Assessing the Cooling Effects of Urban Parks and Their Potential Influencing Factors: Perspectives on Maximum Impact and Accumulation Effects

Abstract

Urban parks play an essential role in mitigating the urban heat island (UHI) effect driven by urbanization. A rigorous understanding of the cooling effects of urban parks can support urban planning efforts aimed at mitigating the UHI effect and enhancing urban sustainability. However, previous research has primarily focused on the maximum cooling impact, often overlooking the accumulative effects arising from spatial continuity. The present study fills this gap by investigating 74 urban parks located in the central area of Jinan and constructing a comprehensive cooling evaluation framework through two dimensions: maximum impact (Park Cooling Area, PCA; Park Cooling Efficiency, PCE) and cumulative impact (Park Cooling Intensity, PCI; Park Cooling Gradient, PCG). We further systematically examined the influence of park attributes and the surrounding urban structures on these metrics. The findings indicate that urban parks, as a whole, significantly contribute to lowering the ambient temperatures in their vicinity: 62.3% are located in surface temperature cold spots, reducing ambient temperatures by up to 7.77 °C. However, cooling intensity, range, and efficiency vary significantly across parks, with an average PCI of 0.0280, PCG of 0.99 °C, PCA of 46.00 ha, and PCE of 5.34. For maximum impact, PCA is jointly determined by park area, boundary length, and shape complexity, while smaller parks generally exhibit higher PCE—reflecting diminished cooling efficiency at excessive scales. For cumulative impact, building density and spatial enclosure degree surrounding parks critically regulate PCI and PCG by influencing cool-air aggregation and diffusion. Based on these findings, this study classified urban parks according to their cooling characteristics, clarified the functional differences among different park types, and proposed targeted recommendations.

1. Introduction

The ongoing process of rapid urbanization has profoundly altered surface thermal characteristics, contributing to the intensification of the urban heat island effect [1,2]. As a key component of urban green infrastructure, parks can modify surface conditions through vegetation coverage and enhance air circulation through appropriate spatial configurations [3,4], thereby playing a critical role in regulating urban microclimates and alleviating thermal stress [5,6]. Previous research has demonstrated that urban parks frequently generate a distinct “park cool island” effect [7], creating more comfortable spaces for recreation and outdoor activities [8]. Nevertheless, in the face of accelerating urban expansion and escalating heat island effects, improving the cooling efficiency of urban parks and optimizing their spatial distribution have become urgent priorities to improve urban environmental quality and promote sustainable urban growth [9,10,11].

Previous research has demonstrated that the cooling performance of urban parks presents clear spatial continuity and nonlinear distribution patterns, indicating that the cooling effect is shaped by the interaction of multiple factors and results in complex spatial variations [12]. However, traditional research has often focused on a single indicator, such as cooling intensity or cooling extent, making it challenging to comprehensively evaluate parks’ combined capacity to regulate the nearby thermal environment [13]. To address this limitation, recent research has highlighted the importance of assessing park cooling effects from two complementary perspectives: maximum impact and accumulative impact. The maximum impact perspective reflects the broader geographic extent of cooling, capturing both temperature regulation benefits and the effective cooling reach on a larger scale [14,15,16,17]. In contrast, the accumulative impact emphasizes the peak cooling intensity observed at specific sites, representing the most pronounced localized cooling effect under certain geographic and environmental conditions. This parameter is frequently utilized to evaluate the maximum temperature reduction potential of parks on a fine spatial scale [18,19]. These two dimensions reveal the spatial characteristics and functional differences of park cooling effects from different perspectives and are highly complementary. Nevertheless, most existing studies have concentrated on either a single indicator or a single spatial scale, often ignoring the relationship between cooling intensity and spatial extent. In addition, a comprehensive assessment framework that integrates both dimensions remains lacking, which limits the understanding of how park size, spatial structure, and surrounding environmental context collectively shape cooling effectiveness [14,20,21]. Thus, it is necessary to examine the cooling performance of urban parks and its driving mechanisms through the lenses of both maximum and accumulative impacts to gain a more holistic understanding of their temperature regulation functions.

Urban parks’ cooling benefits are shaped through the interaction of landscape composition characteristics and adjacent environmental factors. Existing studies indicate that park size has a direct influence on cooling intensity, with empirical evidence showing that larger parks tend to generate stronger cooling outcomes [3,22]. However, this relationship typically follows a logarithmic trend, where the incremental cooling benefit diminishes as the park area expands [14,23]. The park perimeter is also closely related to cooling performance, as a longer boundary enhances interaction with the surrounding environment and promotes heat exchange [24]. Additionally, the landscape shape index (LSI) is widely applied to quantify the morphological intricacy of park boundaries. Some studies have suggested that more complex park shapes are associated with lower cooling intensity, indicating a negative correlation [22]. However, other research has reported a positive relationship between LSI and cooling intensity [25], or found no significant correlation at all [8], highlighting uncertainty in how park shape complexity affects cooling performance. In addition to park size and shape, ecological factors, such as vegetation, water, and impervious surfaces within and around the park, also play critical roles. Previous studies revealed that the normalized difference vegetation index (NDVI) of parkland greenery is significantly positively correlated with cooling intensity, meaning that higher vegetation health contributes to stronger cooling [26,27]. Similarly, cooling intensity shows significant positive correlations with fractional vegetation cover (FVC) and the normalized difference water index (NDWI), whereas there is a significant negative correlation with the normalized difference impervious surface index (NDISI) [28]. Meanwhile, the built environment surrounding the park, including buildings and roads, also affects the park’s cooling ability due to the heat absorption and storage properties of artificial materials [29]. Numerical simulations further demonstrated that the surrounding environment influences the convection and diffusion of cool air within parks. Changes in building density can either enhance or restrict the dispersion of cool air, and a higher building density is often associated with a broader cooling range [19]. For instance, parks situated in central urban areas typically exhibit greater cooling intensity and broader zones of influence compared to those in other parts of the city, largely as a result of the increased density in the nearby built landscape [30]. Furthermore, the buffering effect of surrounding buildings on park cooling becomes increasingly significant during periods of extreme heat [31]. Research conducted in Xi’an indicated that the optimal cooling performance of urban green spaces occurs at building densities within 0.2 to 0.3 and floor area ratios of 2.5 to 3.0 [32]. In these conditions, nearby buildings indirectly enhance park cooling by regulating cold air diffusion pathways [33]. Together, these factors constitute a complex mechanism influencing the cooling performance of urban parks, offering a theoretical basis for further investigation through the dual lenses of maximum and accumulative impact.

With the increasing scarcity of urban land, maximizing the cooling benefits of urban parks within limited space has become a core issue in green infrastructure planning [4]. Previous studies have indicated that the cooling effects of urban green spaces do not exhibit a simple linear relationship with their size, but rather follow a pattern of diminishing marginal returns. To address this, the concept of the Threshold Value of Efficiency (TVoE) has been proposed [34], which defines the optimal size range that achieves resource allocation efficiency while meeting functional requirements. This approach has been widely applied to identify size thresholds for various urban landscape elements, such as green spaces and water bodies [11,35]. However, research on determining the optimal size threshold for urban parks remains limited, particularly studies that integrate both maximum cooling effects and cumulative cooling effects. Therefore, this study introduces the TVoE concept to identify critical park size thresholds under different cooling indicators, aiming to provide a scientific basis for urban heat mitigation and green space system optimization.

Jinan city, a typical inland provincial capital city in eastern China, has recently experienced frequent extreme heat events during summer, with urban temperatures often exceeding 37 °C and localized areas surpassing 40 °C [36,37]. This study examines 74 urban parks within Jinan’s central district to systematically evaluate their cooling effects, identify primary influencing factors, and explore how internal and external landscape elements regulate the urban thermal environment. The study aims to (1) quantitatively assess the cooling effects of urban parks based on two dimensions—maximum impact and accumulative impact; (2) determine the critical driving factors that influence the park cooling effect; and (3) classify parks based on cooling characteristics, reveal functional differences among different park types, and propose targeted optimization strategies. The findings aim to provide a scientific understanding and actionable recommendations for the design of urban parks and strategies to mitigate heat.

2. Materials and Methods

2.1. Study Area

Jinan, located in Shandong Province, China (36°01′–37°32′ N, 116°11′–117°44′ E), lies in the central part of the North China Plain and serves as the economic center of the Shandong Peninsula urban agglomeration. According to the Köppen climate classification, Jinan belongs to the temperate monsoon climate zone, characterized by hot summers with significant rainfall and cold, dry winters. The city has a mean annual temperature of approximately 15 °C and total annual rainfall between 650 and 700 mm. During summer, maximum air temperatures often reach 37–39 °C, with some areas exceeding 40 °C. The permanent population grew from 5.92 million in 2000 to 9.44 million in 2024, accompanied by an urbanization rate rise from 36.91% to 75.3%. This study concentrates on Jinan’s central urban area (Figure 1), encompassing approximately 536 km².

2.2. Data Sources

In this study, remote sensing data were acquired from Landsat 8 OLI_TIRS imagery captured on 16 September 2021, with an average cloud cover of 2.56%. The images were downloaded from the Geospatial Data Cloud Platform (http://www.gscloud.cn/) and applied to land surface temperature (LST) retrieval and analysis. Park boundary data were sourced from the Baidu Maps Open Platform (https://lbsyun.baidu.com/). Using the API interface, park-related data within the area of interest (AOI) for 2021 were extracted. High-resolution remote sensing images and park location data were combined, and visual interpretation methods were applied to identify the location and boundary vector information of urban parks located in Jinan’s city center. Following established research screening criteria [21], several parks were excluded to minimize the influence of external natural factors and reduce errors caused by remote sensing resolution. These excluded parks included small parks located within large parks or surrounded by green spaces, parks adjacent to large rivers or mountainous green areas, parks surrounded by farmland, as well as parks smaller than 0.5 ha in size. After screening, a total of 74 urban parks met the selection requirements for this study. In addition, building contour data and building height information for the central area of Jinan were collected using Python (Version 3.9) and web crawler technology based on the Baidu Maps Open Platform. Using geospatial analysis methods, both two-dimensional and three-dimensional urban building information was systematically extracted.

2.3. Methods

This study focused on 74 urban parks situated within the downtown area of Jinan as the primary research sites. The cooling effects of these parks were evaluated from two dimensions: maximum impact and accumulative impact. Based on this, 10 representative potential influencing factors were identified, incorporating both internal park attributes and characteristics of the surrounding built environment. A systematic analysis was conducted to identify the primary factors influencing park cooling performance. Furthermore, the study investigated how different factors influence park cooling mechanisms and proposed targeted strategies for improving park planning and design. The overall research framework is illustrated in Figure 2.

2.3.1. LST Retrieval

In this study, remote sensing images were first processed for cloud masking. Radiometric calibration and atmospheric correction were applied to improve the radiometric accuracy of the imagery [38]. Considering the calibration difficulties associated with Landsat 8’s TIRS Band 11, Band 10 was adopted as the main thermal infrared data source for retrieving LST in this study [39]. The Radiative Transfer Equation (RTE) method was subsequently employed to derive the LST values [40]. This algorithm simulates the radiative transfer process between the land surface and atmosphere, removing the atmospheric influence on thermal radiation and estimating LST based on the retrieved surface radiance values [41,42,43,44]. On this basis, the blackbody radiance$B\left(T_S\right)$ was calculated using the inverse function of Planck’s law [21]. Finally, the LST was derived through the Planck equation (Equation (3)) [45].

$$L_{\lambda }=\left[\epsilon B\left(T_S\right)+\left(1-\epsilon \right)L\downarrow \right]\tau +L\uparrow$$

(1)

$$B\left(T_S\right)={\displaystyle \frac{\left[L_{\lambda }-L\uparrow -\tau \left(1-\epsilon \right)L\downarrow \right]}{\tau \epsilon }}$$

(2)

$$T_S={\displaystyle \frac{K_2}{ln\left(\frac{K_1}{B\left(T_S\right)}+1\right)}}$$

(3)

where $L_{\lambda }$ represents the thermal infrared radiance received by the satellite sensor;$\epsilon$ denotes land surface emissivity; $L\uparrow$ refers to the upward atmospheric radiance; $L\downarrow$ describes the downward atmospheric radiation component reflected by the ground surface; and $\tau$ represents atmospheric transmittance in the thermal infrared band. The values of $L\uparrow$, $L\downarrow$, and $\tau$ were obtained from the NASA website (http://hls.gsfc.nasa.gov/). The calibration constants $K_{1 }\left(\mathrm{W}{\mathrm{m}}^{-2}\mathrm{s}{\mathrm{r}}^{-1}\mathsf{\mu }{\mathrm{m}}^{-1}\right)$ and $K_2\left(K\right)$ were set at 774.89 and 1321.08, respectively. In addition, this study applied hotspot analysis in ArcGIS (Version 10.6) to identify high-temperature hotspots and low-temperature cold spots based on LST data. This spatial statistical approach effectively reveals the heterogeneity of park cooling effects and provides a more intuitive representation of urban thermal patterns compared to relying solely on temperature values [46,47].

2.3.2. Quantification of Park Cooling Effects

A 300 m buffer was established around the boundary of each urban park to quantify the spatial extent of its cooling effect. This distance was selected because it aligns with the 30 m spatial resolution of Landsat 8 imagery, allowing the buffer to be divided into 10 equal sub-zones for detailed analysis. This buffer was further divided into 10 sub-buffer zones, each with a width of 30 m. The variation in LST with increasing distance from the park boundary was then examined. Distance to the park boundary (r) was designated as the predictor variable, while mean LST (T) functioned as the response variable. A third-order polynomial function was employed to model the relationship between LST and distance [48] (Figure 3). In this equation, T(r) represents the LST at a given distance r extending outward from the park boundary, where r refers to the distance between the park boundary and the corresponding buffer zone, and $a,b,c$, and $d$ are constants.

$$T\left(r\right)=a{r}^3+b{r}^2+cr+d$$

(4)

As the distance from the park increased, LST generally exhibited a rising pattern; however, the pace of growth gradually declined until stabilizing. The first derivative of the T(r) function, denoted as T′(r), was applied to identify the distance beyond which the park’s cooling influence significantly weakened. The maximum cooling distance (D) was identified as the point at which T′(r) first reaches zero, along with the associated surrounding land surface temperature. If no clear turning point was observed, the location corresponding to the minimum value of T′(r) was used as D. The temperature difference between T_D and the mean land surface temperature within the park (T_P) was calculated as ΔLST. Based on this, four cooling indicators were calculated: Park Cooling Area (PCA), Park Cooling Efficiency (PCE), Park Cooling Intensity (PCI), and Park Cooling Gradient (PCG). Among them, PCI and PCG were used to represent accumulative impacts, while PCA and PCE reflected maximum impacts [12,15,21]. The proposed indices provide a multi-dimensional and objective assessment of urban parks’ cooling performance.

$$PCA=S_{max}$$

(5)

$$PCE=S_{max}/S_{park}$$

(6)

$$PCI={\displaystyle \frac{D\times T_D-{\int }_0^{D}T\left(r\right)dr}{D\times T_D}}$$

(7)

$$PCG={\displaystyle \frac{D\times T_D-{\int }_0^{D}T\left(r\right)dr}{L}}$$

(8)

where PCA represents the maximum area (S_max) around the park where significant cooling effects were observed. PCE describes the cooling efficiency relative to the park’s area, calculated as the proportion of the cooling footprint (S_max) to the actual park area (S_park). The calculations of PCI and PCG describe the degree of spatial continuity in the cooling effect provided by the park. Specifically, the term $D\times T_D$ represents the theoretical cumulative surface temperature within the maximum cooling distance in the absence of the park, while ${\int }_0^{D}T\left(r\right)dr$denotes the corresponding cumulative LST considering the park’s presence. The difference, $D\times T_D-{\int }_0^{D}T\left(r\right)dr$ quantifies the realized cooling benefit provided by the park. PCI indicates the percentage reduction in surface temperature within the maximum cooling extent, with higher values corresponding to more pronounced cooling performance. PCG reflects the average temperature reduction per unit distance, where greater values imply stronger cooling intensity and greater thermal mitigation capacity.

2.3.3. Influential Factors Affecting the Park Cooling Effect

Earlier research has demonstrated that urban park cooling performance is influenced by a range of factors, such as internal landscape configuration, geometric characteristics, and adjacent environmental conditions [49]. Drawing from prior studies [50,51,52,53], 10 candidate influencing variables were initially identified. These factors covered two dimensions: the internal attributes of parks and the attributes of the surrounding built environment. Specifically, they included Park_Area, Park_Perimeter, Park_NDVI, Park_MNDWI, and Park_LSI, as well as Buffer_SVF, Buffer_BD, Buffer_MBH, Buffer_FAR, and Buffer_BHSTD (

Table 1. Factors affecting the cooling effect of urban parks.

Indicator Category	Influencing Factors	Acronyms	Formula and Scope	Definition
Landscape component features of urban parks	Area of the park	Park_Area	>0.5 ha	Area of the urban park (unit: ha).
	Perimeter of the park	Park_Perimeter	>0 km	Area of the urban park (unit: km).
	Normalized difference vegetation index	Park_NDVI	$\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}={\displaystyle \frac{R_{nir}-R_{red}}{R_{nir}+R_{red}}}$	NDVI is used to determine the density of green on a patch of land. Rnir and Rred denote the reflectance of near-infrared and infrared bands, respectively.
	Modified normalized difference water index	Park_MNDWI	$\mathrm{M}\mathrm{N}\mathrm{D}\mathrm{W}\mathrm{I}={\displaystyle \frac{R_{green}-R_{mir}}{R_{green}+R_{mir}}}$	MNDWI is an indicator used to determine the open water area. Rmir and Rgreen denote reflectance in near-infrared and infrared bands, respectively.
	Landscape shape index	Park_LSI	$\mathrm{L}\mathrm{S}\mathrm{I}={\displaystyle \frac{P}{2\sqrt{\pi \times S_{park}}}}$	The landscape shape index of each park. P represents Park_Perimeter, and Spark represents Park_Area.
Built environment characteristics within urban park buffers	Sky view factor	Buffer_SVF	$\mathrm{S}\mathrm{V}\mathrm{F}=1-{\displaystyle \frac{\sum _{i=1}^n\mathit{sin}{\gamma }_i}{\mathrm{n}}}$	SVF quantifies the unobstructed sky proportion within a 300 m buffer around the park. ${\gamma }_i$ denotes the terrain-induced adjustment to the azimuth angle, where i ∈ [1, n] and n = 36.
	Building density	Buffer_BD	$\mathrm{B}\mathrm{D}={\displaystyle \frac{A_{footprint}}{A_{site}}}$	Proportion of building footprint within the 300 m buffer zone to the total area of that buffer zone. $A_{footprint}$ represents the total building footprint; $A_{site}$ represents total site area.
	Mean building height	Buffer_MBH	$\mathrm{M}\mathrm{B}\mathrm{H}={\displaystyle \frac{1}{\mathrm{n}}}{\displaystyle \sum _{i=1}^n}{H}_i$	MBH within the 300 m buffer zone around the park perimeter. Hi represents the height of the building (Building Height), and n is the number of buildings in the buffer zone.
	Floor area ratio	Buffer_FAR	$\mathrm{F}\mathrm{A}\mathrm{R}={\displaystyle \frac{\sum _{i=1}^n{(A}_{footprint,i}\times H_i)}{A_{buffer}}}$	Building FAR within the 300 m buffer zone around the park perimeter. $A_{buffer}$ represents the total area of the 300 m buffer strip (m2).
	Building height standard deviation	Buffer_BHSTD	$\mathrm{B}\mathrm{H}\mathrm{S}\mathrm{T}\mathrm{D}=\sqrt{{\displaystyle \frac{1}{\mathrm{n}}}{\displaystyle \sum _{i=1}^n}{\left(H_i-MBH\right)}^2}$	Standard deviation of building heights within the 300 m buffer zone around the park perimeter.

). To verify the absence of substantial multicollinearity among the predictors, variance inflation factors (VIFs) were calculated using SPSS software (Version 26.0), with the 10 influencing factors designated as predictors and the four cooling effect indicators (PCA, PCE, PCI, and PCG) as responses. A VIF threshold of 5 was used to identify potential multicollinearity issues [4]. The results showed that all factors exhibited VIF scores under 5, indicating an absence of notable multicollinearity. Therefore, these indicators were suitable for subsequent correlation and mechanism analysis.

2.3.4. Typological Classification of Urban Parks

To more accurately discern variations in cooling effectiveness and spatial attributes among urban parks, this study applied a clustering analysis approach [21]. First, the 4 cooling effect indicators (PCA, PCE, PCI, and PCG) were standardized to remove discrepancies caused by differing units and measurement scales. This facilitated a more rigorous and objective comparison of cooling heterogeneity across parks. Based on the normalized data, Ward’s hierarchical clustering algorithm was implemented to categorize the 74 urban parks. This method minimizes the within-cluster sum of squared errors (SSE) to increase similarity among parks within the same cluster while maximizing differences between clusters. The resulting groups were defined as “cooling efficiency clusters” [12]. In Ward’s method, the distance d(u, v) (Equation (9)) between clusters u and v was defined as the increase in SSE (Equation (10)) resulting from merging the two clusters.

$$d\left(u,v\right)=SSE\left(\left(u,v\right)\right)-SSE\left(u\right)-SSE\left(v\right)$$

(9)

$$SSE=\sum _{j\in C_k}\sum _i^n{\left(x_{ij}-{\displaystyle \frac{1}{N_k}}\sum _{j\in C_k}{x}_{ij}\right)}^2$$

(10)

where n represents the dimensionality of x, $N_k$ indicates the sample size within cluster k, and $C_k$ refers to the set of samples within cluster k. To identify the optimal cluster count, multiple tests were conducted by setting the cluster number from 2 to 10. Variance analysis was performed to evaluate the inter-group differences under each clustering scheme. The entire clustering analysis process and analysis of variance were performed using IBM SPSS Statistics 26.0 software. Based on the results of clustering analysis and analysis of variance, the classification of 74 urban parks into three clusters demonstrates stable and reliable grouping, with a significance level of 0.05 for mean differences. The results showed that when the parks were divided into three clusters, the differences between the groups were most significant, and the distribution of indicators within each group was relatively concentrated, supporting the robustness of the clustering solution. This structure provided a good balance between classification accuracy and interpretability. Meanwhile, the three-cluster scheme demonstrated clear distinctions in the cooling effect among different park types and offered practical value for identifying functional differences and guiding park optimization.

3. Results

3.1. The Spatial Heterogeneity of LST and Urban Parks’ Cooling Effect

The land surface temperature in the central urban area of Jinan ranged from 25.08 °C to 47.10 °C, with an average temperature of approximately 35 °C (Figure 4a). Overall, the LST showed significant spatial heterogeneity across the study area. Elevated temperatures were predominantly observed in densely populated residential neighborhoods and industrial districts, particularly within the central and northeastern sections of the city. In comparison, low-temperature zones were primarily distributed around urban parks, mountainous areas, aquatic features, including rivers, lakes, and additional ecological components. The famous Baoshan Park and Jinan Forest Park also showed the presence of significant low-temperature zones within the parks (Figure 4c,d). To better understand the spatial clustering of LST within the study area, a hot spot analysis was conducted to identify high-temperature hot spots and low-temperature cold spots [54]. The results (Figure 4b) indicated that 62.3% of cold spots with a confidence level above 90% overlapped with urban parks, highlighting the significant contribution of parks in mitigating LST and enhancing the local thermal environment. Specifically, the average LST within parks ranged from 28.93 °C to 36.00 °C, with the maximum ΔLST reaching 7.77 °C.

The 4 indicators reflecting the cooling performance of urban parks revealed pronounced spatial disparities, reflecting clear spatial heterogeneity in cooling performance across different parks (Figure 5). Overall, PCI ranged from 0.0004 to 0.0957, with an average of 0.028. PCG varied between 0.01 °C and 3.30 °C, averaging 0.99 °C. PCA varied between 3.51 ha and 308.32 ha, with an average of 46.00 ha. PCE ranged from 0.27 to 25.93, with a mean value of 5.34. High values of PCA were mainly concentrated in large parks located in the city center and peripheral areas, indicating their advantage in expanding the cooling influence area. In contrast, high PCE values were more dispersed, with several concentrated in densely built-up central zones such as Heroes Hill Green Plaza, underscoring the critical role of small and medium-sized parks in maximizing cooling efficiency relative to their area. Meanwhile, PCI and PCG exhibited a more fragmented spatial pattern. Some high-value areas were located near natural terrains or water bodies, suggesting that surrounding environmental conditions had a synergistic effect on cooling intensity and gradient. Notably, although parks in the city center were generally smaller in size, they showed higher PCE and PCG values, indicating their strong capacity to mitigate local temperatures in densely developed urban areas. In contrast, parks located on the urban periphery, owing to their larger spatial extent, made substantial contributions to PCA and PCI, reflecting the dual role of park size in enhancing both the spatial extent and magnitude of cooling effects. Overall, the spatial distribution of park cooling benefits exhibited a clear pattern characterized by “high efficiency in the center, large coverage on the periphery.”

3.2. Influencing Factors of Park’s Cooling Effect

This study employed Pearson correlation analysis to examine the influence of park attributes and surrounding environmental factors on the cooling performance of urban parks (Figure 6). From the perspective of maximum impact, PCA was strongly and positively associated with Park_Perimeter, Park_Area, and Park_LSI, corresponding to correlation coefficients of 0.858, 0.441, and 0.766 (p < 0.01). These results indicate that parks with larger areas, longer perimeters, and more complex shapes are generally associated with broader cooling coverage. In contrast, PCE showed significant negative correlations with both Park_Perimeter and Park_Area, suggesting that as park size increases, cooling efficiency per unit area tends to decline. In larger parks, the diffusion of cold air reduces the localized cooling intensity. No significant correlation was found between Park_LSI and PCE, implying that shape complexity mainly affects cooling area rather than unit cooling efficiency. Meanwhile, Park_NDVI showed weak and negative correlations with both PCA and PCE, suggesting limited influence of vegetation on maximum impact metrics. From the accumulative impact perspective, both PCI and PCG were influenced by multiple factors. Park_Perimeter and Park_Area remained positively correlated with both indicators, confirming that larger parks not only expand cooling area but also enhance cooling intensity and spatial gradient. However, surrounding built environment factors play a more significant role in cumulative impacts. Buffer_FAR, Buffer_MBH, and Buffer_BHSTD demonstrated significant positive correlations with PCI and PCG (p < 0.05), while Buffer_SVF exhibited a significant negative correlation with both indicators (p < 0.01). These results indicate that areas with higher building density and more high-rise structures tend to experience a stronger urban heat island effect, which in turn amplifies the relative cooling effect of parks. In contrast, areas with a higher sky view factor dissipate heat more quickly, reducing cooling intensity and gradient near parks. Park_NDVI showed a weak positive correlation with PCI and PCG, suggesting a mild influence of vegetation on cumulative cooling performance. MNDWI did not exhibit significant correlations (p > 0.1).

In summary, among the maximum impact indicators, PCA was primarily driven by Park_Perimeter, Park_Area, and Park_LSI, highlighting the importance of park geometric features in determining maximum cooling coverage. PCE tended to reach higher levels in smaller-scale parks, reflecting its negative effect on scale change. Among the accumulative impact indicators, beyond park size and boundary characteristics, surrounding built environment factors demonstrated stronger explanatory power in regulating cooling intensity and gradient. In particular, variations in building density and spatial enclosure were key factors influencing the processes of cold air accumulation and dispersion.

3.3. Classification of Cooling Types of Urban Parks

According to the outcomes of cluster analysis and variance testing, which indicated significant mean differences (p < 0.05), the 74 urban parks in this study were categorized into three distinct clusters reflecting varying levels of cooling effectiveness. These groups reflected clear differences in both cooling effectiveness and structural characteristics. To further explore these differences, the mean values of the cooling factors and influencing indicators were compared across the three clusters (Figure 7).

Cluster 1, which comprised 36.49% of all parks, showed a moderate level of overall cooling effectiveness. Most of its normalized indicator values ranged between 0 and 0.5. Although this group did not perform strongly in terms of accumulative cooling impact, it had relatively high mean values for geometric attributes such as Park_Area, Park_Perimeter, and Park_LSI, exhibiting an average area of 40.33 ha and an average perimeter of 2564.27 m. This suggests that the parks in this cluster were generally large, with complex shapes that enhanced their spatial integration. Parks in this group also exhibited higher values for Park_MNDWI, Buffer_FAR, and Buffer_BD, indicating higher surrounding building density and water body proportions, which helped maintain a basic cooling function within a certain range.

Cluster 2 accounted for 35.13% of all parks and demonstrated the strongest accumulative cooling performance. Both PCI and PCG values were the highest among the three clusters. However, this cluster had relatively low values of PCA and PCE, suggesting a limited maximum cooling range and lower cooling efficiency per unit area. Regarding the influencing factors, parks in this group had the smallest values for geometric and ecological variables such as Park_Area, Park_Perimeter, Park_LSI, and Park_NDVI. Despite their smaller size and lower vegetation coverage, these parks had the highest Buffer_SVF values, indicating more open space and better ventilation around them. Such spatial openness promoted cold air diffusion, allowing these parks to achieve significant cooling intensity and gradient within a relatively small area, thus enhancing their ability to regulate the local thermal environment.

Cluster 3, representing 28.38% of the total sample, featured higher mean values for key structural and ecological factors, including Park_Area, Park_Perimeter, Park_LSI, and Park_NDVI, compared to Cluster 2. This reflects their larger overall size and greater vegetation coverage. Parks in this cluster achieved higher PCA values, indicating a more significant cooling effect over a wide area. However, their PCI and PCG levels were lower than those of Cluster 2, suggesting weaker performance in accumulative cooling intensity and gradient. The main advantage of Cluster 3 lies in its ability to expand the spatial extent of cooling through larger sizes and complex shapes. Nevertheless, insufficient openness and poor cold air circulation limited the potential of these parks to improve cooling intensity.

4. Discussion

4.1. Quantify Urban Parks’ Cooling Effects from Both Maximum and Accumulative Perspectives

This study systematically examined the differences in urban park cooling effects and their dominant mechanisms from two dimensions: maximum impact and accumulative impact. Our results further confirmed the findings that geometric attributes such as park size and shape complexity play a decisive role in determining the maximum cooling range [12,55]. In the study area, large parks with complex boundaries showed clear advantages in PCA and PCE. The largest PCA reached 308.32 ha, and the highest PCE was 25.93. This indicates that spatial scale and boundary shape largely dominated the park’s cooling effect at the maximum impact level. However, at the accumulative impact level, the importance of park size and boundary shape weakened significantly. Instead, spatial openness and ventilation conditions became the main factors. Some small but well-ventilated parks performed strongly on PCI and PCG. This suggests that accumulative cooling depends more on cold air diffusion and local thermal regulation. These observations are in strong agreement with the conclusions of Du et al. [15].

Based on the classification of cooling efficiency clusters, this study further revealed functional differences among urban park types in terms of maximum and accumulative impacts. Cluster 2 parks were generally smaller, with lower PCA and PCE values, indicating limited cooling coverage. However, they showed the strongest performance in PCI and PCG, highlighting the positive role of good spatial structure and ventilation openness in enhancing accumulative cooling intensity and gradient. This finding aligns with Li et al.’s study [56], which emphasized improving local wind environments and cooling effects through optimized ventilation corridor layouts. In contrast, Cluster 3 parks were larger and more complex in shape, with clear advantages in PCA and PCE. This supports Blachowski et al.’s view [57] that large-scale green spaces provide stronger cooling coverage and environmental regulation. Nevertheless, Cluster 3 had generally lower PCI and PCG values, suggesting that mere expansion of area, without good structural design and cold air diffusion paths, struggles to achieve synergistic effects in accumulative cooling. This further confirms that improper spatial layout of green spaces weakens overall cooling performance [58,59]. Overall, the differences revealed by the maximum and accumulative impact indicators reflect the varied functional emphases of urban parks under different scales and structural conditions. This underscores the need to consider size, structure, and openness comprehensively in practical planning and design. Minimizing dependence on a single indicator while promoting multi-dimensional optimization strategies is essential to maximizing the overall cooling benefits of urban parks.

4.2. The Driving Factors of Cooling Effect in Urban Parks

This study systematically assessed the relative influence of park landscape composition and the adjacent urban fabric on their cooling performance. The results showed clear differences between the maximum impact and accumulative impact dimensions. Specifically, park area and perimeter, as key geometric attributes, positively influenced both the cooling range and intensity at the maximum impact level. As the area and perimeter increased, the surface temperature inside parks dropped significantly, the cooling influence extended, and the cooling intensity per unit distance strengthened. These findings align with previous studies by Peng [12] and Wu [52]. Additionally, existing research has indicated that boundary complexity can enhance maximum impact cooling by increasing the contact area between the park and its surroundings, promoting heat exchange [55]. This view is generally consistent with the overall trend found in this study. However, some studies have reported negative or non-significant correlations between LSI and cooling intensity [7,22]. Jaganmohan et al. [54] suggested that park size may moderate the effect of shape complexity on maximum impact, which might explain the inconsistencies among different studies.

In contrast, the cooling effects at the accumulative impact level relied more on the combined regulation of spatial structure, ventilation conditions, and surrounding environmental factors. The study found that some small but well-ventilated parks performed strongly in PCI and PCG indicators, indicating that accumulative effects were largely driven by cold air diffusion and continuous local thermal environment improvement. Regarding vegetation, the NDVI showed generally weak correlations with the cooling indicators, which is consistent with the results reported by Qiu et al. [55]. However, other studies have suggested that NDVI might indirectly enhance cooling by improving vegetation conditions. For instance, Xiao et al. [39] reported a positive association between NDVI and PCI, while Cui [60] emphasized that NDVI’s effect varies due to indicators like observation platform, season, and atmospheric conditions, which could explain differences in accumulative impact results across studies. In addition, although this study did not find significant correlations between MNDWI and cooling effects, the existing literature indicates that water bodies positively contribute to local cooling [23,52]. Regarding the surrounding built environment, building height and height variability showed significant negative correlations with accumulative cooling indicators. Tall buildings block ventilation paths and reduce airflow, weakening cold air diffusion [29,61,62]. In this study, Buffer_MBH and Buffer_BHSTD correlated significantly with PCI and PCG, highlighting the important role of building height and spatial variability in accumulative cooling performance. Additionally, both Buffer_FAR and Buffer_SVF exhibited significant correlations with PCI and PCG. FAR reflects building density and vertical intensity; higher values indicate more enclosed spaces, which hinder air exchange and heat dissipation [63]. SVF represents the proportion of visible sky above parks; higher values indicate a more open environment, favorable for radiative cooling and microclimate regulation [64]. In comparison, Buffer_BD had no significant effect on accumulative impacts. This may be because building density mainly captures horizontal distribution and does not fully reflect three-dimensional features like height, shape, or arrangement. As a result, it poorly represents actual microclimate influences such as ventilation and shading.

4.3. Implications for Park Planning and Design

Urban parks serve as essential ecological infrastructure for mitigating urban heat environments, significantly contributing to reductions in LST and improvements in local microclimate conditions [11,65]. However, with growing urban land scarcity, maximizing the cooling benefits of parks within constrained spaces has become a major challenge for green space planning [4]. Previous research has established that the cooling effects of urban green spaces are strongly influenced by their size, spatial structure, and layout, while different cooling indicators often require distinct optimization approaches [66]. Although large-scale parks usually have a more significant cooling capacity [40], this study revealed a “diminishing marginal benefit” pattern for cooling intensity relative to park area and perimeter, with PCI plateauing as park size increases (Figure 8). Consequently, indiscriminate park expansion is neither an efficient nor a sustainable approach to optimization. To address this issue, the Threshold Value of Efficiency (TVoE) concept has been proposed [10,67], aiming to determine an optimal park size range that balances costs and benefits [35]. When park area exceeds the TVoE, cooling effectiveness tends to plateau or even decline [68]. Fitting analysis results (Figure 9) showed that the TVoE varies depending on different cooling indicators. Specifically, at the maximum impact level, PCA increased with park size but tended to stabilize beyond 14.46 ha, which can be considered a reasonable lower limit to maximize cooling coverage. At the accumulative impact level, cooling efficiency decreased with a larger park size, with the optimal PCI observed at 3.98 ha. Therefore, future park planning should flexibly align with different TVoE standards based on land availability and specific cooling demands, ensuring a balance between cooling coverage and intensity. Relying solely on a single indicator may lead to inefficient land use, while a coordinated, multi-dimensional optimization approach is essential to fully realize the comprehensive cooling benefits of urban parks. Based on the results of park cooling performance clustering, this study further proposed differentiated spatial layout strategies for various park types.

(1)

Cluster 1—Large-Scale Parks with Basic Cooling Function

The average area of this type of park reaches 40.33 ha, and its perimeter and morphological complexity are also at a high level, which provides good spatial embedding and cooling coverage. However, their accumulative cooling performance was moderate. A higher water surface ratio (Park_MNDWI) and building density (Buffer_BD) in surrounding areas indicated a certain degree of spatial enclosure. Therefore, for this park type, further internal spatial restructuring is recommended. Optimizing waterbody distribution and enhancing ventilation corridors could help unlock greater potential for improving the local thermal environment.

(2)

Cluster 2—Small-Scale, High-Efficiency Parks

Although this cluster included the smallest parks with the simplest structures, it showed the best performance in accumulative cooling indicators. This reflects the advantage of good spatial openness and effective cold air diffusion, especially with the highest surrounding SVF values, indicating excellent environmental permeability. Thus, this type of small, high-efficiency park is recommended for densely built-up areas or locations with limited land resources. Park size should align with the PCE-based TVoE (3.98 ha) to achieve efficient, localized cooling with minimal space demand and meet thermal comfort needs for high-density populations.

(3)

Cluster 3—Large-Scale Parks with Extensive Coverage

Parks in this group had the largest area, perimeter, and vegetation coverage, with clear advantages in the maximum cooling range (PCA). However, they showed weaknesses in accumulative cooling performance, mainly due to enclosed spatial structures and poor ventilation conditions. In areas where land resources allow, this type of large park is recommended, with a suggested minimum size of 14.46 ha, based on the PCA-related TVoE. During the planning process, more attention should be given to improving internal spatial connectivity and creating external cold air corridors. These measures help avoid structural problems such as “large area but weak effectiveness” and allow the park to fully contribute to both cooling coverage and microclimate regulation.

The significance of this study extends well beyond the case of Jinan and provides critical planning guidance for cities with comparable climatic conditions and urban morphology. This is particularly relevant for rapidly urbanizing regions within the temperate monsoon climate zone, where high building density and severe land constraints make traditional strategies such as simply expanding green space to improve cooling, neither feasible nor sustainable. By introducing the concept of the TVoE within a differentiated park layout framework, this study provides a practical method to maximize cooling benefits in space-constrained cities. This approach improves both ecological performance and land-use efficiency, helping address the twin challenges of climate adaptation and limited urban land. Furthermore, its applicability extends to East and Southeast Asian cities, where accelerating urbanization combined with global warming has intensified thermal stress, which underscores the practical value of this optimization paradigm in guiding climate-sensitive urban green infrastructure planning. Accordingly, to improve the cooling performance of urban parks, both maximum and accumulative impacts should be considered. It is essential to balance the TVoE with the structural characteristics of different park types. A flexible and site-specific approach to green space planning is needed to promote differentiated and targeted layouts. By matching appropriate park size with functional design, it is possible to enhance ecological benefits while optimizing land use efficiency. This strategy provides a more effective solution to address the growing challenges of urban heat environments.

Furthermore, studies on the cooling effects of urban green spaces worldwide have commonly focused on the marginal benefits of area expansion, yet the regulating mechanisms vary significantly across different climatic conditions and urban morphologies. Recent studies indicate that in temperate and humid climates, the expansion of green space, combined with optimized ventilation corridors and enhanced vegetation diversity, can markedly improve cooling efficiency [69,70,71]. However, in arid or semi-arid environments, the cooling benefits of vegetation expansion are often limited by restricted evapotranspiration, making the integration of water features a critical strategy for improving the thermal environment [72,73]. In comparison, this study finds that within the warm temperate monsoon climate zone, cooling intensity tends to plateau once park size exceeds a certain threshold, with surrounding spatial morphology playing a significant role in modulating cooling performance. These findings suggest that although the marginal effects of green space areas are generally observed, their manifestation and dominant drivers are highly dependent on regional climatic characteristics and spatial structure. Therefore, when formulating green space planning strategies, it is essential to consider local conditions and promote designs that are responsive to climatic factors and optimized spatial configurations to maximize cooling benefits.

4.4. Limitations and Recommendations for Future Research

This study conducted a systematic analysis of the cooling effects of urban parks in the central area of Jinan using remote sensing data. The evaluation considered both maximum and accumulative impacts to examine the role of parks in mitigating urban heat stress, aiming to offer evidence-based guidance for the design and administration of urban parks and greenscapes in similar contexts. Nonetheless, several limitations remain. First, the analysis relied predominantly on remote sensing data captured during summer daytime, restricting the ability to capture park cooling performance across other seasons or different times of day. Since surface temperature is highly sensitive to seasonal and diurnal variations, using single-period data may not fully reveal the year-round or all-day cooling effects of urban parks. Future research should incorporate multi-season and multi-period datasets to better capture the spatiotemporal dynamics of park cooling effects and improve the accuracy, timeliness, and applicability of the results. Second, the analysis of the factors influencing park cooling performance in this study remains incomplete, as socioeconomic, cultural, and ecological variables were not fully considered. In addition to indicators such as population density, road density, and urban heat emissions, which reflect spatial patterns and thermal characteristics of the city, ecological attributes such as biodiversity, habitat connectivity, and vegetation quality are also critical. These factors not only influence cooling mechanisms but also contribute to ecosystem services, which are increasingly emphasized in modern urban planning. Future research should integrate these multidimensional indicators to construct a more robust and holistic analytical framework, thereby deepening the understanding of both thermal regulation and ecological benefits while enhancing practical applications. Moreover, given the significant role of water bodies in regulating surface temperatures, this study excluded parks located near large water bodies to avoid interference when evaluating the independent cooling effect of parks, resulting in a final sample of 74 parks. It should be noted that the MNDWI results showed minimal water presence within these parks, and no significant relationship was found between water features and cooling factors, suggesting limited interference from water bodies in this research. Nevertheless, previous studies have shown that the integration of water bodies and green spaces contributes to enhancing cooling performance [34,74,75]. Therefore, future research should explore the integrated layout of water and green spaces to comprehensively assess their combined impact on urban thermal environment regulation.

5. Conclusions

This study focused on 74 urban parks in the central area of Jinan and systematically assessed their cooling effects using a comprehensive evaluation system based on both maximum impact indicators (PCA, PCE) and accumulative impact indicators (PCI, PCG). The findings revealed that urban parks consistently contribute significant cooling effects. About 62.3% of the parks were located within surface temperature cold spots, with the maximum temperature reduction reaching 7.77 °C. However, significant differences were observed among parks in terms of cooling intensity, range, and efficiency. The maximum impact was mainly influenced by the structural characteristics of the parks, while the accumulative impact was closely related to the surrounding built environment. This highlights the differentiated regulatory roles of internal and external factors. Based on this, parks were categorized according to their cooling characteristics to clarify functional differences, and targeted optimization directions were proposed. Drawing on these findings, urban planning should prioritize enhancing spatial ventilation and incorporating water features in large parks to improve overall cooling performance, while strategically allocating small, high-efficiency parks in densely built-up areas to create a well-distributed cool-island network and maximize cooling benefits. This study not only enriches the quantitative research basis of urban thermal environment regulation, but also provides scientific support for the optimization of functional zoning and green infrastructure layout of urban parks, which holds significant practical importance for strengthening urban resilience to climate change.