A Decadal Assessment of the Coordinated Relationship Between Heat Risk and Cooling Resources in Guangzhou, China

Abstract

Global climate change has intensified urban heat exposure risks due to extreme heat events, posing significant health threats, particularly to socially vulnerable groups such as the elderly and children. However, the spatial allocation of urban public cooling resources exhibits heterogeneity, leading to insufficient or mismatched provision of cooling facilities in high heat exposure areas. Taking the central urban area of Guangzhou, China as an example, we employ the hazard–exposure–vulnerability (HEV) framework to evaluate a composite heat risk index (HRI). Using a coupling coordination degree and development coordination coefficient, we identify the matching status and temporal dynamic between heat risk and facility supply across 2010 and 2020. The results indicate that (1) HRI generally exhibits high-value clustering in the core areas of the old city, while peripheral areas show relatively lower levels; (2) the coupling coordination degree (CCD) exhibits clear spatial clustering characteristics, and highly coordinated streets are mostly concentrated in old city areas, whereas newly developed and peripheral districts generally show low coordination; and (3) from 2010 to 2020, cooling facility development in old city districts was generally proactive, while newly developed and peripheral areas exhibited slower progress relative to increasing heat risk. This study highlights the issue of adaptive imbalance in the allocation of cooling resources concerning vulnerable populations, providing guidance for future urban planning.

1. Introduction

In recent years, global warming has led to frequent extreme heat events, making public health problems caused by heat exposure increasingly prominent [1,2]. The World Health Organization (WHO) suggests that heat stress under exposure has become a leading cause of weather-related mortality [3]. Extreme heat not only exacerbates chronic diseases such as cardiovascular [4], respiratory [5], and kidney ailments [6] but also significantly increases the incidence of acute health events like heatstroke [7]. A United Nations climate action report indicates that over 70% of the global workforce (approximately 2.4 billion people) faces threats from extreme heat [8]. Between 2000 and 2019, an average of approximately 249,000 additional deaths annually were attributed to climate-related health issues like high temperatures and heatstroke, with 45% of these deaths occurring in Asia [3]. China also faces severe challenges. Research by Cai et al. [9] shows that nearly 15,000 deaths in 2020 were attributed to heat exposure. The IPCC Sixth Assessment Report further emphasizes that if global warming exceeds 1.5 °C, extreme heat and heatwave events will become more frequent and intense worldwide [10]. Therefore, establishing a scientific heat risk assessment system is crucial not only for identifying the spatial impact patterns of heat events but also for informing policies to address extreme weather.

Nowadays, numerous studies have attempted to integrate remote sensing data, demographic information, and socioeconomic indicators to construct multi-factor comprehensive assessment frameworks for quantifying heat exposure risk [11]. For instance, Estoque et al. [12] assessed heat health risk in 139 Philippine cities by combining remote sensing and socioeconomic data. Yu et al. [13] developed a hazard–exposure–vulnerability (HEV) framework for British Columbia, Canada, by integrating multidimensional health and social indicators. Building on the HEV framework, Yao et al. [14] proposed a social–ecological–economic index (SEEI) to evaluate the spatial distribution of heat health risk (HHR) in Shanghai during extreme heat periods. Although these studies continuously enrich the heat risk assessment system through framework construction, mechanism exploration, and spatial identification, they predominantly focus on general risk identification and driver analysis [14,15,16]. However, they overlook the critical aspect of vulnerable groups and cooling resource supply–demand balance in the community context.

Vulnerable groups (e.g., the elderly) are particularly sensitive to heat-related health risks due to inherent disadvantages in thermoregulation, baseline health, information access, and social support [17,18,19]. Existing physiological and epidemiological studies confirm that the elderly, with declining thermoregulatory function, are more susceptible to dehydration and cardiovascular dysfunction in high-temperature environments [20]. Further, children are more prone to heatstroke due to their immature thermoregulatory systems and insufficient heat dissipation capacity [21]. Also, individuals with lower education levels face greater exposure risk than the general population due to their limited ability to learn risk defense skills and access information [22,23]. Furthermore, the UN 2030 Agenda for Sustainable Development explicitly calls for reducing the unequal exposure of vulnerable groups to climate risks (SDG 10) and enhancing the resilience and adaptive capacity of urban public services (SDG 11) in multiple Sustainable Development Goals (SDGs) [24,25,26].

In addition, the unbalanced allocation of cooling facilities further exacerbates the heat risks for vulnerable groups, as these facilities directly impact the community’s sheltering capacity and health protection during high-temperature periods [27,28,29]. Particularly, public spaces such as libraries, cultural centers, community service centers, and children’s palaces play vital roles as cooling shelters [28,30,31]. Extensive research has systematically explored the role of green infrastructure (e.g., urban parks) in mitigating the urban heat island effect and providing cooling capacity [32], especially with regard to the supply and demand of cooling in urban blue-green spaces, and the relationship between heat risk and the cooling benefits of green spaces [33,34]. However, fewer studies integrate multiple categories of public cooling facility resources to reflect the comprehensive level of cooling space service capacity available to residents. In reality, accessible and usable public spaces are especially critical for vulnerable groups during heatwaves.

Regarding quantitative methods for measuring supply–demand matching, the coupling coordination degree model has been widely applied in fields like ecosystem services, medical resource allocation, and land use coordination [35,36,37]. By quantitatively characterizing the coordination level between supply and demand systems, it provides an effective analytical tool for assessing resource allocation equity [34,38]. However, in studies on heat risk and cooling facility allocation, the practice of systematically measuring supply–demand matching status using the coupling coordination degree model is still rare. There is an even greater lack of empirical exploration integrating composite heat exposure risk indicators with multi-type cooling facility supply into a unified analytical framework.

Furthermore, the spatial allocation of urban cooling resources continuously evolves with urban expansion, planning adjustments, and policy guidance, resulting in dynamic changes in the supply–demand relationship between cooling facilities and heat exposure risk [28,39]. In the long term, disparities in resource allocation may persist and even widen in some high-risk areas due to historically low investment in resources, resulting in persistently inadequate levels of cooling facilities [30,39]. Current studies, however, analyze supply–demand relationships only at a single time point, with few comparing changes across different years [16]. Dynamic analysis of supply–demand coordination across different periods can help to identify long-term trends in resource allocation equity, highlighting areas that have improved and those with persistent mismatches. This provides a basis for subsequent policy adjustments and resource allocation.

This study focuses on identifying and dynamically assessing the coordinated relationship between heat exposure and cooling resources, with a specific focus on vulnerable groups. It integrates multiple categories of public cooling facilities into a single indicator representing overall cooling service capacity at the subdistrict level and links this with a heat risk index through both static coordination analysis using the coupling coordination degree model and dynamic change assessment using the development coordination coefficient model. This combined approach is useful because it allows for a systematic evaluation of the supply–demand relationship at a fine spatial scale while also revealing its temporal evolution over a decade, thereby capturing both persistent mismatches and areas of improvement. By applying this dual-perspective framework, the study extends previous research that has typically examined static assessments, and it provides a transferable reference for exploring the equity of heat risk mitigation resources in other urban contexts.

Guangzhou shares key characteristics with many heat-exposed cities around the world, such as a hot and humid subtropical climate, densely built environments with limited ventilation, and pronounced inequalities in access to cooling resources. These features make it a representative case for studying urban heat exposure and adaptation strategies. In addition, the city’s efforts to address heat risk inequality align with global policy agendas such as the United Nations Sustainable Development Goals (SDG 10 and SDG 11), emphasizing inclusive urban development and climate adaptation for vulnerable populations [40]. In light of its global relevance and representativeness, the central urban area of Guangzhou was selected as the case study site. The research objectives include (1) constructing a heat risk indicator system oriented towards vulnerable populations to identify high heat risk areas; (2) integrating multi-category cooling facility resources to construct cooling service supply indicators and quantitatively evaluating supply–demand coordination using the coupling coordination degree model; and (3) systematically analyzing the evolution of the coordination state between heat risk and cooling resource supply and demand under urban development by comparing data from 2010 and 2020. This study aims to reveal structural mismatch patterns between the spatial distribution of urban heat risk and cooling resources from the perspective of vulnerable groups, providing a theoretical foundation and policy reference for optimizing the allocation of cooling facilities and supporting the development of urban planning strategies.

2. Study Area and Materials

2.1. Study Area

Guangzhou is located in southern China, within the south subtropical monsoon climate zone. Its climate is characterized by high temperatures, high humidity, and long summers. In recent years, driven by both global climate change and rapid urban expansion, Guangzhou has experienced a continuous rise in heat risk, with heat exposure becoming increasingly severe, breaking high-temperature records for multiple consecutive years [41,42]. Since 2020, this city has been one of the first pilot cities in the World Bank’s “Cooling Cities Program” [40]. The city aims to enhance public cooling capacity through urban renewal, green space expansion, and public space transformation, seeking to strengthen spatial planning for the heat exposure adaptation of vulnerable groups [40]. This study selects six central urban districts of Guangzhou (i.e., Liwan, Yuexiu, Haizhu, Tianhe, Baiyun, and Huangpu) as the research area, encompassing a total of 120 communities (Figure 1).

2.2. Data Sources

This study focused on decadal-scale changes in urban heat risk and cooling resource coordination, aiming to capture long-term spatial trends rather than short-term fluctuations. It utilized multi-source data to assess urban heat risk in Guangzhou during the summer months of June through August, consistent with previous studies and reflecting the period of highest seasonal temperatures in the region [16]. Land surface temperature (LST) and normalized difference vegetation index (NDVI) data were derived from Landsat 5 TM (2010) and Landsat 8 OLI/TIRS (2020) imagery (USGS) for June through August of each year. All available images within this period were processed using the maximum value composite (MVC) method to produce summer composites. Summer LST and NDVI values were then calculated from these composites using their respective standard formulas. Population structure data were obtained from the Sixth (2010) and Seventh (2020) National Population Census of Guangzhou, while GDP data were sourced from the China GDP Spatial Distribution Kilometer Grid Dataset (RESDC) at 1 km resolution. Additionally, cooling facility point of interest (POI) data were collected through Amap platform’s API, categorizing facilities into science, education, and cultural venues; sports and recreation; and scenic spots and parks, with 1055 and 12,373 valid POIs identified for 2010 and 2020, respectively.

We integrated these diverse datasets to establish a comprehensive assessment system at the community scale. Spatial indicators, including LST, NDVI, and GDP, were combined with population structure data from the censuses, while the per capita availability of cooling facilities was calculated based on POI data [43]. This multi-dimensional indicator system enabled the evaluation of heat hazard, heat exposure, and heat vulnerability across the study area, as detailed in

Table 1. Summary of indicators for constructing the heat risk index (HRI) framework.

Factors	Id	Indicators	Expected Impact	Refs.
Hazard	-	Land Surface Temperature (LST)	Positive	[44]
Exposure	E-01	Population Density	Positive	[45]
	E-02	Normalized Difference Vegetation Index (NDVI)	Negative	[43]
Vulnerability	V-01	Population Density over 60 years	Positive	[13]
	V-02	Population Density under 15 years	Positive	[13]
	V-03	Low Education Level Density	Positive	[23]
	V-04	Total community GDP	Negative	[46]

, providing a thorough basis for urban heat risk assessment. Spatial visualization maps were classified using the Jenks natural breaks method.

3. Methodology

3.1. Assessment of Heat Risk

The heat risk assessment framework is based on the hazard–exposure–vulnerability (HEV) paradigm [14]. Heat hazard refers to the probability and intensity of a physical event with the potential to cause harm [47]. In this study, heat hazard was represented by LST, a key physical indicator of urban thermal accumulation derived from remote sensing. Higher LST values correspond to greater hazard levels, making LST a positive indicator. The standardized LST value served directly as the heat hazard index (HHI). Additionally, heat exposure quantifies the presence of populations, assets, or ecosystems in areas subject to heat threats [47]. It was assessed using two indicators: population density (positive indicator; higher density increases exposure risk) and the NDVI (negative indicator; higher values indicate better cooling potential). Standardized values for these indicators were weighted using principal component analysis-derived weights to compute the heat exposure index (HEI):

$$\mathrm{H}\mathrm{E}\mathrm{I}={\sum }_{i=1}^n(w_{ei}\times h_{ei})$$

(1)

where $w_{ei}$ is the weight, and $h_{ei}$ is the standardized value for exposure indicator $i$.

Heat vulnerability captures the susceptibility and lack of adaptive capacity of populations or systems experiencing adverse effects from heat exposure [47]. Four indicators were used: elderly population density (positive; increased sensitivity), child population density (positive; lower physiological tolerance), illiterate population density (positive; limited adaptive capacity), and total GDP at the community level (negative; higher economic capacity enhances resilience). Standardized values were weighted using principal component analysis (PCA) derived weights to calculate the heat vulnerability index (HVI):

$$\mathrm{H}\mathrm{V}\mathrm{I}={\sum }_{i=1}^n(w_{vi}\times h_{vi})$$

(2)

where $w_{ei}$ is the weight, and $h_{ei}$ is the standardized value for vulnerability indicator $i$.

In addition, heat risk was assessed by integrating the three dimensions (H, E, and V) within the HEV framework, where hazard reflects the physical thermal potential, exposure describes the spatial intensity of population presence, and vulnerability characterizes the susceptibility of exposed populations. The composite heat risk index (HRI) was calculated as

$$\mathrm{H}\mathrm{R}\mathrm{I}=w_H\times \mathrm{H}\mathrm{H}\mathrm{I}+w_E\times \mathrm{H}\mathrm{E}\mathrm{I}+w_V\times \mathrm{H}\mathrm{V}\mathrm{I}$$

(3)

where $HHI$, $HEI$, and $HVI$ are the standardized indices, and $w_H$, $w_E$, and $w_V$ are their respective PCA-derived weights. A higher HRI indicates a greater level of integrated heat risk.

We employed PCA for dimensionality reduction and standardization to unify different evaluation indicators [13]. To eliminate dimensional differences, all indicator data underwent standardization, with indicators classified as positive or negative based on their influence on the heat risk system. The min–max normalization method transformed all indicators into a 0–1 range [48]. Meanwhile, to objectively determine relative weights, PCA integrated potentially correlated variables into uncorrelated principal components while retaining the majority of original information variance. Principal components with eigenvalues greater than 1 were extracted, and variance contribution rates were calculated to ensure a cumulative contribution rate exceeding 70%. The comprehensive weight was then computed using the following formula:

$$w_j={\displaystyle \frac{{\sum }_{k=1}^m{\lambda }_k\cdot |a_{jk}|}{{\sum }_{j=1}^n{\sum }_{k=1}^m{\lambda }_k\cdot |a_{jk}|}}$$

(4)

where ${\lambda }_k$ is the variance contribution rate of the $k$-th principal component, $|a_{jk}|$ is the absolute loading coefficient of indicator $j$ on component $k$, $m$ is the number of extracted components, and $n$ is the total number of indicators.

3.2. Coupling Coordination Degree

To evaluate the coordination between urban heat risk and cooling resource supply, the coupling coordination degree (CCD) model was applied [34]. This model quantifies the interaction and coordinated development level between two subsystems. The HRI and cooling facilities per capita were selected as the representative variables for the heat risk subsystem $U_H$ and cooling resource supply subsystem $U_C$, respectively. Both variables were standardized prior to analysis. The CCD calculation involved three steps. First, the comprehensive development index $T$ was computed as

$$T=\alpha \cdot U_H+\beta \cdot U_C$$

(5)

where weights $\alpha$ and $\beta$ represent the relative importance of each subsystem. This study assigned equal weight (0.5) to both subsystems. Second, the coupling degree $C$, reflecting the intensity of interaction between the subsystems, was calculated as

$$C=2\times {\displaystyle \frac{\sqrt{U_H\times U_C}}{U_H+U_C}}$$

(6)

Finally, the CCD result, integrating the level of development $T$ the interaction intensity $C$, was derived as

$$D=\sqrt{C\times T}$$

(7)

Spatial autocorrelation analysis, comprising global Moran’s I and local Moran’s I, was employed to investigate the spatial clustering patterns and heterogeneity of the CCD across community units within Guangzhou’s central urban area [48]. Global Moran’s I measures the overall spatial dependence across the entire study region. Values range between −1 and 1; a value significantly greater than 0 indicates positive spatial autocorrelation (clustering of similar high or low values), a value significantly less than 0 indicates negative spatial autocorrelation (spatial dispersion or checkerboard pattern), and a value near 0 suggests a random spatial pattern. Following the detection of global spatial dependence, local spatial association analysis was performed to identify local spatial clusters and spatial outliers at the community level. The local Moran’s I statistic for a spatial unit $i$ is given by

$$I_i=z_i\sum _{j=1}^n{w}_{ij}{z}_j$$

(8)

where $z_i$ and $z_j$ are the standardized values of the CCD for units $i$ and $j$; $w_{ij}$ is the spatial weight matrix element defining the neighborhood relationship between units $i$ and $j$. $n$ is the total number of spatial units.

3.3. Development Coordination Coefficient

To evaluate how cooling resource adaptation evolves with changing heat risk, we designed a development coordination coefficient (DCC). This coefficient measures the relative growth rates between cooling facility provision and heat risk intensity from 2010 to 2020, defined as

$$\mathrm{D}\mathrm{C}\mathrm{C}={\displaystyle \frac{\mathrm{l}\mathrm{o}\mathrm{g}({\mathrm{C}\mathrm{R}\mathrm{D}}_{2020}-{\mathrm{C}\mathrm{R}\mathrm{D}}_{2010})}{\mathrm{l}\mathrm{o}\mathrm{g}({\mathrm{H}\mathrm{R}\mathrm{I}}_{2020}-{\mathrm{H}\mathrm{R}\mathrm{I}}_{2010})}}$$

(9)

where $\mathrm{C}\mathrm{R}\mathrm{D}$ represent cooling resource density, and $\mathrm{H}\mathrm{R}\mathrm{I}$ represent heat risk index. The DCC values reveal distinct temporal patterns. A higher DCC indicates a greater balance of cooling resource development. Additionally, a “Coordination–Development” quadrant analysis was implemented to map the current coordination status and development trends. The four quadrants represent: Quadrant I (HH: high CCD–high DCC), denoting areas with favorable current coordination and sustainable development alignment; Quadrant II (HL: high CCD–low DCC), serving as an early warning zone where good coordination hides a critical imbalance, with risk escalating much faster than facility growth, indicating potential future deterioration; Quadrant III (LL: low CCD–low DCC), highlighting lagging zones characterized by the most severe challenges, exhibiting both poor current coordination and insufficient facility development speed relative to intensifying risk; and Quadrant IV (LH: low CCD–high DCC), identifying potential zones where, despite current coordination deficiencies, the positive trajectory of facility development relative to risk offers promise for future improvement. This quadrant analysis framework provides a comprehensive tool for identifying priority areas for intervention and informing targeted policy recommendations to enhance urban cooling resource adaptation.

The overall methodological framework is summarized in Figure 2.

4. Results

4.1. Heat Hazard

Figure 3 illustrates the spatial distribution characteristics of the HHI. High HHI values are concentrated in the western, southern, southwestern, and southeastern communities, primarily encompassing the central-western Baiyun District, western Haizhu District, the entire Liwan District, western Yuexiu District, the southern section of Tianhe District, and southern Huangpu District (e.g., Helong, Fengyang, Longjin, Chepo, and Nangang communities). These high-value zones exhibit pronounced surface temperature accumulation during extreme heat events in summer. In contrast, lower HHI values are prevalent in the northern periphery and outer western areas of the central city, including the northeastern part of Baiyun District and the northern part of Huangpu District. Communities such as Tonghe (which encompasses Baiyun Mountain) and newly developed areas in the outer suburbs of Huangpu benefit from ample green space and lower population density, resulting in significantly reduced heat hazard levels.

4.2. Heat Exposure

Figure 4 shows the spatial distribution of two exposure indicators (NDVI and population density) alongside the HEI. The HEI shows significant spatial disparities, with higher levels in the western and southwestern communities. Specifically, northeastern Yuexiu, western Haizhu, and southwestern Yuexiu show high exposure levels, particularly in Liwan’s Longjin, Fengyuan, and Hualin communities and in Yuexiu’s Meihuacun, Zhuguang, and Guangta communities. These areas are characterized by high population density and low vegetation coverage, leading to substantially increased heat exposure risk. Conversely, HEI values are generally low in the northeastern part of Baiyun District and the northern part of Huangpu District, where population density is lower and vegetation coverage better. The inverse spatial distribution of population density and vegetation coverage drives the high concentration of heat exposure within the central urban core. Heat exposure is significantly higher in the old urban areas than in peripheral communities, forming a prominent, spatially decreasing gradient that radiates outwards.

4.3. Heat Vulnerability

Figure 5 illustrates the spatial distribution of four vulnerability indicators (elderly population density, child population density, illiteracy rate, and GDP) and the HVI. High HVI values are concentrated in the historic core, particularly in the northern part of Liwan District, the central part of Yuexiu District, and the northwestern part of Haizhu District (e.g., Longjin Subdistrict, Guangta Subdistrict, and Jiangnan Zhong Subdistrict). These areas typically have high concentrations of elderly people and high numbers of children and illiterate individuals, resulting in higher overall vulnerability. Peripheral areas generally exhibit lower HVI levels, though the underlying causes vary regionally. In northern and northwestern Baiyun and Huangpu, low HVI levels are primarily due to small population bases and low densities of vulnerable groups, which mitigates the impact of lower economic development. In contrast, northern/eastern Tianhe (e.g., the Wushan, Longdong, and Yuangang communities) benefits from higher levels of economic development and a younger population, leading to a lower HVI. Overall, the high vulnerability in the historic core is primarily driven by demographic factors (an aging population, high child ratios, and low education levels), while lower vulnerability in peripheral areas is influenced by smaller populations, stronger economies, and younger demographics.

4.4. Heat Risk Analysis

Based on HEV results, Figure 6 shows the spatial distribution of the HRI. Areas with high HRI values are concentrated in the historic urban core, particularly at the junction of the Yuexiu, Liwan, and Haizhu Districts as well as in the central Haizhu communities (e.g., Changgang, Shayuan, and Longjin). These zones have high surface temperatures, dense populations, and a large proportion of vulnerable groups, as reflected by high HHI, HEI, and HVI values, which lead to significantly elevated heat risk. Most peripheral communities exhibit lower overall HRI values, with regional variations. Communities in the northeast of Baiyun and the north of Huangpu benefit from good vegetation cover and smaller populations, maintaining low HHI, HEI, and HVI values. Communities with a low HRI in Tianhe (e.g., Wushan, Longdong, and Yuangang) primarily owe their status to younger populations and good social adaptive capacity (low HVI), which effectively mitigates overall heat risk. Overall, the HRI forms a high-value ring centered on the historic core, decreasing progressively outward in the typical “high central–low periphery” concentric pattern. This reflects the compounded heat risk pressure faced by the historic urban core.

4.5. Coupling Coordination Degree Analysis

Based on HRI identification, we analyzed the spatial distribution of the CCD between heat risk and cooling facility supply using a coupling coordination model (Figure 7). Overall, few communities (12.5%) achieved high coordination, mainly in northern Yuexiu and southwestern Tianhe. Moderate incoordination clusters in northern, northwestern, and northeastern Baiyun. Barely coordinated communities (34.1%) are widely scattered across all districts. Moderately coordinated communities (41.7%) dominate, concentrated in central-southern Baiyun, southern Yuexiu, western Liwan, and northwestern and central Haizhu. Severely uncoordinated communities are rare (1.7%), located primarily in Huangpu’s Changling and Xinlong Town, warranting focused attention. Additionally, analysis at the district level reveals distinct CCD patterns (Figure 7b). Yuexiu exhibits the highest coordination (CCD = 0.50), followed by Tianhe (0.46), Haizhu (0.44), and Liwan (0.40) in the barely coordinated range. Baiyun (0.34) and Huangpu (0.32) show the lowest coordination, indicating persistent resource allocation imbalances.

Furthermore, global and local Moran’s I analyses were conducted to further explore CCD spatial patterns. Global Moran’s I is 0.59 (

Table 2. Global Moran’s I calculation results.

Category Value	Value
Moran’s I	0.59
Z-value	10.85
p-value	<0.01

), statistically significant at the 99.99% confidence level, confirming significant positive spatial autocorrelation of CCD. H-H clusters are concentrated in most of Yuexiu and southwestern Tianhe (e.g., Dadong, Datang, Meihuacun, Chebei, and Yuancun communities), indicating high overall CCD and good system coordination. L-L clusters dominate northern/northeastern Baiyun and northern Huangpu (e.g., Dayuan, Lianhe, Changling, and Longgui communities), signifying low system coordination and imbalance (Figure 8).

5. Discussion

5.1. Heat Risk Under the Hazard–Exposure–Vulnerability Framework

The formation of urban heat risk is a complex outcome of the long-term interactive evolution of natural environmental factors, human activities, and socio-structural elements [15]. Identification of heat risk in Guangzhou’s central urban area based on the HEV framework reveals a distinct triple convergence of hazard, exposure, and vulnerability in the old city districts. The high HHI is predominantly concentrated in older built-up zones. Historically, these areas prioritized excessive construction intensity, leading to diminished green space and increased impervious surface cover, resulting in pronounced urban heat island effects [49]. Furthermore, dense road networks and building layouts obstruct ventilation corridors, impairing local heat dissipation capacity [42]. This causes the thermal load to accumulate in traditional urban districts like Liwan and Haizhu persistently. This enclosed spatial morphology and historical planning legacies constitute the fundamental cause of spatially elevated heat hazard levels [50].

The HEI exhibits a spatial pattern characterized by high values in core old districts (Liwan, Yuexiu, and Haizhu) that decrease outwards. These areas feature high population density but extremely low green space coverage. This not only results in excessively exposed individuals per unit area but also exacerbates heat exposure due to the lack of ecological buffering functions like shading and cooling [33]. This essentially reflects the imbalance between supply and demand for urban public ecological services under high population pressure, highlighting a structural misalignment between the layout of ecological infrastructure and human settlement distribution [34]. The spatial distribution of the HVI aligns with findings by Chen et al. [51], showing high values concentrated primarily in old communities and economically disadvantaged neighborhoods. This occurs because vulnerable populations, constrained by limited economic capacity during urban development, are often concentrated in older neighborhoods with outdated infrastructure and poor housing conditions. These groups exhibit heightened sensitivity to high temperatures due to physiological status, limited information access, and constrained behavioral responses, forming a non-physical mechanism of high susceptibility through their limited adaptive capacity [19].

To more precisely identify areas of elevated heat risk for vulnerable populations, the results of the hazard–exposure–vulnerability framework are integrated into a unified and comprehensive heat risk index (HRI). Heat risk mapping analysis at the community scale reveals a concentric “high center–low periphery” distribution pattern in Guangzhou’s central urban area, consistent with the conclusions of Man et al. [16]. As the earliest developed and most intensively built areas, the urban core is characterized by insufficient green space coverage and high population concentration. The combined effect of intense human activity and limited ecological buffering sustains elevated land surface temperatures in these zones, accompanied by significant heat hazard and exposure levels [16]. Furthermore, historical factors like long-term urban development, population migration, and immigration have led to increasing proportions of elderly populations, children, and individuals with limited literacy, resulting in generally higher heat vulnerability [22]. This superposition of multiple risk factors makes the core subdistricts of the old city areas pronounced hotspots for the HRI.

While peripheral subdistricts generally exhibit lower HRI levels, the underlying causes show regional variation. Subdistricts in northeastern Baiyun and the northern Huangpu District benefit from smaller population bases, maintaining lower exposure levels. Concurrently, ample vegetation coverage enhances the ecological buffering effect of green spaces, which not only reduces surface temperature [52] but also improves thermal comfort for vulnerable populations [53]. In contrast, some low-value subdistricts in Tianhe District, despite having substantial populations, demonstrate good social adaptive capacity in the vulnerability dimension due to more recent development, younger demographic structures, and higher economic development levels. This effectively mitigates overall heat risk, showcasing a socially advantaged low-risk profile [16]. This study shows that the disparity in heat risk levels between old and newly developed urban areas, revealed through a heat risk index (HRI) focused on vulnerable populations, reflects a representative spatial pattern. This pattern offers valuable implications for urban heat risk identification and policymaking.

5.2. Implications of the CCD and DCC for Urban Planning

Assessing the spatial configuration coordination and decadal developmental adaptation of cooling shelters in Guangzhou’s central urban area from a supply–demand matching perspective holds significant importance for urban planning and decision making. The CCD analysis reveals a concentric hierarchical pattern of “high core–low periphery” for the coupling coordination between cooling shelters and HRI. Core subdistricts in old urban areas, despite facing high heat risk, are equipped with correspondingly dense cooling shelters, forming a spatial match of “high risk–high supply”. This highlights the proactive response capability and reflects the maturity of its spatial governance level concerning heat risk in the urban core [27]. In contrast, newly developed peripheral areas like northern Baiyun and northern Huangpu District exhibit a “low risk–low supply” pattern but generally suffer from low CCD. This lack of coordination may stem from structural imbalances between facility provision and risk levels or overall underdevelopment, leading to diminished system efficiency [34]. This phenomenon suggests that even in areas with currently relatively low heat risk, it is necessary to proactively establish a robust cooling shelter network to build resilience for future climate change impacts. To test the robustness of our model, we recalculated the HRI and CCD using entropy weight-derived weights instead of PCA-derived weights. The results showed that the HRI and CCD outcomes showed no substantial differences between the two weighting schemes, and the overall spatial patterns and trends remained consistent, confirming the stability of our approach (Figures S1 and S2).

While the CCD offers a static representation of the coordination between heat risk and cooling resources, it remains uncertain whether the current spatial pattern is the outcome of sustained proactive planning or short-term reactive adjustments. To further examine its developmental trajectory and adaptive capacity, we introduce the development coordination coefficient (DCC), designed to evaluate the dynamic relationship between the rate of cooling facility construction and changes in heat risk over the period from 2010 to 2020 (Figure 9). The DCC analysis shows clear differences in how well different areas of Guangzhou’s central urban districts can adapt to rising heat risk. Communities with higher coordination levels are mostly found in the old city. In these areas, cooling facilities have been added faster than heat risk has increased, showing that these places have a strong ability to plan ahead and respond. This is likely due to solid early-stage planning, stronger community structures, and relatively better public investment.

In contrast, areas with lower DCC values are typically newly developed or located on the urban fringe. These places currently have low heat risk and are not yet highly exposed, but they also lack enough cooling infrastructure, forming a “low risk–low supply” pattern. Although this is not an urgent issue now, it points to potential problems. As these areas grow quickly, experience more extreme heat events, or attract more vulnerable populations, heat risk could rise sharply. Without early infrastructure planning, they may become future hotspots of risk. Overall, the DCC highlights that managing heat risk depends not only on current coordination but also on whether infrastructure can keep pace with rising risk.

To more comprehensively identify the status and trajectories of thermal risk management across different urban areas and to inform more targeted governance and planning strategies, this study further integrates the current CCD and DCC to construct a four-quadrant analytical framework (Figure 10). The results show that core areas of old districts and some peripheral areas fall within the “high coordination–high matching” or “low coordination–high matching” quadrants. This indicates that these areas, leveraging well-established infrastructure and policy support, demonstrate strong adaptive resilience in heat risk management. However, areas like southern Huangpu and eastern Tianhe contain numerous communities clustered in the “low coordination–low matching” quadrant. These exhibit an imbalanced state characterized by rapidly escalating heat risk and severely lagging facility supply, reflecting structural deficiencies in public service coverage during rapid urbanization [34]. Furthermore, some communities falling into the “high coordination–low matching” quadrant, while exhibiting good current coordination, face a latent risk of imbalance as cooling shelter growth has demonstrably lagged behind the rising heat risk. Adaptive planning and supplementary facility construction should be initiated promptly for these areas.

The combined interpretation of CCD and DCC provides a valuable analytical perspective for formulating forward-looking and spatially responsive urban planning strategies. The findings highlight the importance of adequately considering the coordination between differentiated heat risk levels among vulnerable populations across different areas (e.g., old and new urban districts) and the development of cooling infrastructure. Such consideration is essential to enable adaptive and context-specific interventions. The analytical framework and differentiated planning strategies proposed in this study may also offer transferable strategies for other rapidly urbanizing cities facing mismatches between heat risk and resource provision for vulnerable groups.

5.3. Practical Implications for the General Public

The spatial patterns revealed in this study provide targeted guidance for residents in managing heat exposure. In old urban core areas where heat risk is high and cooling facilities relatively dense, residents, particularly vulnerable groups, can reduce exposure by actively using nearby public cooling spaces such as libraries, cultural centers, and parks and by choosing facility types that best meet their needs during extreme heat (e.g., indoor air-conditioned venues for rest or outdoor shaded parks for light activity). In peripheral areas where current heat risk is lower, but coordination between risk and resources is poor, residents should be aware of the limited availability of nearby cooling options and prepare alternative plans. The dynamic analysis further shows that some areas have persistent mismatches between risk and cooling resources, indicating the need for residents in these locations to remain attentive to changes in facility availability and urban development that may affect their access to cooling services. By aligning personal coping strategies with both the spatial distribution and temporal changes of public cooling resources, residents can enhance their ability to manage heat exposure risks in their local context.

5.4. Limitations

Although this study conducted a systematic exploration of heat risk assessment and spatial coordination analysis, several limitations remain to be addressed in future research: (1) The study focuses only on the central urban area of Guangzhou, with limited spatial and data coverage, which may weaken the representativeness of the findings. (2) There are certain constraints in data acquisition. This study was unable to obtain more representative indicators of individual adaptability and social vulnerability at the community level, such as resident income levels. (3) The analysis of cooling facilities was primarily based on POI density, without considering factors such as facility accessibility and service capacity. (4) Finally, while the coupling coordination degree model used in this study can assess the developmental synchronicity between heat risk and cooling facility provision, it still has significant methodological limitations. The model essentially measures the overall development level and synchronicity of the system, and lower coordination degrees may arise from different scenarios: on one hand, there may be typically unbalanced areas with high demand but insufficient supply; on the other hand, there may be developmentally lagging areas where both demand and supply levels are low. Therefore, it is necessary to avoid simply equating low values directly with resource allocation imbalance; instead, comprehensive judgment should be made by combining specific demand and supply levels.

6. Conclusions

This research developed an urban heat risk assessment system based on the HEV framework from the perspective of vulnerable population adaptation. By introducing coupling coordination models and development coordination coefficients, it systematically evaluated the matching relationship between heat risks and public cooling facilities and their temporal evolution trends, reaching the following conclusions: (1) heat risk in Guangzhou’s central urban area exhibits a typical concentric pattern of “high center–low periphery”, with densely populated old districts bearing the highest risk, while newly developed or peripheral zones remain at relatively low risk due to smaller population bases; (2) over the past decade, older urban areas have generally maintained good coordination between heat risk and cooling facility provision, while newly developed and peripheral areas show clear deficiencies, and some currently well-coordinated areas are also lagging in facility development, pointing to potential risks that warrant early intervention.

This study provides practical insights for urban managers in formulating heat-related environmental policies and holds valuable significance for improving urban thermal conditions and optimizing the planning of cooling resources, and it also offers transferable strategies for other rapidly developing cities facing similar challenges.