Urgency, Feasibility, Synergy, and Typology: A Framework for Identifying Priority of Urban Green Infrastructure Intervention in Sustainable Urban Renewal

Abstract

Urban built-up areas confront significant environmental challenges and growing demand for enhanced residents’ well-being. Prioritizing urban green infrastructure (UGI) interventions is crucial for sustainable urban renewal. We propose a six-step framework that integrates urgency, synergy, feasibility, and typology to identify UGI intervention priorities. The framework targets detailed planning units (DPUs) and was applied to Xi’an, China. First, we assess the risks of supply–demand mismatches related to four key urban ecosystem services (UESs), namely air purification, temperature regulation, runoff regulation, and recreation. K-means clustering analysis is utilized to classify the risk typology. Next, we use the dynamic weighting method to diagnose the urgency of comprehensive risk, then evaluate the potential for synergy optimization between DPUs using local univariate and bivariate spatial autocorrelation analysis. The proportion of urban renewal land area in DPUs is employed as an indicator to evaluate the feasibility of the method. Lastly, we adopt the TOPSIS method to identify the priority ranking. Our research reveals that 51.7% of DPUs in Xi’an are at high risk of multiple supply–demand mismatches for UESs and that seven risk types need targeted optimization strategies. The DPUs ranked in the top 30 can be selected as priority UGI intervention units based on urgency, synergy, and feasibility. This study provides a scientific basis for decision making on UGI interventions in sustainable urban renewal.

1. Introduction

Currently, the urban ecosystem is facing dual challenges of high population and intensified environmental risks caused by climate change. As of 2020, approximately 55% of the global population lived in urban areas. This number is expected to reach 70% by 2050 [1]. Urban expansion and densification have resulted in a significant loss and fragmentation of ecological space [2], leading to a decline in the functionality of urban ecosystem services (UESs) [3]. This has resulted in significant spatial heterogeneity of urban issues [4], such as heat islands, urban flooding, and air pollution, further exacerbating social equity issues [5,6]. The severity of environmental problems underscores the importance of maintaining and enhancing the ecosystem services that support human well-being and urban sustainability [7,8,9]. Given that urban populations mainly gather in built-up areas, improving UESs in these areas is crucial in order to enhance human well-being [10]. Urban green infrastructure (UGI) plays a critical role in providing essential local services [2,10,11,12] that cannot be replaced by peripheral ecological space, including runoff regulation [13], air quality regulation [6], climate regulation [14], cultural and aesthetic values [15], etc. UGI is a cornerstone that contributes to a sense of security, happiness, and accomplishment felt by urban residents [16]. Given the scarcity of land resources in built-up areas, identifying priorities for UGI interventions is crucial in order to promote rational investment timing and support relevant policies and funding [8,17,18], enhancing the effectiveness of urban renewal.

Previous studies have proposed various prioritization methods for UGI interventions, from overlay mapping [19]; to utilizing multicriteria decision analysis (MCDA) [20] tools such as the equal-weighted method [21,22,23], the analytic hierarchy process method (AHP) [24] weighted from planning policy [17], expert opinion [25] or different stakeholders [26]; and the technique for order of preference by similarity to ideal solution (TOPSIS) [27]. The risk assessment method [28] has been utilized to specifically evaluate the demand for a single ecosystem service, such as heat mitigation [29,30] or waterlogging regulation [31]. Recently, many researchers have focused on assessing UESs to identify spatial mismatch in supply and demand to allocate UGI [4,17,32], as this method it can link multiple ecosystem services to social welfare benefits [12,33]. For example, Kremer [17] et al. applied spatial multicriteria evaluation (MCE) to evaluate five ecosystem services in New York City to identify priority areas for UGI in four different scenarios. de Manuel et al. [34] proposed an index to assess the mismatch between ecosystem service supply and demand to develop functional UGI. Tran et al. [21] used the GIST model to evaluate multiple ecosystem services to maximize UGI multifunctionality. Wu et al. [32] evaluated the excess demand for ecosystem services in Wuhan, China, to provide suggestions for UGI planning.

In evaluating the supply and demand of UESs, the common approach is to separately quantify the supply and demand of UESs [4,23,35,36]. These analyses often rely on highly precise land use/land cover data. Estimating UESs can be inaccurate if rough data are used in areas with insufficient fine data [37]. Furthermore, due to the difference in the measurement for each UES, the normalized overlay may lead to issues such as ambiguity and inaccuracy of the results [38]. In urban areas where UESs are in high demand but low supply [39], effectively characterizing the supply–demand mismatch and identifying additional demand space is crucial to optimize urban spatial structures [40,41]. Baró et al. [42] used environmental quality standards and other proxy indicators to assess the satisfaction level of ecosystem service demands, which provided a better link with planning management and environmental monitoring standards [32]. By viewing the supply–demand process of UESs as a whole and using the output results of the system, that is, whether it can meet the needs or expectations of society [43], the degree of risk of supply–demand mismatch of ecosystem services can be succinctly and efficiently revealed, providing a solution for areas lacking fine data. Moreover, there is a gap between the scale and accuracy of research units and their practical systems [4,8,37], such as mismatch between evaluation units and planning units, as well as inconsistency between analysis units and data resources. These defects make it challenging to effectively translate research results into planning practices. In recent years, following the reform of the territorial spatial planning system [44], a detailed planning based on detailed planning units (DPUs) has become an urgent priority for practical implementation in China. However, despite the great practical value it holds, there is currently a lack of research focusing on DPUs.

While current research has mostly focused on assessing UESs, it is important to consider not only the urgency of risks but also the constraints of land renewal feasibility [45,46] and the potential of spatial synergies [45,47,48] when assessing the spatial priority of UGI interventions in built-up areas. Although the synergies and tradeoffs between multiple UESs have been commonly analyzed [19,48,49], limited studies have examined the local spatial relation of synergy between DPUs [45]. Implementing UGI requires significant human, material, and financial resources, and synergy can help to build stronger partnerships, improve communication, and foster trust among stakeholders. Cooperation among multiple units facing similar problems simultaneously can help to implement spatially continuous UGI strategies between units, such as greenways and other linear UGI elements that may need to traverse multiple units. Spatial autocorrelation analysis is a useful tool for identifying potential synergies and informing the development of comprehensive and collaborative UGI strategies.

To address the limitations mentioned above, in this study, we propose a stepwise framework that integrates the urgency, synergy, and feasibility of UGI interventions in sustainable urban renewal. The framework comprises six steps: (1) constructing a risk assessment model for the supply–demand mismatch of four major UESs, namely air purification, temperature regulation, runoff regulation, and recreation; (2) distinguishing risk clusters through K-means clustering analysis to categorize the type of UGI interventions based on risk typology; (3) identifying a comprehensive risk assessment of UES supply–demand mismatch using the dynamic weighting method to determine the urgency of UGI interventions; (4) utilizing univariate and bivariate local spatial autocorrelation analysis to identify the synergy potential between DPUs; (5) using the proportion of urban renewable land area in DPUs as an indicator to evaluate the feasibility of UGI implementation; and (6) determining the priority of DPUs for UGI interventions based on the TOPSIS method.

In summary, this study presents a decision-making framework (Figure 1) that integrates urgency, synergy, and feasibility to identify priority DPUs for UGI interventions in urban planning. The proposed framework has demonstrated its practicality and effectiveness in 145 detailed planning units (DPUs) within the built-up area of Xi’an, China. The six-step approach is flexible and adaptable to diverse urban contexts, taking into account the specific challenges and opportunities faced by each city. Overall, this study contributes to ongoing global discussions on sustainable urban renewal, providing a comprehensive and practical solution for decision makers and planners to achieve a more livable and resilient urban environment.

Figure 1. Framework for identifying UGI priority.

2. Materials and Methods

2.1. Study Area

As the core city of the Guanzhong Plain urban agglomeration in western China, Xi’an is not only the capital of Shaanxi Province but also an important national central city. In this study, we focused on the built-up area (563 km²) in the central urban region of Xi’an (Figure 2), which consists of 6 administrative districts and 4 development districts. We selected the DPUs as the research object to link with detailed planning; thus, a total of 145 units were selected. The study area has a warm, temperate, semihumid, and continental monsoon climate, with distinct seasons. During the summer months (June to August) in Xi’an, the average maximum temperature remains consistently between 35 and 40 °C, with the highest recorded temperature reaching 42.9 °C. The average total precipitation from April to October is 488.6 mm.

Figure 2. (a–c) Study area.

Over the past three decades, the built-up area in Xi’an has more than tripled in size due to rapid urban expansion, with the speed of urban land expansion surpassing population growth [50]. Previous natural features, such as vegetation and water bodies, have been replaced or fragmented by the growing density of the impervious surface, including buildings and roads. As a result, the urban ecological space is decreasing at an alarming pace, lowering the functionality of urban systems. This has led to urban diseases, such as urban waterlogging, heat islands, and air pollution, which are now becoming stark realities for the citizens.

2.2. Risk Assessment of UES Supply–Demand Mismatch

Based on the system theory of disasters, a risk assessment model (risk, R) [12,51] was introduced to evaluate the risks of UES supply–demand mismatch on each of three dimensions: hazard (H), exposure (E), and vulnerability (V) (Table 1). The formula is shown below:

R = H × E × V,

(1)

where H represents the level of hazard, which refers to the severity of the disaster’s impact. Environmental and management standards were selected as key threshold values to measure the degree of UES supply–demand mismatch risk [32,42]. E indicates the degree of exposure, which describes the primary vulnerable groups exposed to risks in terms of population and building density and infrastructure. V denotes the stage of vulnerability, which measures the differences in ability to withstand disasters, which can be characterized by the proportion of vulnerable populations, such as children and the elderly.

Table 1. Risk assessment indicators, thresholds, and data resources.

UES	Risk	Indicator	Secondary Indicators	Threshold	Data Sources
Regulating Services	Air Pollution	H	Average Annual AQI Concentration	SO2: ≤60 μg/m3	Xi’an Environmental Quality Monitoring Stations Daily Data (2020) (http://www.cnemc.cn/en/, accessed on 14 July 2021)
				NO2: ≤150 μg/m3
				PM10: ≤70 μg/m3
				PM2.5: ≤35 μg/m3
		E	Population Density	—	Population Data of DPUs (2022) *
		V	Social Vulnerability	—	The Seventh Census Data, Proportion of Elderly and Children (2020) [52]
	Urban Heat Island	H	Urban Heat Island Proportion Index (UHPI)	0.1	Landsat8 Remote Sensing Image (2019) (http://www.gscloud.cn, accessed on 3 August 2021)
		E	Population Density	—	Population Data of DPUs (2022) *
		V	Social Vulnerability	—	The Seventh Census Data, Proportion of Elderly and Children (2020) [52]
	Urban Waterlogging	H	The Density of Waterlogging Points	>0	Waterlogging Point Data (2021) (https://map.baidu.com/, accessed on 28 August 2022)
		E	Population Density	—	Population Data of DPUs (2022) *
			Building Density	—	Building Data (2021) (https://map.baidu.com/, accessed on 28 April 2022)
			Road Density	—	Road Data (2021) (https://map.baidu.com/, accessed on 28 April 2022)
		V	Social Vulnerability	—	The Seventh Census Data, Proportion of Elderly and Children (2020) [52]
Cultural Service	Insufficient Recreation	H	Residential land area not covered by parks and green spaces within the service radius	<1 ha, 300 m	Park Entrance and Exit Data (2022) (https://map.baidu.com/, accessed on 12 September 2021) Road Data (2021) (https://map.baidu.com/, accessed on 28 April 2022) Third National Land Survey Data, Residential Land (2021) *
				1–5 ha, 500 m
				>5 ha, 1000 m
		E	Population Density	—	Population Data of DPUs (2022) *
		V	Social Vulnerability	—	The Seventh Census Data, Proportion of Elderly and Children (2020) [52]

Notes: The data marked with * are not publicly available in China and were obtained from the Xi’an Urban Planning and Design Institute. See Appendix A for detailed overview of the data.

2.2.1. Stress Assessment

• Air Pollution

The AQI (Air Quality Index) is a dimensionless index that quantitatively describes the condition of air quality by integrating primary pollutants, including SO₂, NO₂, PM₁₀, PM_2.5, O₃, and CO, along with the key indicators of exceeding pollutant standards [53]. In this study, AQI was chosen as the comprehensive indicator to evaluate the stress level of air pollution. Daily AQI data from all monitoring stations in Xi’an City in 2020 (http://www.cnemc.cn/en/, accessed on 14 July 2021) were statistically analyzed to derive the annual average concentration. The Kriging interpolation method was used to map the spatial distribution of the annual average AQI concentration, while the zonal statistic method was employed to characterize the stress level of air pollution for each DPU. Since AQI, O₃, and CO do not have lower annual concentration limits, the lower limits of pollutants such as SO₂, NO₂, PM₁₀, and PM_2.5 were used as key threshold values to measure service mismatch levels. DPUs with annual average concentrations of those four pollutants below the threshold value were assigned a stress level of 0, while those with values above the threshold value were reclassified into five levels based on the annual average concentration using the natural breaks method.

• Urban Heat Island

We elected the Urban Heat Island Proportion Index (UHPI) [54] as the index for heat island risk. The UHPI determines the level of heat islands by calculating the proportion of pixel area of each heat island intensity level within the statistical unit. The level of surface urban heat island (SUHI) was calculated as the difference between the surface temperature of a given pixel and the average temperature of suburban farmland [55] and classified into seven levels (Table 2). The formula is expressed as follows:

$${SUHI}_i=T_i-\frac{1}{n}\sum _{j=1}^n{Tsub}_j,$$

(2)

where ${SUHI}_i$ represents the intensity of the surface heat island (°C) of the $i$-th pixel, $T_i$ represents the surface temperature of the $i$-th pixel, $n$ represents the number of effective pixels in suburban farmland, and ${Tsub}_j$ represents the surface temperature (°C) of the $j$-th suburban farmland pixel. The Landsat 8 satellite remote sensing image from 12 August 2019, was obtained (http://www.gscloud.cn, accessed on 3 August 2021), and the surface temperature was derived using the atmospheric correction method with ENVI 5.3 [56]. Subsequently, the average surface temperature of suburban farmland in Xi’an was calculated to be 33.7 °C. Using Formula (2), the heat island intensity for each pixel was computed and classified into seven levels based on relevant literature [44,45] (Table 2). The UHPI of each DPU was calculated as follows (Formula (3)):

$$UHPI=\frac{1}{100m}\sum _i^m{w}_i{p}_i,$$

(3)

where $UHPI$ represents the Urban Heat Island Proportion Index; $m$ is the number of heat island levels; $i$ refers to the ordinal number of the level for which the temperature is higher than that of suburban areas; $w_i$ represents the weight of the $i$-th level, which is equal to the level value; and $p_i$ represents the percentage of the area of the $i$-th level. According to Table 2, $m$ equaled 7, and $i$ equaled 5, 6, or 7. The UHPI value ranged from 0 to 1, with higher values indicating a more severe heat island. Based on the evaluation criteria in the guidelines mentioned above, a key threshold value of 0.1 was set. DPUs with UHPI values below 0.1 were assigned a stress level of 0, while those above this threshold were classified and assigned specific values according to Table 3.

Table 2. The level of surface urban heat island (SUHI) [44,45].

Level	SUHI (°C)	Description
1	≤−7.0	Strong cold island
2	−7.0–−5.0	Moderate cold island
3	−5.0–3.0	Weak cold island
4	−3.0–3.0	No heat island
5	3.0–5.0	Weak heat island
6	5.0–7.0	Moderate heat island
7	>7.0	Strong heat island

Table 3. Stress levels of heat islands [54].

Level	UHPI	Heat Island Level	Stress Level
1	0–0.1	None	0
2	0.1–0.3	Slight	1
3	0.3–0.5	Moderate	2
4	0.5–0.7	Severe	3
5	0.7–0.9	Extremely severe	4
6	0.9–1	Very extremely severe	5

• Urban Waterlogging

The density of waterlogging points was selected as the indicator of the risk caused by urban waterlogging according to the “Urban Physical Examination Indicator System” developed by the Ministry of Housing and Urban-Rural Development [57]. A total of 148 waterlogging points were collected from Baidu Map (https://map.baidu.com/, accessed on 28 August 2022) and cross-checked against data from the Xi’an Urban Administration Bureau and media reports on urban waterlogging over the past five years. The point density of waterlogging was calculated, while the zonal statistic method was employed to characterize the stress level of urban waterlogging. DPUs with a point density of 0 were assigned a stress level of 0, while those above the threshold value were reclassified based on the point density of waterlogging.

• Insufficient Recreation

The accessibility of urban recreation services was mainly related to the size of public green spaces and population density [23]. In this study, accessibility was evaluated using the service radius, which was divided into three categories based on the size of public green spaces [58] within and adjacent to the study area: community park with an area of less than 1 hectare, community park with an area of 1–5 hectares, or comprehensive park with an area of more than 5 hectares. The 5 min, 10 min, and 15 min service radii had corresponding threshold values of 300 m, 500 m, and 1000 m. The entrance and exit points of each park were visually identified using Baidu Map (https://map.baidu.com/, accessed on 12 September 2021), and network analysis was conducted in ArcGIS to determine the actual service areas of the public green spaces. The proportion of residential areas not covered by the service radius of public green spaces within each DPU was selected as the indicator of risks and indicated insufficient recreation. The natural breaks method was employed to classify the DPUs into five levels.

2.2.2. Exposure Assessment

We characterized the level of exposure by selecting population density and obtaining population data from the current statistics of DPUs. It is worth mentioning that the calculation of exposure caused by waterlogging differs from other calculations due to its impact, which not only affects residents’ lives but also causes damage to social resources and infrastructure [31]. Therefore, the density maps of population, building, and road data were normalized, equally weighted, overlaid, and reclassified into five levels to reflect the comprehensive exposure to waterlogging risk.

2.2.3. Vulnerability Assessment

Vulnerability was used to measure differences in the ability to withstand disasters. On the one hand, high-temperature heat waves, air pollution, and flood risks more severely affected the elderly and children, making them more vulnerable [5]. On the other hand, recreation accessibility had a more significant impact on the frequency of use by vulnerable groups. Thus, the normalized proportions of the elderly population and child population were selected and equally weighted, and the social vulnerability index was obtained through reclassification.

2.3. Priority Identification of UGI Intervention

2.3.1. Risk Typology

SPSS Statistics 25.0 was employed to perform the K-means clustering analysis, which classified the risk types of each DPU. The optimal clustering number was determined using the “elbow method” [23], and the 145 DPUs were categorized into distinct risk types. Additionally, eCognition 10.2 software was utilized to analyze the UGI distribution data extracted from Gaofen-1 (GF-1) imagery (http://www.gscloud.cn, accessed on 5 October 2021) with a spatial resolution of 2 m in the study area. In addition, the correlation between the proportion of the UGI area and the risk level of each DPU was calculated.

2.3.2. Urgency Assessment

To identify the urgency of UGI interventions, comprehensive risk assessment requires the weighting of the four risks. At present, static weight assignment methods, including equal weighting, AHP, and entropy methods, are primarily used [46]. In this study, the dynamic weighting method proposed by Chen et al. [45] was employed. This method calculates the ratio of each risk value within the unit relative to the total risk value and adjusts it by computing the ratio of each risk value within the study area relative to the total risk value within the study area. This could help in developing unique weights for each DPU and highlighting the primary issues of each DPU. The comprehensive mismatch risk was calculated by summing four risks based on dynamic weight values using the following formulas:

$$W_k^i=\raisebox{1ex}{$\frac{Z_k^i}{\sum _{k=1}^m{Z}^i}$}\!\left/ \!\raisebox{-1ex}{$\frac{\sum _{i=1}^n{Z}_k}{\sum _{k=1}^m\sum _{i=1}^{n}Z}$}\right.$$

(4)

$$V^i={\sum }_{k=1}^m{W}_k^i{Z}_k^i$$

(5)

where $W_k^i$ is the weight of the $k$-th risk of the $i$-th DPU, $V^i$ represents the sum of the four risk values for the $i$-th DPU, $Z_k^i$ refers to the standardized value for the $k$-th risk of the $i$-th DPU, $n$ denotes the number of DPUs, and $m$ is the number of risk types. Based on Formula (1), the risks of supply–demand mismatch of the four UESs were calculated, and the results were normalized. Then, based on Formulas (4) and (5), the dynamic weights of each DPU were calculated, and the comprehensive risk was obtained and subsequently categorized into five levels using the natural breaks method. DPUs with high and medium–high risks were identified as high-urgency units.

2.3.3. Synergetic Potential Identification

Local Moran’s I can reveal the spatial clustering characteristics of single and bivariate indicators [59]. Using GeoDa V1.2 software, we performed local univariate spatial autocorrelation analysis to explore four risks and local bivariate spatial autocorrelation analysis to study the combinations of two risks. LISA maps were generated to visualize the identified clusters and outliers, highlighting high–high (HH), low–low (LL), high–low (HL), and low–high (LH) clustering. The sum of HH types from the univariate and bivariate results of each DPU’s autocorrelation was used to estimate the potential synergistic effects of urban green infrastructure (UGI) interventions. The total sum of univariate Moran’s I spatial autocorrelation indicated the ability of an individual DPU to cluster high-value risks with its surrounding units, while the total sum of bivariate Moran’s I spatial autocorrelation revealed the capability of a DPU and its surroundings to aggregate different high-value risks. Such DPUs and their surroundings faced multiple risks of mismatching urban ecosystem services (UESs), which called for more comprehensive planning and optimization of UGI distribution to achieve stronger synergistic effects.

2.3.4. Prioritizing UGI Interventions

When identifying the priority of UGI interventions of DPUs, two essential components were analyzed. On the one hand, the spatial ranking of DPUs is crucial for decision making; on the other hand, the use of adaptive strategies that match the risk typology is essential for UGI planning. To address the matching of UGI strategies with risk types, we developed adaptive UGI strategies that can match risk typology based on previous studies [11,19,49,60,61].

To determine spatial ranking, details should be taken into account. On one hand, the high risk of UES supply–demand mismatch should be addressed urgently. On the other hand, the synergetic potential of DPUs should also be noticed, such as the common need for UGI updating, as well as the positive spillover effects resulting from implemented UGI measures. Additionally, given the scarcity of spatial resources, the feasibility of implementing UGI interventions should be assessed based on the potential for renewal. Our research identified “the proportion of urban renewal land area in DPUs” that were surveyed by the Xi’an Urban Planning and Design Institute, which is a suitable indicator of feasibility.

The TOPSIS method [62] was utilized with SPSS Statistics 25.0 to identify the optimal sequence of DPUs that integrates urgency, synergy, and feasibility, as mentioned above. To be specific, all DPUs need to be represented in a decision matrix ($X$) with dimensions $m\times n$ (m = 145, n = 3), where $X_{ij}$ represents the score of the $i$-th DPU under the $j$-th criteria. The weight of each criterion can be derived from expert opinion [25,27,62], which is a vector ($W$) with length $n$ (See Appendix B for a list of experts consulted for opinions and the result of the questionnaire to determine the weights). Then, the TOPSIS score of each DPU can be calculated through the following steps:

1.

Normalize each column of matrix $X$ so that the sum of each column is 1, and transform each non-negative value ($X_{ij}$) into its relative size within the current column;

2.

Calculate the weighted value of each element in matrix $X$ and transfer them into a new matrix. This weighted matrix ($V$) can be calculated according to the following formula:

$$\begin{array}{l}{V}_{ij}=W_j\cdot X^{\prime }=\left(\begin{array}{ccc}{x}_{11}^{\prime }\cdot w_1& \dots & x_{1n}^{\prime }\cdot w_n\\ ⋮& \ddots & ⋮\\ x_{m1}^{\prime }\cdot w_1& \dots & x_{mn}^{\prime }\cdot w_n\end{array}\right)\\ i=\left(\mathrm{1,2},\dots ,m\right),j=(\mathrm{1,2},\dots ,n)\end{array}$$

(6)

3.

The positive ideal solution ($S_i^{+}$) and negative ideal solution ($S_i^{-}$) can be respectively calculated using the following formulas:

$$S^{+}=\left\{\left(\underset{i}{\max }{V}_{ij} \mid j\in J\right)\mid i=1,2,\dots m\right\}=\left\{V_1^{+},V_2^{+},\dots ,V_n^{+}\right\}$$

(7)

$$S^{-}=\left\{\left(\underset{i}{\min }{V}_{ij} \mid j\in J\right)\mid i=1,2,\dots m\right\}=\left\{V_1^{-},V_2^{-},\dots ,V_n^{-}\right\}$$

(8)

where $J$ represents the criteria; ${V_j}^{+}$ represents the maximum weighted score in the $j$-th criteria, which is the positive ideal solution; and ${V_j}^{-}$ represents the minimum weighted score in the $j$-th criteria, which is the negative ideal solution;

4.

Calculate the Euclidean distance of each DPU to the positive ideal solution ($S_i^{+}$) in Formula (9) and the negative ideal solution ($S_i^{-}$) in Formula (10):

$$S_i^{+}=\sqrt{{\sum }_{j=1}^n{\left(V_{ij}-{V_j}^{+}\right)}^2},i=1,\dots ,m$$

(9)

$$S_i^{-}=\sqrt{\sum _{j=1}^n{\left(V_{ij}-{V_j}^{-}\right)}^2},i=1,\dots ,m$$

(10)

5.

Calculate the relative position of each DPU compared to all other DPUs in Formula (11):

$$C_i=\frac{S_i^{-}}{S_i^{+}+S_i^{-}},0<C_i<1,i=(1,2,\dots ,m)$$

(11)

where $C_i$ represents the TOPSIS score of the $i$-th DPU, and a higher value indicates a more favorable DPU [62,63].

3. Results and Analysis

3.1. Assessment of Stress, Exposure, and Vulnerability

The evaluation results of stress, exposure, vulnerability, and risk for each DPU are displayed in Figure 3. The absence of low-stress areas for air pollution in the study area indicates that overall levels exceeded standards, with high-risk areas concentrated in the northwest of the city. The risk of urban waterlogging exhibited an inside–out trend decreasing trend, and there were three aggregation areas located within the inner-city wall, to the south of the second ring road, and to the east of the third ring road. Low-stress areas were mainly distributed in the Chang’an district in the south and the Chanba ecological district in the northeast. The risk of heat island exhibited features of gathering in the central city, with high-risk areas mainly distributed within and along the second ring road and to the southwest and northeast of the third ring road, while low-risk areas were located in the Qindu district in the southeast. The risk of insufficient recreation was scattered along the second ring road.

Figure 3. The evaluation results of stress, exposure, vulnerability, and comprehensive risk.

3.2. Risk Typology

Based on the K-means clustering analysis of four service supply–demand mismatch risks, the “elbow method” was used to determine the optimal number of clusters as seven (Figure 4a,b). As shown in Figure 4a, the determined comprehensive risk level was employed to evaluate the severity of risk for different risk clusters. Cluster I consisted of two DPUs with extremely high cumulative risk, accounting for 1.4%, primarily located in the Hongmiaopo subdistrict of Lianhu district and the Fangzhicheng subdistrict of Baqiao district. Cluster II included 14 DPUs that were at extremely high risk of waterlogging, as well as heat islands, making up 9.7% of the total, mainly distributed within the first ring wall and its western and eastern sides. Cluster III comprised four DPUs with air-pollution-oriented high risk, representing 2.8% of the total, and situated between the second and third rings on the west side of the city. Cluster IV consisted of eight DPUs with insufficient recreation-oriented high risks, accounting for 5.5%, scattered along the north, southwest, and southeast sides of the second ring. Cluster V contained 20 DPUs with heat-island-oriented high risk, accounting for 13.8% of the total, and distributed along the east and south sides of the second ring. Cluster VI included 35 DPUs with medium risk, making up 24.1% of the total, located between the second and third rings and on the south side of the third ring. Cluster VII consisted of 62 DPUs that were at low risk, making up 42.8% of the total, mainly distributed along the third ring of the city and in the outer areas. The data presented in Figure 4b show that high-risk DPUs mostly had a low proportion of UGI area, indicating the problem of insufficient supply of UESs.

Figure 4. (a) The risk typology of DPUS; (b) Relationship between the proportion of UGI area and comprehensive risk level of different types of DPUs.

3.3. Urgency of Risk

We employed the dynamic weighting method to calculate the weights of each DPU (Figure 5a) and applied them to comprehensively evaluate the risks (Figure 5b). A total of 75 DPUs had medium risk, medium–high-risk, or high risk, accounting for 51.7% of the total, and were mainly distributed within the second ring road, extending towards the third ring road, indicating the outward spillover feature of risks. Among them, 20 DPUs had high risk, making up 13.8% of the total, primarily located within the city wall and to the north, west, and northeast of the second ring road, as well as to the east of the third ring road. Additionally, there were 24 medium–high-risk DPUs, representing 16.6% of the total, mainly distributed along the second ring road and the east of the third ring road. Low–medium-risk and low-risk areas were mostly situated at the edge of the outer three ring roads.

Figure 5. (a) The dynamic weight for each DPU. (b) The comprehensive risk of UES supply–demand mismatch.

3.4. Synergetic Potential

The results of the local Moran’s I for both univariate and bivariate calculations are presented in Figure 6 and Table 4. All spatial autocorrelation results for univariate and bivariate variables show a significant positive correlation (p < 0.001), indicating high-value aggregation of single or multiple risks between DPSs and their neighboring units. As illustrated in Figure 6, the high-value aggregation areas (HH) of three regulating services’ mismatch risks were observed in the central city, while those related to the risk of insufficient recreation were scattered along the second ring. To further investigate the degree of correlation between risks, the HH types of univariate and bivariate spatial autocorrelation analysis results were superimposed independently (Figure 7a,b). The high-value DPUs of both were located on the north side within the first ring wall and on the west and northwest sides of the second ring road. Although they overlapped mostly, there were still differences between them. By equally superimposing two maps and dividing the result into five levels, a comprehensive synergy potential was obtained, as displayed in Figure 7c. A total of 19 units with high and medium–high synergy potential were identified, indicating that approximately 13.1% of the total area should be specially considered.

Figure 6. Clustering maps of four risks.

Table 4. The Moran’s I and p values for univariate and bivariate spatial autocorrelation.

Variable	Heat Island		Urban Waterlogging		Air Pollution		Insufficient Recreation
	Moran’s I	p Value	Moran’s I	p Value	Moran’s I	p Value	Moran’s I	p Value
Heat Island	0.724 **	0.001	0.631 **	0.001	0.427 **	0.001	0.401 **	0.001
Urban Waterlogging	0.626 **	0.001	0.699 **	0.001	0.421 **	0.001	0.367 **	0.001
Air Pollution	0.429 **	0.001	0.434 **	0.001	0.679 **	0.001	0.354 **	0.001
Insufficient Recreation	0.432 **	0.001	0.396 **	0.001	0.375 **	0.001	0.414 **	0.001

Note: The symbols ** represents highly significant spatial autocorrelation at the level of 0.05.

Figure 7. The results of spatial autocorrelation in high–high regions ((a) univariate, (b) bivariate, (c) comprehensive).

3.5. Priority of UGI Interventions

3.5.1. Spatial Ranking of DPUs for UGI Interventions

After considering the TOPSIS method, we conducted a ranking of every DPU (See Appendix C for detailed result of the TOPSIS method for all DPUs) and selected the top 20% DPUs (a total of 30) as priority DPUs, which have more needs and potential for UGI interventions (Figure 8). The priority DPUs are primarily located within the first ring road and scattered along the second ring road, with a higher concentration on the west and east sides. Additionally, three DPUs are distributed on the north and east sides of the Third Ring Road. For details of the ranking, please refer to Figure 8 and Table 5. Three different kinds of DPUs were selected. As shown in Figure 8, the first kind of DPUs occupied high urgency, feasibility, and synergy simultaneously and were clustered within the first ring road, as well as the western section of the second ring road, such as LH-07, LH-02, and DMG-07. The second kind of DPUs only highlighted two factors. The DPUs situated in the east section of the second ring road, such as XC-07, YT-08, and XC-08, have a high level of urgency and feasibility. Meanwhile, the DPUs within the first ring road, such as LH-12 and LH-13, exhibited high levels of urgency and synergy.

Figure 8. Spatial ranking of DPUs for UGI interventions.

Table 5. The priority ranking of TOP30 DPUs based on TOPSIS method.

Code	R	U	S	F	TS	Code	R	U	S	F	TS
LH-07	1	1.00	0.81	0.50	0.75	YT-08	16	0.78	0.00	0.67	0.55
LH-02	2	0.93	0.48	0.67	0.71	XC-08	17	0.59	0.00	1.00	0.55
DMG-01	3	0.94	0.58	0.50	0.69	XC-07	18	0.54	0.02	1.00	0.54
LH-04	4	0.70	0.37	1.00	0.67	WY-05	19	0.52	0.00	1.00	0.53
LH-09	5	0.74	0.49	0.67	0.66	BL-03	20	0.85	0.38	0.17	0.53
LH-08	6	0.82	0.40	0.67	0.66	BQ-03	21	1.00	0.14	0.17	0.53
LH-14	7	0.85	0.87	0.33	0.66	XC-04	22	0.85	0.13	0.33	0.52
LH-03	8	0.90	0.30	0.67	0.65	LH-13	23	0.85	0.51	0.00	0.51
LH-06	9	0.82	0.45	0.50	0.63	BQ-04	24	0.84	0.00	0.33	0.50
LH-05	10	0.82	0.30	0.67	0.63	BQ-01	25	0.84	0.00	0.33	0.50
XC-01	11	0.84	0.72	0.33	0.63	XC-09	26	0.62	0.04	0.67	0.50
LH-01	12	0.82	0.14	0.67	0.59	BL-05	27	0.85	0.34	0.00	0.49
LH-12	13	0.91	1.00	0.00	0.59	YT-05	28	0.87	0.10	0.17	0.49
XC-06	14	0.67	0.00	1.00	0.57	WY-13	29	0.83	0.32	0.00	0.48
XC-02	15	0.84	0.55	0.17	0.56	DMG-04	30	0.89	0.00	0.17	0.48

Note: The abbreviations in the table headers include R (ranking), U (urgency), S (synergy), F (feasibility), and TS (TOPSIS score).

These 30 DPUs faced multiple risks of UES supply–demand mismatch and high spatial aggregation of risk, indicating higher transforming effectiveness and high feasibility of implementation. Policy makers should make decisions according to the clear ranking result of the TOPSIS score, as well as additional factors to be considered.

3.5.2. The Categorization Strategy of UGI Interventions Based on Risk Typology

In response to different risk typologies, we developed a categorization UGI intervention strategy tailored to their implementation feasibility and matching priority UGI types. As shown in Figure 9, the strategies can be divided into three categories: (1) embed and integrate, (2) create and restore, and (3) preserve and protect. We can match the strategies according to the feasibility of the seven risk types. UGI types should also be targeted and matched to specific risk types to improve responsiveness to certain risks.

Figure 9. Categorization Strategy of UGI interventions based on risk typology.

4. Discussion

The key objective of sustainable urban renewal in built-up areas is to achieve synergistic optimization of spatial performance and resident well-being by addressing urban environmental issues and enhancing the functionality of UESs. In addition, to effectively allocate UGI as a public asset to areas in need during urban renewal, determining priority locations and timing is essential [23,32,45]. Therefore, we proposed a framework that identifies the priority of UGI interventions in DPUs based on the assessment of risks of UES supply–demand mismatch. This framework evaluates the urgency, typology, synergy, and feasibility of UGI interventions in a step-by-step manner, which helps to optimize the spatial layout of UGI at the city level, target UGI intervention types, and integrate DPUs to achieve synergy effects.

4.1. Spatial Priority Identification Integrating Urgency, Synergy, and Feasibility

UGI interventions in built-up areas, with the objective of sustainable renewal, have garnered significant research attention. In this study we proposed a stepwise framework integrating urgency, synergy, and feasibility to identify the priority of DPUs for UGI interventions, which is crucial for UGI planning.

The first step is to assess the risk of UES supply–demand mismatches, which requires the scientific integration of multiple risks [46]. We adopted the dynamic weighting method [45,47,64], which can dynamically adjust weights based on the four UES risk characteristics of each DPU. This method addresses the problem of weakening single high-risk value areas during the overlay process, which might be caused by the equal weight method, entropy methods, or other methods that use globally uniform weights [45,46,47]. Currently, areas in need of urgent intervention due to high risks of UES supply–demand mismatch are mainly concentrated in the old city center, which is consistent with the research conclusions reported in [32]. This reflects the trap of compact urban development, namely the compact city paradox [4]. These DPUs not only represent comprehensive high-risk areas but also areas with the highest level of social vulnerability (Figure 3), particularly those with high proportions of elderly residents. UGI intervention strategies should be formulated based on the characteristics and needs of vulnerable groups in order to improve spatial justice [65].

The second step involves evaluating the synergy among DPUs to enhance the multifunctional optimization effect [19,25,47,66] of UGI and promote collaboration between them. Numerous studies have confirmed the multifunctionality of UGI, which provides synergistic effects among multiple UESs [9,23,66]. However, from a planning and coordination perspective, it is important to assess the local relationship of synergy among DPUs [45]. DPUs facing similar high risks are more likely to collaborate and implement city-level UGI interventions, which can significantly enhance UGI connectivity. There were 19 DPUs (13%) with high synergy potentials, reflecting the reality of the high risk of UES supply–demand mismatch in both DPUs and the surrounding areas. On the one hand, this indicates that residents in these DPUs had persistently faced multiple environmental risk pressures, which may worsen or spread non-linearly if not addressed in time [45]. On the other hand, there was significant demand and potential for synergy optimization among these DPUs and their surrounding areas. The advantages of integrating UGI interventions among DPUs should be fully utilized to enhance spatial overflow effects after the implementation of UGI measures.

Thirdly, amidst a backdrop of stock renewal, feasibility is crucial for UGI implementation. The built-up areas were facing escalating developmental pressure [47], so it was challenging to implement an incremental logic-based optimization strategy for the construction an ideal UGI blueprint. Thus, proportions of vacant land, low-efficiency land, and land awaiting renewal within DPUs could serve as key indicators of the feasibility of implementation [25,38].

Finally, we utilized the TOPSIS method, which can comprehensively rank alternatives by considering multiple criteria simultaneously, making it well-suited to integrate the urgency, synergy, and feasibility factors we mentioned earlier. Furthermore, the TOPSIS method allows for the inclusion of subjective preferences and weights [62,63], which means that we can gather opinions from experts [25] and different stakeholders [26]. This gives decision makers more control over the decision-making process and could potentially enhance the effectiveness and efficiency of planning practice [62].

4.2. Categorization of UGI Interventions Based on Risk Typology

The current 145 DPUs can be classified into seven categories based on clustering results, with each category corresponding to different levels of risk and dominant issues. Figure 4b shows that the risk level of DPUs was closely related to their internal proportion of the UGI area. High-risk DPUs usually had a low proportion of UGI areas and a high population demand, while peripheral DPUs had a higher proportion of UGI areas with less demand. This indicates a mismatch between green space supply and residents’ demand in the city [32].

As shown in Figure 9, DPUs with high and extremely high risk levels should maximize the use of available space, improve both the quantity and quality of UGI, and choose effective UGI intervention strategies and types based on risk typology. Targeted and precise UGI intervention strategies should be formulated accordingly. For DPUs with a high proportion of renewal land, the “create and restore” approach [19] can be employed, and innovative approaches should be carried out using policy opportunities such as urban renewal, ecological restoration, sponge city [13], brownfield restoration [67], transformation of gray infrastructure into green infrastructure [68], rooftop and vertical greening [2,4], etc. While for DPUs with limited available land, a gradual “embed and integrate” approach can be implemented. In addition to top–down updates, bottom–up spontaneous public action should not be overlooked. Cities such as Berlin and London [69] encouraged public participation through their “tree planting plan” policy, which helped residents rebuild their connection with nature by providing ecological engagement opportunities while reshaping urban residents’ perceptions, memories, and attitudes towards nature. This promoted ecological awareness and encouraged green actions [9].

DPUs with a medium risk level may face increasing risk of supply–demand mismatch. Therefore, to avoid further exacerbation, a “create and restore” approach can be chosen for the early creation and protection of high-quality green space, as well as the optimization and upgrading of existing UGI quality. For DPUs currently at the low risk level, the relative balance of supply and demand is due to undeveloped land and abundant green space supply, in addition to low population density and less demand pressure. However, with further urban development, they still face the risk of supply–demand mismatch in the future. Therefore, the “preserve and protect” strategy should be adopted to identify key UGI in advance, and green space should be maximally protected to prevent green spaces from being encroached upon during the process of urban development. Methods should be taken into account to minimize the degradation of ecosystem service functions before and after planning, such as an assessment tool for setting ecosystem performance baselines for urban transformation [23].

4.3. Limitations and Furutre Directions

The present study is subject to several limitations. First, the assessment of risks of UES supply–demand mismatch only explored four major environmental issues in the built-up areas of medium and large Chinese cities. However, the research framework is highly flexible and can be expanded to include evaluations of other services, such as carbon storage and biodiversity, based on changes in research objects, scope, and the availability of data. Secondly, while factor weights for spatial priority recognition in this study were established based on expert opinions, future research could benefit from a comparison of scenarios [17] that involve different stakeholder perspectives to enhance the comprehensiveness and defensibility of spatial priority recognition outcomes. Thirdly, the study mainly focused on diagnosing the priority of UGI inventions, which is a crucial step in identifying typical units and developing targeted approaches. Due to length limitations, we did not thoroughly explore detailed strategies for responding to different levels and types of risks. Finally, our evaluation relied on city-scale data from different sources, the precision and quality of which are subject to certain limitations. Therefore, enhancing the accuracy and reliability of available data is crucial for future studies.

5. Conclusions

In the context of sustainable urban renewal, UGI interventions have attracted increasing attention due to their ability to provide targeted, multifunctional improvements that are essential for achieving high-quality development and enhancing residents’ well-being. We propose a six-step framework integrating urgency, synergy, feasibility, and typology to identify priority UPDs that require UGI interventions. In Xi’an’s built-up area, we evaluated 145 DPUs and assessed the risks of UES supply–demand mismatch. DPUs were classified into different risk typologies using clustering analysis, and the synergy potential between DPUs was determined through univariate and bivariable local spatial autocorrelation analysis. Finally, we determined the spatial ranking of DPUs and matched adaptive strategies to prioritize UGI interventions.

Our study is characterized by two main strengths. First, we proposed a stepwise framework to address the limitations associated with the current UGI spatial prioritization evaluation framework. The proposed framework utilizes the dynamic weighting method in the first step, identifying the urgency of risks that highlight serious risks. Then, the TOPSIS method is used to integrate “urgency, synergy, and feasibility” and identify the spatial ranking of DPUs, enriching the dimensions prioritizing UGI interventions in built-up areas and increasing clarity in decision making. Secondly, we focused on DPUs to better characterize the spatial heterogeneity and enhance effective linkages with the current spatial planning system. This framework is designed to be flexible in utilizing obtainable data and national standards, making it applicable to most cities. Our study can provide scientific evidence for the comprehensive implementation of spatial planning, urban renewal, and high-quality sustainable development.