the central urban area scope of the Guangzhou Territorial Spatial Master Plan (2021–2035) and delineates the study area as the four districts of Liwan, Yuexiu, Haizhu, and Tianhe, which contain 79 sub-districts. Sub-districts are the basic research unit of this study (Figure 1).

Four data sources were used in this study. The first data source is the China Land Cover Dataset (CLCD), published by Mr Xin Huang of Wuhan University. Based on the 335,709 Landsat data in Google Earth Engine, Mr Huang Xin produced the annual land cover dataset of China, which contains the yearly land cover information of China in the years 1985 + 1990–2020, with an overall accuracy of 80%. The dataset classifies land use types into cropland, forest, shrub, grassland, water, snow/ice, barren, impervious, and wetland, and it identifies cropland, forest, shrub, and grassland as GS [74].

The second data source is remote sensing data downloaded from Google Earth Engine (GEE) for the NDVI mean values in 2000 and 2020 in the downtown area of Guangzhou City, which are all obtained based on Landsat TM/ETM image analysis, with cloud cover less than 5% in the remote sensing images.

The third source of data is the Fifth Census data and the Seventh Census data published by the National Statistical Office [75,76], with the smallest statistical unit of data being the sub-district.

The fourth data source is the China GDP spatial distribution kilometer grid dataset, which is derived from the Resource and Environmental Science Data Registration and Publishing System (RESDPS) and comprises 1 km raster data, with each raster representing the total value of GDP within the grid (1 square kilometer) in units of CNY 10,000/square kilometer [77].

2.2. Methods

2.2.1. Calculation of GS per Capita

The indicator of GS per capita can visually reflect the supply of GS resources at the per capita level and is an important measure of the equity and adequacy of GS, which can help to assess whether GS is able to satisfy the basic needs of the population and whether there are any imbalances between regions [78]. Its unit is usually expressed in square meters per person. The calculation formula is (1)

$$A_{gm}=A_g/N_p$$

(1)

Here, $A_{gm}$ represents the per capita GS area. $A_g$ represents the total area of GS, including park GS, sub-district GS, waterfront GS, etc. $N_p$ represents the total + resident population.

2.2.2. Equity Evaluation

This study employs the Theil index [79] to evaluate the equity of GS availability, NDVI, and FVC distributions, with comparative analysis using the Gini index [80] to ensure methodological robustness. It should be noted that the Theil index proves more effective than simple disparity metrics for decomposing inequalities into between-group (Tb) and within-group (Tw) components. The index calculation follows (2):

$$T(g,p{)}_r={\sum }_{i=1}^n{g}_{i,r}\mathrm{l}\mathrm{n}\left({\displaystyle \frac{g_{i,r}}{p_i}}\right)$$

(2)

where n indicates total residential units, g(i, r) corresponds to per capita GS/NDVI/FVC values for unit i, and pi reflects its population proportion (g denoting the mean). Higher Theil values represent more pronounced distributional inequalities.

The decomposition formula provides greater analytical depth than non-divisible indices (3) [79,81]:

$$T(g,p{)}_r=T_w+T_b={\sum }_{t=1}^m{g}_{t,r}T(g,p{)}_{t,r}+{\sum }_{t=1}^m{g}_{t,r}\mathrm{l}\mathrm{n}\left({\displaystyle \frac{g_{t,r}}{p_t}}\right)$$

(3)

with $m$ being group count and pt, gt representing group-specific proportions and means. For comparative validation, the Gini index was computed through (4):

$$\mathrm{Gini} =1+\left({\displaystyle \frac{1}{l}}\right)-\left({\displaystyle \frac{2}{\bar{G}\times l^2}}\right){\sum }_{i=1}^l\left[(l-i+1)\times G_i\right]$$

(4)

where $G_i$ ranks GS availability values and $\overline{G}$ is the mean GS availability.

3. Results

3.1. Temporal–Spatial Changes and Equity Analysis of GS

LULC transformations between 2000–2020 (

Table 1. LULC change matrix, 2000–2020 (unit: ha).

Land Use Type		2000
		Cropland	Forest	Grassland	Water	Barren	Impervious	Sum (Unit: Hectare)	Percentage (%)
2020	Cropland	4393.80	864.09	7.29	125.91	0.54	10.26	5401.89	16.68%
	Forest	561.96	2502.09	10.98	6.93	1.08	1.44	3084.48	9.53%
	Grassland	2.79	0.00	1.08	0.18	0.27	0.00	4.32	0.01%
	Water	42.39	5.49	0.00	2009.52	0.00	102.69	2160.09	6.67%
	Barren	1.17	0.00	0.09	0.09	1.08	0.00	2.43	0.01%
	Impervious	4411.62	158.4	9.36	181.98	0.45	16,961.76	21,723.57	67.10%
	Sum (Unit: hectare)	9413.73	3530.07	28.8	2324.61	3.42	17,076.15	32,376.78	100.00%
	Percentage (%)	29.08%	10.90%	0.09%	7.18%	0.01%	52.74%	100.00%

) exerted substantial impacts on GS equity in the research area. Impervious surfaces constituted the dominant LULC category (>50% coverage), followed by cropland. As Figure 2 demonstrates, the conversion of cropland to impervious surfaces (4411.62 ha) accounted for 98.43% of total GS attrition (4481.91 ha), primarily driving spatial deprivation of GS resources. This systemic transition disproportionately affected densely populated sub-districts where GS accessibility was already constrained. Secondary impacts stemmed from forest-to-cropland conversion (864.09 ha), which degraded ecosystem service capacity despite nominal GS retention in peri-urban zones.

Figure 3 shows the Spatial distribution of GS in 2000 and 2020. In 2000, the total GS area in the central urban area of Guangzhou was 12,964.25 ha, while in 2020, it was 8483.26 ha, representing a 34.6% reduction relative to the baseline year.

Figure 4 presents the spatiotemporal dynamics of GS distribution at the street-level within four central urban districts of Guangzhou: Haizhu, Liwan, Tianhe, and Yuexiu. Among them, in 2000, the Fenghuang Sub-district (2037.00 hectares), Longdong Sub-district (963.27 hectares), Changxing Sub-district (920.38 hectares), Xintang Sub-district (827.90 hectares), and Huangcun Sub-district (739.97 hectares) in Tianhe District had the largest areas. The total area of these five sub-districts accounted for 42.34% of the total GS area. In 2020, the GS area of each sub-district showed a decline compared with that in 2000.

Figure 5 shows the size relationships and changes in per capita GS area of each sub-district in 2000 and 2020. Although the distributions in the two years showed significant spatial heterogeneity, the spatial patterns of the two years were relatively similar. Specifically, higher values of per capita GS area were concentrated in the urban fringe, while lower values appeared in the urban center, indicating that residents in the urban fringe obtained more GS than those in the urban center. In particular, sub-districts at the urban fringe, such as Fenghuang Sub-district, Guanzhou Sub-district, Longdong Sub-district, Xintang Sub-district, Huazhou Sub-district, and Changxing Sub-district, had higher per capita GS areas, while sub-districts in the urban center, such as Duobao Sub-district, Longjin Sub-district, and Hualin Sub-district, had lower GS areas. The values for these sub-districts in 2000 were 0, indicating that access to GS in these areas was extremely limited.

Figure 5C further demonstrates the changes in the spatial distribution of per capita GS area from 2000 to 2020. The results show that the per capita GS area of the vast majority of sub-districts decreased, among which Fenghuang Sub-district (658.12 m²/capita), Guanzhou Sub-district (421.31 m²/capita), Longdong Sub-district (402.39 m²/capita), Xintang Sub-district (268.23 m²/capita), and Huazhou Sub-district (215.57 m²/capita) at the urban fringe showed the largest declines. The per capita GS areas of Duobao Sub-district, Longjin Sub-district, and Hualin Sub-district remained unchanged, while those of Renmin Sub-district (0.04 m²/capita), Liurong Sub-district (0.13 m²/capita), Beijing Sub-district (0.48 m²/capita), Shamian Sub-district (1.54 m²/capita), and Liuhua Sub-district (5.35 m²/capita) increased slightly.

Given that NDVI is more sensitive to vegetation coverage changes than land use cover data, the NDVI analysis is not limited to GS but extends to all land cover types. The overall average NDVI value increased from 0.3663 in 2000 to 0.4021 in 2020.

Figure 6 shows the Spatial distribution of NDVI in 2000 and 2020. The change in NDVI values were calculated by subtracting the normalized 2000 NDVI composites from the normalized 2020 NDVI composites, both derived from spatially averaged GEE datasets. Comparison with satellite imagery shows that the old urban areas of Guangzhou had significantly higher NDVI increases. The sub-districts with the most obvious average NDVI growth were Datang Sub-district (0.1360), Tianhe South Sub-district (0.1357), Xinghua Sub-district (0.1354), Dadong Sub-district (0.1354), and Sushe Sub-district (0.1316). In contrast, NDVI values decreased in some urban fringe areas, including Zhongnan Sub-district (−0.0730), Hailong Sub-district (−0.0674), Zhuji Sub-district (0.0556), Guanzhou Sub-district (−0.0438), Huazhou Sub-district (−0.0330), Pazhou Sub-district (−0.0285), Dongsha Sub-district (−0.0269), Xintang Sub-district (−0.0142), Chajiao Sub-district (−0.0064), and Huangcun Sub-district (−0.0014).

Figure 7 shows that in 2000, the Gini index of per capita GS was 0.816. The higher the Gini index, the greater the degree of inequality. The value of 0.816 indicates that the inequality in the distribution of GS per capita in Guangzhou was extremely high in 2000. The Theil index of per capita GS in 2000 was 1.397. Similar to the Gini index, a higher Theil index also indicates greater inequality. The value of 1.397 further confirms the significant gap in GS in 2000. In 2020, the Gini index of per capita GS dropped to 0.759. Although the degree of inequality remained high, the decline from 0.816 suggests that the equity in GS distribution improved slightly compared with 2000. The Theil index of per capita GS in 2020 also decreased to 1.110. This decline, compared with 1.397, further supports the observation that the equity in GS distribution had improved by 2020.

The inequality remained high in both years: The Gini index and Theil indices, measured in both 2000 and 2020, were relatively high (close to 1), indicating considerable inequality in the distribution of per capita GS in Guangzhou during both periods. This means that GS was not evenly distributed among residents. Despite persistent distribution inequality, GS equity exhibited consistent improvement across Guangzhou’s central districts from 2000 to 2020. Both the Gini index (from 0.816 to 0.759) and the Theil index (from 1.397 to 1.110) showed a downward trend. This indicates that although significant inequality still existed, Guangzhou made some progress in narrowing the gap in per capita GS use over the past two decades.

3.2. Changes in Urban Economic Development Level

Figure 8 shows the Spatial distribution of GDP in 2000 and 2020. The total GDP for the urban area in 2000 was CNY 40,800 million, increasing significantly to CNY 1,160,000 million by 2020, reflecting substantial economic growth over the two-decade period. In 2000, the five sub-districts with the highest GDP values were Kuangquan (CNY 161.6233 million), Xicun (CNY 161.54 million), Zhanqian (CNY 161.54 million), Nanyuan (CNY 161.51 million), and Liuhua (CNY 161.49 million). Conversely, the five sub-districts with the lowest GDP were Hailong (CNY 73.8444 million), Fenghuang (CNY 83.9342 million), Liede (CNY 97.9466 million), Huazhou (CNY 101.9381 million), and Dongshan (CNY 103.3666 million).

In 2020, the five sub-districts with the highest GDP values were Zhanqian (CNY 114.0174 billion), Guangta (CNY 108.386 billion), Meihuacun (CNY 105.4262 billion), Dongshan (CNY 103.0216 billion), and Nonglin (CNY 101.8794 billion). Conversely, the five sub-districts with the lowest GDP values were Xicun (CNY 12.6196 billion), Huadi (CNY 12.9302 billion), Nanhuaxi (CNY 13.9405 billion), Qiaozhong (CNY 14.1791 billion), and Shiweitang (CNY 15.3763 billion).

The spatial distribution of per capita GDP exhibits a divergent pattern from that of aggregate GDP (Figure 9). In 2020, the sub-districts with the highest per capita GDP were concentrated in Yuexiu District. Notably, Zhuji Sub-district in Tianhe District registered per capita GDP values substantially higher than those in surrounding areas.

In 2000, the five sub-districts with the highest per capita GDP were Shamian (CNY 36,180/capita), Liuhua (CNY 8497/capita), Dongjiao (CNY 7282/capita), Guanzhou (CNY 7227/capita), and Dongsha (CNY 6322/capita). Conversely, the five sub-districts with the lowest values were Hailong (CNY 899/capita), Shipai (CNY 1199/capita), Wushan (CNY 1330/capita), Dongshan (CNY 1473/capita), and Chigang (CNY 1570/capita).

By 2020, the ranking shifted significantly: Liuhua (CNY 605,489/capita) ranked first, followed by Shamian (CNY 561,506/capita), Zhuji (CNY 351,845/capita), Zhanqian (CNY 326,960/capita), and Hongqiao (CNY 267,973/capita). The lowest per capita GDP values were observed in Nanzhou (CNY 15,409/capita), Shiweitang (CNY 16,083/capita), Ruibao (CNY 16,583/capita), Fengyang (CNY 16,976/capita), and Tangxia (CNY 18,227/capita).

3.3. Comparative Analysis of Characteristics Across Urban Regions

In 2000, the median population per sub-district (or statistical unit) was approximately 50,000 residents. Population distribution exhibited relative concentration, with most sub-districts having comparable population sizes. By 2020, the median population increased significantly to approximately 71,000 residents—nearly doubling. The distribution became broader and shifted demonstrably towards higher values, indicating substantial population growth in numerous sub-districts. The 2000 median GDP density was extremely low (approaching zero), signifying limited economic output per unit area across most of the study area. Data distribution was heavily concentrated in the low-value range. In stark contrast, the 2020 median GDP density surged to approximately 4.21 × 10⁹ CNY. Data distribution was considerably broader and strongly skewed towards higher values, reflecting explosive growth. Guangzhou’s economy underwent remarkable development, with GDP density exhibiting statistically significant exponential growth over the 20-year period, significantly boosting economic vitality across all districts.

The median GS area per sub-district in 2000 was approximately 80 ha, with a relatively broad data distribution showing several sub-districts possessing large green areas. By 2020, the median GS area had declined to approximately 51 ha. The data distribution became more concentrated at lower values, indicating a reduction in the number of sub-districts with extensive GS. Despite rapid economic expansion, total GS area generally decreased, with a statistically significant reduction in the median, suggesting urban development and land conversion may have encroached upon existing green areas. The median maximum NDVI value in 2000 was approximately 0.28. By 2020, the median maximum NDVI value had risen to approximately 0.325. Contrasting with the decline in GS area, this increase in median maximum NDVI may indicate enhanced vegetation density, health, or canopy coverage within the remaining or newly established GS—potentially signifying an improvement in GS quality.

The median per capita GS area in 2000 was approximately 30 m²/capita, displaying a relatively broad distribution, with certain sub-districts maintaining comparatively high levels. By 2020, this median had undergone a dramatic, statistically significant (p < 0.001) decline to approximately 10 m²/capita. The distribution became highly concentrated at very low values, indicating a critical shortage of per capita GS resources. The change in per capita GS represents one of the most pronounced shifts observed. Despite potential improvements in NDVI, rapid population growth vastly outpaced any efforts to preserve or augment GS area. This resulted in a precipitous decline of nearly two-thirds in per capita GS, substantially reducing the relative availability of green resources for Guangzhou residents.

Reductions in GS were as follows: Liwan District (71.01%) > Haizhu District (35.78%) > Tianhe District (27.44%) > Yuexiu District (13.44%). Notably, Liwan District experienced a pronounced decline in GS area. These declines, particularly stark in Liwan and Haizhu, stem from intertwined factors: Liwan, a historic district with dense old buildings and villages, saw extensive GS converted to residential and commercial use during urban renewal to boost land efficiency, while Haizhu’s scattered green areas were squeezed by village redevelopment for housing and industrial upgrading. Population growth amplified housing demand—Haizhu’s riverside appeal drew migrants, prompting developers to occupy marginal GS for apartments, while Liwan transformed community greenery into affordable housing or parking lots amid high density. Economic priorities also played a role: Haizhu converted industrial-zone protective green belts for its Pazhou tech hub, and Liwan built commercial complexes around parks to boost commerce. Infrastructure projects like Liwan’s Metro Line 11 and road widening occupied green corridors (some not fully restored), while Haizhu’s flood control works encroached on waterfront greenery. Weaker policy enforcement compared to Tianhe exacerbated this—Liwan’s unregulated historic land use led to illegal occupation, and Haizhu saw construction near its Orchard Reserve, further shrinking green areas. Spatial analysis reveals persistent but diminished disparities in GS coverage across districts between 2000 and 2020 (Figure 10). While the relative ranking of administrative districts by GS proportion remained consistent (Tianhe > Haizhu > Liwan > Yuexiu in 2000; Tianhe > Haizhu > Yuexiu > Liwan in 2020), all districts experienced substantial absolute declines (Tianhe: 55.33%→40.15%; Haizhu: 33.64%→21.61%; Yuexiu: 18.21%→15.77%; Liwan: 27.68%→8.02%). Critically, Liwan District exhibited the most severe reduction (>71% loss since 2000), collapsing to the lowest proportional coverage (8.02%) by 2020—underscoring spatially uneven degradation patterns (

Table 2. GS coverage by district: 2000 vs. 2020.

District	Aggregate Area (Unit: Hectare)	2000 GS Area (Unit: Hectare)	Percentage (%)	2020 GS Area (Unit: Hectare)	Percentage (%)
Tianhe	13,644.83	7550.05	55.33%	5478.45	40.15%
Haizhu	9117.35	3067.42	33.64%	1969.86	21.61%
Liwan	6253.64	1730.84	27.68%	501.81	8.02%
Yuexiu	3381.63	615.93	18.21%	533.14	15.77%

).

In 2000, GS coverage descended as follows: Tianhe District (55.33%) > Haizhu District (33.64%) > Liwan District (27.68%) > Yuexiu District (18.21%). By 2020, the hierarchy remained largely consistent but with substantially diminished proportions: Tianhe District (40.15%) > Haizhu District (21.61%) > Yuexiu District (15.77%) > Liwan District (8.02%). Notably, Liwan District experienced the most dramatic decline—from 27.68% to 8.02% coverage—and consistently occupied the lowest position in both years.

Spatial analysis revealed pronounced inter-district heterogeneity in Guangzhou (2000): Tianhe District functioned as a demographic–economic core, exhibiting both the highest population concentration (densest data distribution) and overwhelming GDP dominance (elevated median, mean, and long right tail). Despite containing the largest absolute GS stock, Tianhe’s per capita GS allocation was constrained by its massive population base. Yuexiu District demonstrated optimal per capita GS provision alongside stable above-average NDVI despite moderate economic/demographic metrics with compact distribution. Haizhu District registered peak NDVI values indicating superior vegetative potential but showed intermediate performance across population, GDP, and per capita UGS metrics. Conversely, Liwan District exhibited systemic deprivation: lowest-ranked in population, GDP (minimal median/mean), NDVI (weakest vegetation potential), and per capita UGS; its uniformly narrow distributional patterns across all metrics revealed homogeneous, district-wide disadvantage.

As is illustrated in Figure 11, over the period of 2000–2020, Guangzhou’s population surged 88% (9.94 M to 18.68 M) via rural–urban migration, concentrating in core districts: Tianhe’s population doubled (1.1 M to 2.24 M), and Haizhu grew 101% (0.9 M to 1.81 M) amid job and riverside appeal. These inflows spiked demand for housing/land, straining GS. In contrast, Liwan saw stagnant growth (2.8%, 0.7 M to 0.72 M), with its 60+ population rising from 12% to 20% and slower income growth. Such shifts worsened spatial disparities: migrant hubs faced green resource pressure from expansion, while stagnant areas like Liwan lacked momentum to maintain green infrastructure, deepening inequities in GS access and quality. In 2020, Tianhe District emerged as the absolute demographic core with significantly higher population density than the other three districts (Figure 12). Liwan and Yuexiu exhibited the lowest population counts, with Yuexiu demonstrating tighter data clustering. Haizhu District occupied an intermediate position but displayed internal population heterogeneity, suggesting uneven intra-district distribution. Economically, Tianhe’s GDP distribution showed severe right-skewness, featuring high-value outliers indicating localized economic hyper-intensity. Yuexiu ranked second with symmetrical GDP distribution, while Haizhu and Liwan lagged considerably—Liwan recorded the lowest mean GDP. Regarding green infrastructure, Tianhe maintained the largest absolute GS area but with high dispersion, reflecting intra-district inequity. Haizhu followed with relatively concentrated distribution. Yuexiu and Liwan possessed minimal GS reserves, though Yuexiu’s distribution was notably uniform. For per capita GS, Haizhu ascended to the leading position. Tianhe showed comparable median values to Haizhu but with right-skewed distribution and elite enclaves (high-value outliers), while Yuexiu declined to third place with clustered distribution. Haizhu demonstrated moderate dispersion with localized high-potential zones. Tianhe displayed intermediate NDVI values and broad dispersion, while Liwan recorded the lowest NDVI without high-value fluctuations. Tianhe’s sustained expansion in economic scale and population carrying capacity necessitates urgent attention to green resource allocation equity. Haizhu’s ecological revitalization achievements—leading in both per capita GS and NDVI potential—likely benefit from strategic riverfront greening initiatives. Yuexiu preserved premium vegetation quality as the historic urban core but faces strict spatial constraints limiting further green/population growth. Liwan requires vigilant intervention against systemic stagnation, with homogeneous low-level performance across indicators signaling inadequate development momentum. Crucially, the shift in the leading position of per capita GS from Yuexiu (2000) to Haizhu (2020) reflects the shift of Guangzhou’s ecological focus towards riverbank corridors.

As illustrated in Figure 13, all districts exhibited improved per capita GS equity by 2020, though intra-district disparities among sub-districts persisted as the primary driver of overall inequity. NDVI distribution demonstrated relatively greater fairness compared to per capita GS yet experienced increased inequity between 2000 and 2020. Mirroring the per capita GS pattern, street-level variations within districts similarly dominated NDVI-based inequities. Notably, Tianhe District recorded the highest Theil index for per capita GS (1.465) in 2000—reflecting pronounced intra-district polarization between core zones and peripheral corridors during rapid urbanization. By 2020, this index decreased by 16.5% to 1.224, signaling measurable but partial improvement.

4. Discussion

4.1. Quantifiable Enhancement in GS Equity

This study demonstrates that rapid population growth and high-density urbanization have exerted pronounced compression effects on Guangzhou’s GS. Marked spatial disparities characterize the green infrastructure system between urban cores and peripheries: Liwan, Yuexiu, and Haizhu districts exhibit critically diminished per capita GS (<15 m²/capita), while suburban areas maintain significantly higher allocations. Accelerated urbanization has driven substantial migration into central urban zones, intensifying demand for built environments and triggering large-scale GS conversion for residential, commercial, and industrial uses [82]. In high-density historic districts like Liwan, GS has undergone significant reduction and fragmentation [83]. This spatial mismatch between population distribution and green infrastructure has generated an uneven resource distribution gradient [84], with core areas experiencing severe scarcity while peripheries retain relative abundance. Moreover, concentrated populations amplify demand for green amenities, while declining aggregate GS has precipitated precipitous drops in per capita availability (>65% reduction in central districts). Consequently, this dual pressure compromises both residential well-being and ecological integrity [85].

Guangzhou’s rapid economic development and GDP growth have generated complex synergistic effects on GS [86]. Positively, economic expansion provides substantial fiscal resources for urban infrastructure, enabling enhancements in greening standards and vegetation quality [87]. This is exemplified by Tianhe District—the emerging CBD and economic core—where select zones maintain comparatively higher NDVI values indicating robust vegetation cover. Paradoxically, areas of intense economic agglomeration simultaneously exhibit critical deficits in per capita GS. High-GDP zones frequently undergo high-intensity development, triggering large-scale GS conversion for built environments [88]. Yuexiu District’s traditional commercial corridors illustrate this contradiction: elevated GDP coexists with diminished per capita green allocation. This spatial mismatch between economic activity and ecological provisioning reflects systemic neglect of environmental stewardship during urbanization, revealing an inherent conflict between green resource supply and the demands of high-density economic/population hubs [89].

Governmental policies have played a critical regulatory and guiding role in the evolution of GS in Guangzhou [90]. Through urban planning frameworks and land use regulations, authorities have standardized the allocation and conservation of urban green infrastructure. The Guangzhou GS System Plan (2021–2035) exemplifies this approach by establishing quantifiable targets—including built-up area greening coverage and per capita park space—providing clear benchmarks for GS development. However, these measures have proven insufficient to counteract intense development pressures in central urban districts, where GS continues to face substantial encroachment.

To promote actionable and equitable GS policies in dense Chinese cities, recommendations are categorized into planning strategies, design interventions, and governance mechanisms, guided by the three dimensions of environmental justice.

Planning strategies (distributive and procedural justice): Implement integrated underground GS systems in dense urban cores to ensure fair spatial distribution [91]. Strengthen cross-jurisdictional regional coordination to enhance green network connectivity and access equity [92]. Regenerate underutilized urban land—especially in areas like Tianhe CBD and Yuexiu—to redistribute GS where it is most needed [92].

Design interventions (distributive justice): Strategically demolish or repurpose non-compliant structures for micro-GS in underserved zones [93]. Deploy rooftop gardens and vertical greening as adaptable, inclusive greening strategies [93]. Convert closed or accessory green areas into accessible pocket parks to enhance everyday interaction with nature [85].

Governance mechanisms (procedural and interactional justice): Promote public–private co-governance models (e.g., “Government Guidance + Enterprise + Community”) to ensure participatory, transparent GS planning [94]. Create incentive systems that engage businesses and communities in long-term, locally responsive stewardship [94]. These strategies offer practical, just, and operable solutions for enhancing urban green equity in China.

4.2. The Land Economics Mechanism Behind GS Shrinkage

The paradox of GDP growth co-occurring with declining per capita GS stems from fundamental land economics mechanisms. According to Ricardo’s Theory of Differential Rent, land allocation hinges on marginal productivity differentials [95]. GDP escalation elevates aggregate land rents, accelerating rent inflation in urban cores and peri-urban zones. As non-revenue-generating assets, GS demonstrates significant bid–rent gaps relative to commercial/residential uses, rendering it expendable during urban expansion(Figure 14).

Within China’s institutional context, local governments face heightened fiscal reliance on land concession revenues. Guangzhou incurred substantial debt obligations following its 2004 Asian Games bid [96,97], intensifying municipal pursuit of land-based revenue. Post-2004, land concession revenues surged from 27.5 percent of municipal fiscal income to 57.9 percent by 2009, peaking at CNY 73.68 billion (constituting 52.11 percent of 2020 fiscal revenue) [98]. Cities experiencing faster GDP growth demonstrate stronger incentives for “revenue-maximizing land disposition,” prioritizing leasable commercial/residential/industrial parcels. Public GS is systematically marginalized under this regime [99]. Consequently, the spatial distribution of GS manifests the following paradoxical duality: economic expansion coupled with degraded per capita green provision.

4.3. GS Equity Improvement

Our results demonstrate that despite decreases in both total GS area and per capita GS availability from 2000 to 2020, Guangzhou has achieved notable improvements in GS equity. The synchronous decline in both the Gini index (0.816→0.759) and Theil index (1.397→1.110) indicates measurable progress in distributional equity. The Gini coefficient and Theil index, while both used to assess inequality, capture different aspects of distribution. The Gini index provides a single-value summary of inequality based on cumulative distributions, making it intuitive and widely used for comparative assessments. In contrast, the Theil index is decomposable, allowing researchers to distinguish between inter-group and intra-group disparities—an essential feature for multi-level spatial equity analysis. Their combined use in this study allows for both a general understanding of GS inequality and a deeper diagnosis of how much of that inequality is due to differences within or between districts. This trend reflects the combined effects of multiple driving factors.

The Gini and Theil indices in our study reveal persistent inequities in Guangzhou’s GS distribution (Figure 7 and Figure 13), consistent with findings from most urban ecosystem service equity assessments [43,61,100]. While all districts showed improved equity metrics, intra-district disparities remain the dominant factor, accounting for over 70% of total inequality. Notably, despite these improvements, both indices remain above levels typically considered equitable, indicating that significant inequities persist in Guangzhou’s central urban areas. This improvement may reflect more scientifically informed urban planning approaches implemented since 2000.

Existing research identifies population density growth as the primary driver of declining per capita green coverage and park provision rates [101]. In Guangzhou’s case, suburban new towns (Panyu, Huangpu) absorbed 62% of the metropolitan area’s 9 million population increase [102]. Initial GS provision lags exacerbated regional disparities (e.g., Tianhe District’s 2000 Theil index of 1.465 revealed core–periphery polarization), though subsequent compensatory projects (e.g., Huangpu Science City Greenbelt) accelerated equity improvements in new towns, while micro-GS interventions addressed historical service deficits in central areas.

A tenfold GDP increase (CNY 250 billion to CNY 2.5 trillion) [95,98] enabled sustained government investment in large-scale parks (e.g., Haizhu National Wetland Park) and an extensive greenway network (3700 km by 2020) [103], significantly expanding service coverage. The 2010 Asian Games served as a pivotal intervention point, with the “Ten-Year Transformation” plan upgrading central urban green systems (e.g., Zhujiang New Town’s Flower City Square). Subsequent “pocket park” initiatives and urban micro-renewals specifically targeted service gaps in older districts, adding 24.6 km² of parkland from 2010 to 2015 [104]. These targeted strategies reduced service deficits in high-density areas, contributing to a 16.5% decline in Tianhe District’s Theil index.

Persistent challenges in Guangzhou’s GS development include old urban cores such as Yuexiu and Liwan districts, where population densities exceeding 35,000 persons/km² [105] severely constrain redevelopment opportunities, forcing reliance on fragmented GS updates. Emerging residential clusters frequently become “equity depressions” as rapid population growth consistently outpaces infrastructure development timelines. Compounding these issues, NDVI metrics reveal lagging equity improvements that reflect an imbalanced distribution favoring service-oriented GS (primarily lawns and shrubs) over ecologically valuable tree canopy coverage, particularly evident in Baiyun District’s 12% loss of ecological land [106]. Current limitations further include the fundamental unsustainability of expansion-based urban models, demonstrated by the decline in built-up area growth to just 1.2% by 2020, coupled with market mechanisms that promote spatial inequities through the privatization of GS in premium developments and persistent land tenure barriers exemplified by urban villages occupying 55% of the available land. These challenges necessitate future strategies focused on implementing precision governance approaches for existing urban spaces while strengthening equity provisions within the “Green Beauty Guangdong” policy framework, particularly for urban village renewal initiatives and brownfield remediation projects.

While this study introduces a novel sub-district-scale equity evaluation framework using open-source data, several limitations warrant deeper discussion. First, the analysis insufficiently distinguishes between intra-street and inter-street equity dimensions, potentially overlooking finer-grained disparities in GS distribution. Second, this study relies primarily on per capita area as a proxy for accessibility, omitting more nuanced accessibility-based indicators such as walking time, barriers, or service area catchments that are increasingly recognized in urban equity research. Third, the model does not account for critical confounding factors—such as housing prices and microclimate variations—which are known to influence both the distribution and development patterns of GS. For instance, higher housing prices often correlate with better-maintained and more abundant GS, while microclimatic differences (e.g., urban heat islands) can shape vegetation coverage and usability. These limitations point to important directions for future research, including the integration of socioeconomic and environmental datasets to improve the explanatory power and applicability of the framework.

5. Conclusions

This study assessed GS equity at the sub-district scale in central Guangzhou from 2000 to 2020 using per capita GS area and NDVI metrics while explicitly incorporating demographic changes. Our results show that rapid urbanization and population growth significantly reduced total GS area in the city, with the most severe losses observed in Liwan District. In contrast, Tianhe District experienced relatively minor declines.

Interestingly, the average NDVI increased from 0.3663 to 0.4021, reflecting improved vegetation density—especially in historic urban centers. However, this greening trend was accompanied by worsening distributional equity in vegetation quality. Despite a decrease in both total and per capita GS availability, equity indicators, including the Gini and Theil indices, improved across all districts. This positive shift likely reflects the impact of post-2000 planning reforms and increased fiscal investment enabled by GDP growth.

Theil index decomposition revealed that intra-district disparities remained the dominant form of inequality, contributing more to overall GS inequity than inter-district differences. Furthermore, the analysis highlighted the dual influence of economic and demographic drivers: while GDP growth supported vegetation improvements and large-scale GS projects, it also intensified development pressure in core urban areas, reducing per capita GS. At the same time, rapid population growth was the main factor behind declining per capita GS availability and increasing equity strain.

This study advances micro-scale equity analysis by integrating demographic data with spatial GS metrics at the sub-district level. The combined use of NDVI and per capita GS indicators, along with decomposable inequality measures such as the Theil index, offers a scalable, open-source framework for diagnosing urban environmental justice issues.

These findings can guide policymakers and urban planners in designing targeted interventions. For instance, areas with high population growth or low per capita GS should be prioritized for micro-GS development or accessibility improvements. Recognizing intra-district inequality is especially important for resource allocation and equity-focused planning.

Future studies should explore GS equity at finer spatial resolutions, such as intra-street levels, to better capture hyperlocal disparities. To establish robust equity frameworks, studies must systematically integrate multisource datasets encompassing pedestrian network-based accessibility metrics (e.g., real-time walking impedance to GS entrances); contextual variables like land price gradients and microclimate exposure indices (including thermal stress gradients); and, crucially, composite social vulnerability indices synthesizing poverty density, renter occupancy rates, and elderly population concentration. This multidimensional integration of built-environment, socioeconomic, and climatic determinants will significantly enhance the precision and policy relevance of equity assessments for targeted interventions in vulnerable communities.