Xi’an is located in the central part of the Guanzhong Plain between the 107 ° 40′ and 109 ° 49′ E longitudes and 33 ° 39′ and 34 ° 45′ N latitudes. It is characterized by the moderate temperate semi-humid continental monsoon environment with hot and humid summers and cold and dry winters. The annual average temperature is 13.3 °C; the annual average rainfall, which is largely concentrated from July to September, is 613.7 mm.

The urban area consists of Lianhu District, Beilin District, Yanta District, Baqiao District, Weiyang District, and Xincheng District. The area has various unique plant species ranging from more than 233 species in 183 genera spread throughout 79 groups. In 2022, Xi’an city spanned 807.57 square kilometers, with a green covering rating of 41.85%. Focusing on the metropolitan area of Xi’an city in 2020, this study covers approximately 600 square kilometers.

Using the normalized vegetation index (NDVI) and vegetation cover statistics derived from satellite remote-sensing data, in this paper, we map the expansion of urban GS vegetation in Xi’an. Studies indicate that the NDVI in the Shaanxi area is rather high in July and August. Thus, for the computation of the NDVI and vegetation coverage data, in this work, we choose satellite remote-sensing images from Xi’an with low cloud cover from July to August.

This research utilizes the Gaofen-1 satellite remote-sensing picture of Xi’an on 6 July 2020, which was sourced from the Geospatial Data Cloud website (http://www.gscloud.cn/). The downloaded photos, at a 30 m × 30 m resolution, show under 5% cloud cover across the study region. ENVI was used for remote-sensing picture preprocessing, including geometry correction, radiometric, and atmospheric adjustment. The aboveground carbon storage in cities will subsequently be calculated using the NDVI.

Research connected the AGC expected outcomes and landscape pattern indicators of many kinds of GSs. The findings of the study revealed a very notable positive link between urban GS AGC and CA, PLAND, as shown in Table 7. The more complicated the patches around the border, the more noticeable the “edge effect”; the bigger the AGC, the more efficient the CS. Looking at the landscape pattern index reveals that the regional GS with the most PLAND, with many huge patches and complicated borders, is the primary source of the amount of carbon storage in the spatial pattern of the five types of GSs. Core patches have the lowest PD and landscape segmentation index; unit patches have the greatest LSI and LPI. Park GSs have an LPI that is second only to that of regional GSs; hence, the AGC of regional GS is the greatest. Although this sample is from a large metropolitan area with less GS and PLAND and less NP, hence lowering the AGC, the LPI of park GSs is second only to regional GSs. Although the PD and DIVISION of the protected GS are the highest, the AGC is somewhat low (Table 8).

In order to compare the carbon sequestration effects of different community types, based on the actual plant communities in the study area, 40 typical plant communities were divided into three types, namely, low-density, medium-density, and high-density, according to their planting density (Table 3) according to their planting density (Table 9).

A significant component of urban carbon sinks, urban GSs, exhibit remarkable spatial variation in carbon storage and density all around (Nowak et al., 2013; Strohbach et al., 2012). The factors include climatic conditions, urbanization development, kinds of GSs, and urban management practices that lead cities to differ substantially in the ratio of GSs carbon storage to total carbon storage (Zhao et al., 2023; Fan et al., 2022; Wang et al., 2021). Earlier studies (Strohbach et al., 2012; Liu and Li, 2012; Wang et al., 2021) indicate that cities with higher carbon density are Sydney (1.74 kg C/m²), New York (1.68 kg C/m²), and Beijing (1.52 kg C/m²), largely as a result of increased green cover and a large number of urban parks. Averaging 1.77 kg/m², Xi’an’s urban GS system in 2020 sequestered 376.97 tons of carbon, which is much less than that of other cities. The Xi’an urban GS’s relatively low overall coverage, which is influenced by climatic factors such as precipitation and urbanization development, is the main reason.

According to the present study, both arbitrarily constituted green spaces (AGS) and protective green spaces (PGS) have low carbon densities compared to most other types of carbon storage areas.

The only drawback with this research is that the argument that landscape connectivity might improve their carbon sequestration potential is based primarily on indirect landscape pattern metrics such as DIVISION and PD, which indicate high fragmentation and isolation in these GS types. Because of the limitations in high-resolution movement or ecological flow data for urban vegetation, the current analysis did not include direct quantitative evidence linking connectivity to carbon density; for example, functional connectivity models or spatial autocorrelation analyses. Figure 11 illustrates the relationship between the average plant height (m) and carbon storage per unit area (kg C/m2) in plant communities of varying density levels.

However, the negative correlation between DIVISION and carbon density across GS types (Table 7) was not statistically significant when viewed for AGS and PGS alone due to sample constraints, yet it conforms to established ecological principles; fragmented patches usually experience edge effects, microclimatic stress, and low ecological resilience, which restrain biomass accumulation, thus decreasing the carbon storage potential. In support of this, previous research showed evidence for connected GSs facilitating good tree growth, diversity, and long-term stability of vegetation, which, taken together, result in increased carbon sequestration (Nowak et al., 2013; Chen et al., 2024).

For example, Chen et al. (2024) showed that GSs of contemporary values in cities store carbon more efficiently than constructed small patches of comparable size, with efficiencies reaching 18%.

Although these authors were not able to carry out direct modeling of connectivity impacts, somewhat integrated landscape pattern analysis and findings from the existing literature represent the best reasonable ground for recommending the enhancement of spatial connectivity between small-scale GSs to enhance their CS efficiency.

The examined area reveals an uneven distribution of GS AGC in Xi’an city. Most of the medium, medium-high, and high coverage areas are in the Chanba region or park GSs, which creates a spatial distribution of high carbon storage around the periphery and low in the core of Xi’an city. Conversely, urbanization has led to consistent construction and hard surface cover on the ground in the city center, hence lowering plant cover. Only in areas with significant plant density, including parks and GSs, can high carbon storage sites be discovered. On the outskirts of the city, where the plant cover is considerable, low development intensity has resulted in a vast number of new parks and regional GSs being constructed. Its significant carbon storage increases the AGC on the outskirts of the city as well. Thus, the more the plant cover, the higher the aboveground carbon storage and carbon density. A good approach is thought to be increasing the carbon storage capacity of urban plants using increased urban GS coverage.

Urban GS carbon storage is highly influenced by many distinct types of GS carbon storage. The main carbon sink for urban GSs is the carbon storage of urban forests. More than 80% of the carbon storage in urban GSs is provided by North American urban forests (Nowak et al., 2013). Park GSs in Sheffield, United Kingdom, make up 40% of the carbon storage of urban GSs (Davies et al., 2011). On the other hand, studies have shown that recommending GS design optimization strategies for different kinds of GSs might boost the CS capability of other GSs. Although their carbon density is lower than that of other types of GSs, Chen et al.'s 2024 study showed that connected GSs make up 56% of carbon storage to 65% of the GS area. Examining the plant structure and the traits of many AGS kinds in conjunction with the recommendation of several optimization design options for different GS types might help to maximize the potential of AGS. The results show that although regional GSs and associated GSs have more overall AGC for Xi’an city than park GSs and regional GSs, the latter have more carbon density. The spatial distribution of carbon storage in urban GSs is influenced by the city size, GS coverage, and management practices (Nowak et al., 2013; Zhao et al., 2023; Liu and Li, 2012). Different types of GSs should be studied without delay, along with optimization strategies to boost the AGS carbon storage.

Although the current study identifies that AGS and PGS exhibit relatively low carbon density despite contributing significantly to total carbon storage, the suggestion that enhancing landscape connectivity could improve their carbon sequestration potential is primarily based on indirect landscape pattern metrics such as DIVISION and PD, which indicate high fragmentation and isolation in these GS types. Direct quantitative evidence linking connectivity to carbon density—such as functional connectivity models or spatial autocorrelation analyses—was not included in this study due to limitations in high-resolution movement or ecological flow data for urban vegetation. However, the observed negative correlation between DIVISION and carbon density across GS types (Table 7), although not statistically significant for AGS and PGS alone due to sample constraints, aligns with established ecological principles; fragmented patches often experience edge effects, microclimatic stress, and reduced ecological resilience, all of which can limit biomass accumulation and carbon storage capacity. This interpretation is supported by prior research showing that connected GSs facilitate improved tree growth conditions, species diversity, and long-term vegetation stability, thereby enhancing CS (Nowak et al., 2013; Chen et al., 2024). For instance, Chen et al. (2024) demonstrated that structurally connected GSs in urban settings increased carbon storage efficiency by up to 18% compared to isolated patches of similar size. Therefore, while direct modeling of connectivity effects is beyond the current scope, the integration of landscape pattern analysis with findings from existing literature provides a reasonable basis for recommending improved spatial linkage among small-scale GSs as a strategy to enhance their CS performance.

Multiple kinds of the GS landscape design affect urban GS carbon storage. Generally, increasing the GS size will increase the CS capacity of GSs (Wang and Shi, 2015). The CS was also much higher than that in other GSs, as the regional GS and its share of the total GS in this study were far larger than that of other GSs. Their size closely relates to the higher carbon storage of protected GSs and other GSs. However, for areas with low GS percentages, it is advised that the GS size be increased. Given the geographical spread of the carbon storage in Xi’an urban GS in this study, it is recommended to stress the layout of larger park GSs in the core region. Low carbon density results from the uneven distribution of supplemental and protective GSs, as well as inadequate fragmentation of GSs. Considering the interaction between small and micro GSs as additional GSs is also rather important when designing a compact, spatially organized GS as a whole.

Park GSs are significant carbon sinks in the majority of cities, typically accounting for 20%–35% of urban GS carbon storage (Davies et al., 2011; Zhao et al., 2023; Wang et al., 2021). The composition of tree species, vegetation structure, and human management methods are the primary factors contributing to the high carbon storage in park GSs. Consequently, it is imperative to assess them independently of the attributes and structure of the vegetation.

This study looked at the relationship between carbon storage in typical plant communities and other markers of associated vegetation structure and characteristics. The results revealed a notable link between carbon storage in typical plant communities and measures of vegetation structure and traits in the study area. The main factors of community carbon storage were the average plant height, average DBH, and community density, which correspond with other findings. Trees’ capacity for CS is significantly influenced by their age and physiological state. Research has shown that young trees develop quickly and fix more carbon per unit time; nevertheless, their size is tiny, and their total carbon storage capacity is restricted. Conversely, elder trees may store more carbon because of their notable biomass even if their annual growth is declining (Stephenson et al., 2014; Nowak et al., 2013). There is a notable positive correlation between the mean DBH and carbon storage. Increasing tree biomass directly enhances the potential for carbon sequestration, hence increasing the carbon storage capacity as the DBH grows (Nowak et al., 2013). Nevertheless, the interaction between the two is not totally clear-cut. Stephenson et al. (2014) demonstrated that the marginal increase of carbon storage slows when the DBH reaches a certain threshold, hence generating a power function or logarithmic curve growth pattern. According to Gülçin and Cecil. (2021), trees in the 20 cm–40 cm diameter range, which make up 60% of the total, are largely responsible for the carbon storage of campus GSs. The rate of carbon storage accumulation decreases as the DBH surpasses 45 cm. The rate of biomass growth decreases and the carbon storage stabilizes when trees’ DBH reaches a certain standard, hence clarifying this. The total CS of plant communities is substantially influenced by the density of the community plants (Zhang et al., 2024). Under the same circumstances, higher density will directly improve biomass and CS. Research has shown that when the density of plant communities increases, so does their CS rate; moreover, there is a very significant positive link between the carbon density of the community and the CS rate (Yang et al., 2024; Zhao et al., 2023; Liu and Li, 2012). Based on their research, Yi and Wang. (2019) discovered that the CS of sites with varying planting densities per unit area is not always preferable. Medium-density plots store more carbon than high-density and low-density plots. This work’s daily CS and planting density revealed an S-shaped curve. Daily carbon sink per unit area of the plant community increased gradually with increasing tree density when the seeding density fell between 0.13 and 0.23 plants/m². Increasing the plant density per unit area appeared to enhance the community’s total carbon fixation capacity. On the other hand, high density might impede the development of plant communities by generating resource limitation and competition for light. Plants thrive in low-resource environments, including water, temperature, light, and space. Members of the plant community create mutual challenges and boundaries among themselves to get the most appropriate environmental resources for their growth. High planting density may squeeze the trees’ growing area and increase mortality rates. Thus, the selection of an appropriate planting density for the community might increase the photosynthetic intensity in the plants, enhance the biomass of the flora, and eventually increase the carbon storage of the plant community.

To maximize the CS potential of GS systems, the GS design should take into account the whole quantitative relationship between spatial landscape patterns and carbon storage in different kinds of GSs. Among urban GSs, regional GSs, which have a high percentage of plant cover, good connectivity, and large GSs, provide the most carbon storage. The site exhibits degraded green space (GS) functions, relies on single-plant planting types, and suffers from poor connectivity among protective green spaces. The impermeable area is very large and the degree of fragmentation is notable even if the number of connected GSs is the highest. Among the park’s most remarkable features are its vast GS, varied species, and high carbon-storage capacity per unit area. Usually made up of single-layer tree planting methods and a lot of hard surfaces, square green landscapes have little plant cover. There are no multilayered tree, shrub, and grass planting plans. Therefore, improving the connection of the whole landscape pattern from the viewpoint of the landscape pattern index study is very crucial if urban GSs are to have more CS capacity. Apart from raising the general carbon storage capacity of urban ecosystems, it might lower the patch separation index, reduce the problem of habitat fragmentation in cities, and increase the patch area.

RS: reduce the impact of the urbanization process and increase high-CS tree species. PS: increase the connectivity of GSs, avoid single-level planting, and ensure the health of trees.

AS: increase the proportion of high carbon-storage species and vertical greening in GSs, improve tree management and maintenance, and enhance CS capacity per unit area. GS: increase the GS of central urban parks or increase their vegetation coverage, add high-CS tree species, and reduce human interference.

SS: optimize the design of the multilayered community structure, increase the proportion of shrubs, and reduce the impermeable area.

The optimization of the DBH structure of trees in urban GSs is also advocated in this study to guarantee that the community structure is made up of tree species with different diameter classes, therefore raising the proportion of trees with a DBH in the 20 cm–40 cm range. The dual goals of species diversity and carbon storage can be attained in urban settings with limited GS resources by choosing suitable community planting densities, boosting the multilayered planting structure of trees and shrubs, and maximizing the CS capacity of plant communities.

The criticisms or the points highlighted as suggestions in the valuable feedback of the reviewers have indeed contributed significantly toward improving the methodological clarity and the scientific basis of the study.

In order to respond to those, the specific paragraphs on the integration and way each approach, that is, remote sensing and ground survey, complementing the others, have been revised to explicitly state that NDVI-based models offer a macroscopic spatial perspective of the AGC distribution across Xi’an, whereas field surveys provide microscopic mechanistic insights into plant community structure and CS processes.

Direct calibration of the NDVI model with field data was restricted by spatial and temporal differences. Field observations, however, interpret and validate the ecological patterns reflected in the remote-sensing results.

Moreover, in addition to strengthening the case regarding landscape connectivity and carbon storage in small-scale GSs (e.g., accessory and protective GS), we extended the discussion with indirect quantitative support, for example, a negative correlation trend between the DIVISION and carbon density, and added recent evidence showing increased carbon sequestration in connected GSs.

Finally, standardization of all cross-city carbon density values in a uniform unit of measurement (kg C/m²) has enabled valid comparisons to be made: Xi’an’s average carbon density (1.77 kg/m²) is now compared with that of Beijing (1.52 kg/m²) and New York (1.68 kg/m²), showing that Xi’an performs slightly better—all this while recognizing differences in GS typology and sources of data. In summary, these revisions together increase the methodological lucidity, analytical rigor, and general applicability of our findings for urban carbon management.

An important aspect of carbon dynamics in urban GS reviewed in this paper is carbon storage variation across different GS types and the impact of plant community structure on sequestration efficiency. In response to the reviewers’ concerns, the manuscript has been improved in three important aspects to support the rationale with higher methodological rigor and greater scientific clarity.

First, by clearly specifying the complementary roles of both remote-sensing and field methodologies, we specify how the NDVI-derived models provide macro-level spatial assessments of the AGC distribution across the entire metropolitan area, whereas the field surveys in the Xi’an World Expo Park provide micro-mechanistic insight into how structural traits such as DBH or planting density impact carbon storage.

Although direct model calibration was limited by spatial mismatches, the field data serve to interpret and validate the ecological patterns observed in the remote-sensing results. Second, we have added indirect quantitative evidence—an observed negative trend between DIVISION and carbon density—to strengthen the argument linking landscape connectivity to CS in small-scale GSs (e.g., accessory and protective GS), along with recent literature showing that ecologically connected GSs enhance carbon storage.

Finally, we have standardized all intercity comparisons on one common unit (kg C/m²)—Xi’an’s average of 1.77 kg/m² was thereafter compared with 1.52 kg/m² for Beijing and 1.68 kg/m² for New York, revealing that although Xi’an was slightly better off, the authors still insist on the fact that this is contextualized with differences in the definition of GS and, hence, the methodology. The revisions made in this manuscript aid in improving the transparency, comparability, and wider applicability of our findings for urban carbon-management strategies.