How to Improve Blue–Green–Gray Infrastructure to Optimize River Cooling Island Effect on Riparian Zone for Outdoor Activities in Summer

Abstract

As important urban green spaces, rivers enhance cooling island effects significantly by leveraging environmental factors. This study selected Suzhou River in Shanghai as the subject to explore how to improve blue–green–gray infrastructure to optimize the river cooling island effect on the riparian zone for outdoor activities in summer. A total of 77 samples, including 36 control groups and 41 experimental groups, were categorized into 12 types of blue–green–gray infrastructure composite features. ENVI-met was used to simulate summer thermal comfort, while redundancy analysis and boosted regression trees were used to identify significant factors and thresholds influencing the river’s cooling island effect. The results showed that for Suzhou River, the green–blue–green–gray–green composition most effectively optimizes the river cooling island effect. It is recommended to select construction sites where the river width is 55 m and the percentage of green infrastructure exceeds 40% and keep the distance between green infrastructure and the water body to within 3 m. Additionally, limiting gray infrastructure to less than 10%, with an average building height of 37 m and a building undulation of 25 m, is recommended to achieve the optimal cooling effect. This study finally proposes optimization strategies to maximize the cooling island effect of urban rivers, offering insights for the development of climate-adaptive urban riparian zones.

1. Introduction

The urban heat island (UHI) effect has intensified with global warming, leading to more frequent extreme heat events that threaten public health and restrict outdoor activities. Urban green spaces play a vital role in supporting residents’ recreational needs, contributing to physical and mental well-being, fostering social interaction and cultural exchange, and strengthening citizens’ sense of belonging and identity within cities [1]. However, rapid urbanization has led to a significant increase in the use of cement and asphalt for rooftops and roadways, materials that absorb and retain large amounts of solar radiation [2]. Simultaneously, dense clusters of high-rise buildings impede natural airflow, resulting in the accumulation of heat and pollutants in urban areas [3]. These factors collectively contribute to the UHI effect, undermining urban livability and posing serious challenges to public health and environmental sustainability [4]. According to the China Climate Change Blue Book (2024), the warming trend in China’s climate system is ongoing and exceeds the global average, making the country one of the most vulnerable regions to climate change worldwide [5,6].

To mitigate the adverse effects of urban heat islands (UHIs) on the environment and public health, cities and institutions across China have begun implementing various cooling strategies. Among these, the creation of urban cooling islands (UCIs) has emerged as one of the most effective approaches. UCIs are defined as areas within the urban core where temperatures are significantly lower than in the surrounding urban landscape. The concept originates from studies of desert oases and has been adopted as a strategic response to the intensifying UHI effect [7]. The urban cooling island effect refers to localized areas where surface or air temperatures are cooler than adjacent urban zones due to a combination of factors such as high vegetation coverage, the presence of large water bodies, low building density, and permeable surfaces [8]. Typical urban spaces exhibiting UCI effects include lakes, rivers, forests, greenbelts, parks, urban green spaces, and green roofs. These are integral components of blue–green–gray infrastructure (BGGI), which plays a critical role in reducing urban overheating and lowering ambient temperatures during summer months [9]. Green infrastructure primarily refers to urban vegetation that enhances outdoor thermal comfort through shading and increased evapotranspiration [10]. Numerous studies have qualitatively examined the relationships between vegetation characteristics—such as canopy cover, community structure, and vegetation density—and surface temperature or thermal comfort [11]. Blue infrastructure, encompassing rivers, lakes, reservoirs, ponds, coastal waters, and wetlands, functions as a cooling element due to water’s high specific heat capacity, which results in relatively small temperature increases under solar radiation [10]. As part of nature-based solutions (NBSs), current research on blue–green infrastructure (BGI) mainly focuses on its morphological impacts on the urban microclimate [12]. Gray infrastructure refers to the built environment, including roads, buildings, hardscapes, and functional gray urban elements such as parking lots [9]. Differences in albedo, heat capacity, and thermal conductivity among these elements result in varied impacts on outdoor thermal comfort. Existing studies on waterside gray infrastructure largely focus on community or building scale [13], while key morphological features—such as river valley aspect ratio and the continuity of waterfront building interfaces—remain underexplored. In this study, gray infrastructure includes roads, pavements, urban buildings, landscape structures, and waterside gray elements such as floodwalls, docks, and platforms.

As one of the most stable urban cooling spaces, water bodies play a critical role in mitigating the urban heat island (UHI) effect. While previous studies have often focused on individual environmental elements, a more comprehensive approach should consider the interactive effects of surrounding infrastructure on the cooling capacity of water bodies. Blue and green spaces—including urban vegetation and water bodies—serve as the core ecological components of the urban environment, enhancing landscape aesthetics, regulating microclimates, and improving outdoor thermal comfort [14]. Due to water’s high specific heat capacity, its temperature increases more slowly than land surfaces, making it a more effective thermal regulator compared to green spaces [15]. In high-density cities, water bodies are particularly valuable as they remain relatively thermally stable across seasons. The cooling effect of urban water bodies is influenced by various factors, such as their size, shape, spatial concentration, and landscape configuration [16]. For example, a study of three water bodies in Changsha found that larger surface areas were associated with broader cooling zones [17]. However, in densely built environments where land is scarce, expanding water surface areas is often unfeasible. Therefore, identifying other environmental variables that enhance the cooling performance of existing water bodies has become a central research focus. With advances in remote sensing technologies, many researchers now utilize land surface temperature (LST) data to analyze the spatial distribution, temporal dynamics, and intensity of cooling effects at large scales [18,19]. Studies report that urban water bodies can reduce temperatures across a radius ranging from 72.57 m to 465.42 m, with more pronounced cooling effects observed closer to the shoreline [20]. Thermal comfort, a key determinant of outdoor space usability, is also a critical metric for assessing microclimatic experience. Although many studies estimate cooling thresholds in waterside areas using LST inversion [21], few have examined thermal perception from a human-scale perspective. Recent research has highlighted the synergistic cooling effects arising from spatial connectivity between green spaces and water bodies, offering promising strategies for UHI mitigation [14]. Empirical measurements and simulations have shown that the width of riparian corridors and their degree of connectivity to water bodies significantly enhance the cooling performance of adjacent green spaces [22]. Additionally, pavement materials and underlying surface structures influence neighborhood-scale thermal conditions. In complex urban systems, the interactions among different infrastructure types—particularly between gray, green, and blue systems—generate composite effects that are increasingly recognized as critical [23]. While substantial progress has been made in understanding the individual and combined cooling functions of blue, green, and gray spaces, studies on the integrated cooling performance of blue–green–gray Infrastructure (BGGI) remain limited. Through spatial integration, structural synergy, and functional complementarity, BGGI offers a pathway for reconciling urban development with environmental sustainability [24]. Future research should prioritize the exploration of composite characteristics and collaborative mechanisms among the three systems to fully harness their synergistic potential. Developing multi-system integrated urban strategies will be essential for maximizing the cooling capacity of water bodies in dense urban settings.

In high-density urban areas, rivers are often the largest and most widely distributed natural water bodies within the urban core. They play an irreplaceable role in providing public waterfront activity spaces and enhancing urban ecological environments [25]. Empirical studies comparing static water bodies (e.g., lakes, wetlands) and dynamic ones (e.g., urban rivers) have shown that rivers, due to their continuous flow, exhibit more efficient heat exchange between water and air, which promotes sustained evaporation. Additionally, flowing water prevents localized overheating caused by solar radiation, maintaining a more stable and lower temperature, approximately 1 °C cooler than still water surfaces on average [26]. Compared to the patch-like morphology of lakes, the linear form of rivers facilitates the inflow of cooler rural air into urban interiors, effectively disrupting the continuity of urban heat islands [27]. Waterfront spaces connected to urban rivers are highly public and accessible, making them among the most active zones for outdoor activities. High-quality riverside environments contribute to a sense of belonging and well-being through four dimensions: physical health, mental well-being, healthy behavior, and social cohesion [28,29,30]. These spaces also support urban public health. In summer, urban rivers act as natural cooling agents by enhancing evaporation, which helps lower air temperature, increase humidity, and promote local air circulation—key processes for mitigating the urban heat island effect and regulating microclimates [31]. In recent years, numerous waterfront revitalization projects have aimed to repair the human–water relationship, restoring public vitality to riverfront areas and improving overall urban livability [32]. However, urbanization has gradually compressed waterfront activity zones in dense cities, shifting them from broad riverfront areas to narrower edges along the water, thereby reducing the functional range of outdoor use [33]. As a result, a key challenge for urban planners today is how to optimize and design the “transitional zone” that stretches from the riverbank to the outer boundary of the river’s cooling influence, especially under conditions of limited space and high urban demand.

Urban riparian zones represent the interface between water and land, where blue–green–gray infrastructure is highly concentrated [34]. This study aimed to scientifically improve the thermal environment of these areas in response to the pressing challenges posed by the urban heat island effect. The objective was to maximize the river cooling island effect while simultaneously creating thermally comfortable outdoor recreational spaces. To this end, the present study took the Suzhou River in Shanghai, China, as the research site. Thermal comfort indicators under extreme heat conditions were simulated using ENVI-met 5.5.1 based on data corresponding to periods of high outdoor activity. Redundancy analysis (RDA) and boosted regression trees (BRTs) were employed to quantify and characterize the key composite features of blue, green, and gray infrastructure. Based on a control group comparison, effective urban riparian design interventions were proposed, providing a scientific foundation for the development of climate-resilient riparian zones.

2. Materials and Methods

2.1. Research Objects

Shanghai is a quintessential high-density global city [35], through which the Suzhou River flows across its central urban area. The central section of the Suzhou River spans approximately 21 km, extending from the estuary at the Huangpu River in the east to the Outer Ring Road in the west. It passes through six districts: Huangpu, Hongkou, Jing’an, Putuo, Changning, and Jiading [36]. This river segment serves as a vital ecological corridor within Shanghai’s densely populated urban core (Figure 1). According to the Shanghai Master Plan (2017–2035), the central section of the Suzhou River and its riparian zone have been designated as a model area for enhancing livability in megacities. Riparian zones in urban settings perform multiple integrated functions: ecological, economic, aesthetic, recreational, and cultural [37]. Each segment of the river reflects different priorities and governance models [38]: the Putuo section focuses on ecological residential development, the Jing’an section highlights historical landmarks, while the Changning section emphasizes healthy communities and cultural identity. The riverbanks are densely developed, with the projected population along the Suzhou River expected to reach up to 4.2 million by 2035 [39], exemplifying the typical characteristics of urban rivers in high-density global cities. Numerous studies have shown that riparian green spaces adjacent to urban rivers function as critical ventilation corridors, facilitating the flow of water vapor and enhancing the river’s cooling and humidifying effects [40,41]. However, the presence of physical barriers such as buildings or walls significantly weakens these cooling benefits [42]. Based on these findings, this study focuses on the riparian zone of the Suzhou River—defined as the area between the first row of buildings on either side—and examines the characteristics of the blue–green–gray infrastructure within this space (Figure 2).

To ensure the accuracy and reliability of this study, river sections beneath elevated bridges were excluded from statistical analysis. These areas exhibit distinct thermal characteristics due to structural shading, which could interfere with accurately identifying the relationship between blue–green infrastructure and thermal comfort [43]. Moreover, as elevated bridges serve primarily as urban transport infrastructure, their structures are difficult to modify within riparian design and lack the attributes of outdoor public spaces; hence, they are beyond the scope of this study [44]. Based on the composite characteristics of blue–green–gray infrastructure, the riparian zone of the Suzhou River was divided into 55 river sections (Figure 3) and categorized into 12 composite types. To minimize experimental errors and avoid the influence of wind direction on the simulation results, sampling points were mainly selected from areas where the river channel aligns closely with the dominant wind directions, specifically southeast–northwest or east–west orientations. For each composite type, three river sections of 100 m length were chosen as study units, resulting in 36 sampling sites, as shown in Figure 4 and

Table 1. Overview of the 12 blue–green–gray infrastructure complex types and the 36 study sites.

Composite Feature Type	Research Sample	Composite Feature Type	Research Sample

.

2.2. Blue-Green-Gray Infrastructure Index

Building on existing research on urban riparian zones, this study developed a comprehensive indicator system to accurately characterize the integrated features of blue–green–gray infrastructure within these areas (

Table 2. Blue–green–gray infrastructure composite characteristic indicators.

Indicator Name		Meaning of the Indicator	Formula	Description of Calculations
Blue infrastructure	River cross-section width (BU1)	Width of water bodies in spatial units	BU1 = Ai/L	Ai is the area of the water body in square i (m2); L is the length of the water body in square i (m)
	Direction of flow (BU2)	A description of the plot orientation of the green space, reflecting in particular the influence of the orientation of the linear waterfront green space corridor on microclimatic effects	——	Category variables: 1 = Southeast; 2 = East; 3 = South; 4 = Northeast
	Percentage of blue infrastructure (BU3)	Percentage of blue infrastructure in cell area	BU3 = Ai/A	Ai is the total area of blue infrastructure in sample i (m2); A is the total area of sample i (m2)
Green infrastructure	Percentage of green infrastructure (GN1)	As a simple indicator of the complexity of the overall shape, the perim-to-area ratio is smallest when the shape is a circle; the longer the strip, the greater the perim-to-area ratio	GN1 = Ci/Si	Ci is the perim of the green infrastructure and Ai is the area of the green infrastructure (m2)
	Green infrastructure fragmentation (GN2)	The degree of fragmentation of the landscape, reflecting the complexity of the spatial structure of the landscape and, to some extent, the degree of human intervention in the landscape	GN2 = Ni/Ai	Ni is the number of green infrastructure patches (nos.) in sample i and Ai is the total area of green infrastructure in sample i (m2)
	Tree closure (GN3)	Ratio of tree cover to total green area	GN3 = Ai/A	Ai is the total area of tree cover in sample i (m2); A is the total area of green infrastructure in sample i (m2)
	Tree coverage (GN4)	Ratio of the area covered by trees to the total area of the unit	GN4 = Ai/A	Ai is the total area covered by trees in sample i (m2); A is the total area of sample i (m2)
Gray infrastructure	Percentage of gray infrastructure (GY1)	Percentage of gray infrastructure in cell area	GY1 = Ai/A	Ai is the total area of green infrastructure in sample i (m2); A is the total area of study sample i (m2)
	Impervious paving coverage (GY2)	Percentage of asphalt paved area	GY2 = Ai/A	Ai is the total area of impervious paving in sample square i (m2); A is the total area of study sample square i (m2)
	Building interface continuity (GY3)	Openness of streets; the higher the ratio, the tidier the streets	GY3 = B/L × 100%	B is the length of the building elevation line (m); L is the length of the building control line (m)
	Valley width-to-height ratio (GY4)	A param used to distinguish between wide river valleys and deep canyons	GY4 = 2 Vfw/[(Eld − Esc) + (Erd − Esc)]	Vfw is the width of the valley floor, Esc is the elevation value of the valley floor, and Eld and Erd are the left and right watersheds of the valley, respectively
	Average building height (GY5)	Average height of the first row of buildings in the river valley area	GY5 = (∑Ai + ∑Bx)/2	Ai is the height (m) of the first row of building i in the sample square; Bx is the height (m) of the first row of building x on the other side of the sample square
	Building undulation (GY6)	The difference between the tallest and lowest building patches, indicating the extent to which building patches within a given area differ in height	GY6 = (Amax − Amin) − (Bmax − Bmin)/2	Amax is the maximum height of the first row of buildings on one side of the sample (m); Amin is the minimum height of the first row of buildings on one side of the sample (m); Bmax is the maximum height of the first row of buildings on the other side of the sample (m); Bmin is the minimum height of the first row of buildings on the other side of the sample (m)
	Building spacing on both sides of the river (GY7)	Distance between buildings in the first row of the river valley area	——	——
	Building staggering (GY8)	Expressed as the ratio of the standard deviation of building patch heights to the average patch height; this can reflect the degree of variation in building landscape heights within a given range, i.e., the higher the value, the more pronounced the building height gradient	GY8 = √(1/n∑(Hi − H−))/H−	Hi is the height (m) of building i in the sample square; n is the sum of the number of buildings on both banks (n)
	Highest building index (GY9)	Expressed as the ratio of the height of the highest point in a building patch to the total height of the landscape, which is used to reflect the height characteristics and spatial congestion of the core building landscape within a given area	GY9 = Hmax/∑Hi	Hmax is the height of the tallest building in the sample (m); Hi is the height of building i in the sample (m)
Blue–green–gray infrastructure	Green infrastructure to water distance (BUGN1)	Vertical average distance from near-water green infrastructure to watershed	——	——
	Percentage of gray–green infrastructure (GYGN1)	Measures the total percentage of gray and green infrastructure in the spatial unit	GYGN1 = (ai + bi)/Ai	ai is the area of gray infrastructure in sample i (m2); bi is the area of green infrastructure in sample i (m2); Ai is the area of gray–green infrastructure in sample i (m2)

). The system comprised 18 indicators organized into four hierarchical levels, ensuring clarity and practical relevance. Previous studies have shown that wider river cross-sections are associated with stronger cooling effects [45]. However, river width and flow direction are constrained by urban planning and are generally unmodifiable. To identify suitable locations for outdoor activities along existing urban rivers, this study selected river width and the proportion of blue infrastructure as key indicators. The angle between corridor orientation and prevailing wind direction significantly affects local thermal environments [40]; thus, river flow direction was included as a critical variable. To examine how green infrastructure enhances cooling effects, metrics such as canopy coverage, tree coverage, and gray–green infrastructure proportion were incorporated. Research indicates that higher continuity and aggregation of riparian green spaces strengthen the synergy between water and vegetation, thereby improving cooling, humidification, and thermal comfort [46]. Consequently, green infrastructure’s perimeter–area ratio and fragmentation degree were also considered. Furthermore, connectivity between green spaces and water bodies enhances cooling effects, though the optimal distance remains unquantified [47]; thus, the spatial distance between green infrastructure and water bodies was included. Regarding gray infrastructure, besides its coverage ratio, permeable pavements have been found to behave similarly to vegetation in heat absorption and storage [2]. Hence, limiting impervious surface coverage is crucial for optimizing thermal comfort in riparian zones. Beyond these two-dimensional metrics, this study incorporated three-dimensional spatial indicators: building façade continuity, average building height, building height variation, riverbank building spacing, building stagger index, and maximum building height index [48]. Temperature differences between the free atmosphere on either side of the river and the river surface cause cool air to flow into low-lying river valleys, forming valley winds and wind corridors that lower local temperatures [49]. Therefore, the valley width-to-height ratio was also included as an additional indicator.

The required data on blue–green–gray infrastructure were sourced from the Geospatial Data Cloud platform. This study utilized Landsat 8 OLI-TIRS satellite imagery acquired on 29 July 2019, with a spatial resolution of 30 m, for interpretation. The data underwent preprocessing steps including radiometric calibration, atmospheric correction, geometric correction, and cloud detection/removal to ensure accuracy. Subsequently, micro-scale spatial features were refined using the satellite imagery (Figure 5). For the remaining unclassified areas, field surveys were conducted to supplement the data, employing GPS positioning, photographic documentation, and sample collection to complete the dataset for 36 plots.

2.3. Thermal Comfort and ENVI-Met

2.3.1. Thermal Comfort Index

Thermal comfort is a key indicator for studying the interaction between thermal environments and human perception. It is influenced by microclimatic conditions, with factors such as air temperature (AT), relative humidity (RH), and wind speed (WS) playing significant roles [50]. Commonly used indices for assessing outdoor thermal comfort include Standard Effective Temperature (SET), Predicted Mean Vote (PMV), and Physiological Equivalent Temperature (PET). Among these, PET better captures the effects of radiation, adapts to changing weather conditions, and more accurately reflects human thermal perception, making it a suitable choice for evaluating outdoor comfort in dynamic environments [51]. SET is less effective in handling uneven outdoor solar radiation, while PMV is more appropriate for relatively stable indoor environments [52]. Furthermore, PET is especially advantageous as it can be understood and applied even by those without a meteorological background [53]. Therefore, this study selected PET, alongside AT, RH, and WS, as the primary indicators for assessing thermal comfort.

$$\mathrm{PET}={\mathrm{T}}_{\mathrm{a}}+0.33\times \left(\mathrm{e}-0.70\times \mathrm{e}\right)+0.70\times \mathrm{ws}-400$$

(1)

Among them, T_a is the ambient air temperature, e is the water vapor pressure, and ws is the wind speed.

2.3.2. Simulation Settings

ENVI-met, a microclimate simulation software, is primarily designed to model interactions among surfaces, vegetation, and soil at the micro-to-meso urban scale [54]. Considering the correlations between landscape elements (such as green spaces, water bodies, and pavements) and meteorological factors (e.g., PET), ENVI-met version 5.5.1 was selected for this study. In the model, the thermal dynamics of water bodies are influenced by multiple factors including initial water temperature, surface albedo, emissivity, heat capacity, latent heat of evaporation, and local wind speed. ENVI-met employs a three-dimensional energy balance algorithm to accurately simulate shortwave and longwave radiation exchanges between water surfaces and surrounding land surfaces, enabling precise calibration. During summer heatwaves, urban riparian zones exhibit similar instantaneous activity fluctuations on both weekdays and weekends, but with pronounced temporal patterns: peak activity occurs between 18:00 and 20:00, followed by 07:00 to 09:00, while activity is lowest from 12:00 to 15:00, coinciding with the hottest period of the day [55]. Shanghai’s spring, autumn, and winter seasons feature mild climates suitable for outdoor activities, whereas summer presents significant thermal stress, posing a major challenge to urban thermal comfort. Summer daily maximum temperatures typically range between 35 °C and 38 °C, with approximately 50 days annually exceeding 35 °C. The hottest period is concentrated in July, with extreme temperatures occasionally surpassing 40 °C [56]. Accordingly, this study focused on the thermal characteristics and mitigation mechanisms during summer heat extremes. The selected representative extreme heat day was 29 July 2019 (maximum temperature 36.9 °C, minimum relative humidity 50%, maximum wind speed 1.7 m/s), used for ENVI-met calibration and scenario simulations. Average thermal comfort values during high-activity periods (07:00–09:00 and 18:00–20:00) in the coastal summer zone were chosen as dependent variables. To minimize simulation errors, observed meteorological data from a weather station were input, and local microclimate was field-calibrated via infrared thermometry, controlling errors within acceptable limits. Meteorological parameters were obtained from the Xujia Meteorological Station at an elevation of 4.6 m. The simulation area was located at latitude 31.2° and longitude 121.4333°, with spatial resolution set to dx = 2 m, dy = 2 m, and dz = 3 m. Considering the influence of building heights on airflow distribution, the vertical simulation domain was set to twice the maximum building height. Horizontally, the model domain was extended by one sample plot length beyond boundaries to reduce spatial autocorrelation and boundary effects, thus enhancing the robustness and representativeness of results. Based on observed data, the wind direction was set to 100.95°, the temperature ranged from 28.9 °C to 36.9 °C, and the relative humidity ranged from 50% to 82%. The simulation time covered 05:00 to 22:00.

2.3.3. Reliability Test

Factors such as the complex urban terrain, special architectural structures, and vegetation diversity can reduce the accuracy of ENVI-met model simulations. To improve the accuracy of the experimental simulations, this study adopted a pre-experiment approach by comparing the simulation results with observed data to validate the model’s reliability. The K1 site (the research point closest to the meteorological station) was selected for model reliability testing. The simulation results were then compared with the 1 km real-time meteorological grid data for 29 July 2019, provided by the National Meteorological Information Center, to verify the model’s accuracy. To assess the model’s precision, the Mean Absolute Percentage Error (MAPE) was calculated. Detailed observed and simulated data can be found in Appendix A. The MAPE values between observed and simulated data for temperature, relative humidity, and wind speed were 0.02%, 0.05%, and 0.19%, respectively. These results indicated that the model provided an excellent fit for these three indicators, demonstrating high accuracy and reliability.

$$MAPA={\displaystyle \frac{1}{n}}{\displaystyle \sum }_{i=0}^n{\displaystyle \frac{\left|y_i^{\prime }-y_i\right|}{y_i^{\prime }}}\times 100\%$$

(2)

The Physiological Equivalent Temperature (PET) was calculated using the Munich Energy-balance Model for Individuals (MEMI) [57] based on three key meteorological parameters: air temperature, water vapor pressure, and wind speed. Since PET is fundamentally derived from these rigorously validated input variables, no separate validation of PET values was conducted in this study. In order to assess the alignment between the simulated PET values and actual thermal perception, this study drew on existing thermal sensation studies conducted in Shanghai. These studies indicate that in summer outdoor environments, PET values ranging from 26 °C to 41 °C generally correspond to Mean Thermal Sensation Vote (MTSV) scores between +1 (“slightly warm”) and +3 (“very hot”) [58]. In this study, the PET values at the K1 site ranged from 29.8 °C to 40.7 °C, which falls within this subjective thermal comfort range, demonstrating that the simulated PET values were both realistic and consistent with human thermal perception.

2.4. Experimental Design

Due to the limitations of empirical methods, the number of study sites within the Suzhou River area was limited. To better understand the mechanisms by which composite blue–green–gray infrastructure influences the river cooling island effect and enhances summer thermal comfort, this study employed numerical simulations incorporating green infrastructure into the riparian zone. The effects of increasing green space area and improving green connectivity were quantified (Figure 6). Thirty-six scenarios (labeled A1 to L3) representing the current conditions of the central riparian zone of the Suzhou River were selected as the baseline control group. Previous studies indicate that increasing green space area is crucial for supporting biodiversity and providing ecosystem services, making it a key strategy for ecological restoration in waterfront areas. Accordingly, the baseline scenario was augmented by densely planting native representative trees—camphor trees (20 m height, 25 m canopy width, spaced 25–30 m apart)—to increase green cover. Existing research shows that different soil types and vegetation species significantly affect thermal comfort simulation outcomes [59]. Given the highly anthropogenically modified banks of Suzhou River, rebuilt under uniform design guidelines, soil and vegetation configurations are relatively consistent, mainly comprising hydromorphic and semi-hydromorphic soils with dominant green hedges and groundcover plants, exhibiting limited structural variation. To control variables and focus on the impact of the built environment on thermal conditions, the model standardized soil and vegetation types and simulated typical summer periods. Leaf area index variations were minimal during this time, thus exerting limited influence on simulation results. Pre-tests confirmed that these settings did not significantly affect spatial thermal distribution patterns, ensuring robustness and interpretability. While most simulation studies adopt green cover increments of 10%, 20%, and 30% [60,61], the high urban density along the Suzhou River banks constrains feasible increases. Through field surveys and preliminary simulations, 5% and 10% increments were determined as practical. A total of 41 experimental scenarios (labeled C1_05 to L3_05) were constructed to enhance the realism of the idealized experiments (

Table 3. List of control and experimental groups.

Control Group	Experimental Group
	5%	10%
A1	/	/
A2	/	/
A3	/	/
B1	/	/
B2	/	/
B3	/	/
C1	C1_05	C1_10
C2	C2_05	/
C3	C3_05	/
D1	D1_05	D1_10
D2	D2_05	/
D3	D3_05	/
E1	E1_05	/
E2	E2_05	/
E3	E3_05	/
F1	F1_05	/
F2	F2_05	/
F3	F3_05	F3_10
G1	G1_05	G1_10
G2	G2_05	/
G3	G3_05	G3_10
H1	H1_05	H1_10
H2	H2_05	H2_10
H3	H3_05	/
I1	I1_05	/
I2	I2_05	I2_10
I3	I3_05	/
J1	J1_05	/
J2	J2_05	/
J3	J3_05	/
K1	K1_05	K1_10
K2	K2_05	K2_10
K3	K3_05	/
L1	L1_05	L1_10
L2	L2_05	/
L3	L3_05	/

).

2.5. Analytical Models: RDA and BRT

Redundancy analysis (RDA) is a regression-based multivariate statistical method that integrates principal component analysis, extending traditional regression to handle multiple response variables simultaneously [62]. Widely used in ecology, RDA explores relationships between multiple environmental variables and species communities [63]. In this study, the dependent variables representing summer thermal comfort included Physiological Equivalent Temperature (PET), air temperature, relative humidity, and mean wind speed. Traditional bivariate correlation or linear regression approaches are inefficient for such multivariate relationships, whereas RDA effectively reveals the overall associations through linear modeling and dimensionality reduction [64]. Accordingly, Canoco 5.0 software was employed to analyze interactions between multiple climatic variables and composite blue–green–gray infrastructure features. Statistical significance was defined as p < 0.01 (highly significant), <0.05 (significant), and <0.1 (marginally significant). Vector length and the cosine of the angle between vectors indicated correlation strength, with longer vectors and smaller cosine angles reflecting the stronger explanatory power of independent variables on response variability.

The use of boosted regression trees (BRTs), combining regression trees with boosting algorithms, is a machine learning technique increasingly applied in urban microclimate research [65,66]. BRTs automatically capture nonlinear relationships and interactions without manual feature engineering [67], making the technique well-suited for complex multivariate datasets. Here, BRTs were used to assess the contribution and influence mechanisms of various variables on four thermal comfort indices, using composite blue–green–gray infrastructure features as predictors. To prevent overfitting, 10-fold cross-validation determined the optimal number of trees, alongside sensitivity analyses and parameter tuning. The final parameter settings were tree complexity = 5, learning rate = 0.01, bag fraction = 0.75, and Gaussian distribution assumption. This approach yielded relative contributions and marginal effect curves of blue–green–gray infrastructure features on thermal comfort indicators.

By integrating RDA for key variable selection and BRTs for threshold extraction, this combined analytical framework balances systematic analysis with practical flexibility, providing a robust foundation for diverse urban renewal applications.

3. Results

3.1. Thermal Comfort Simulation Results

To minimize the experimental simulation errors (see Appendix B), the thermal comfort indices of twelve composite blue–green–gray infrastructure types were analyzed and compared with the average values (see Figure 7). The error bars in the figure represent the standard errors, reflecting the fluctuation range of the mean thermal comfort values for each composite type. Physiological Equivalent Temperature (PET), which considers multiple climatic parameters and physiological factors, provides a comprehensive and accurate assessment of thermal comfort, effectively capturing the river cooling island effect [68]. Among the 77 study units, the simulated average PET was 34.76 °C, which corresponded to the “hot” (+2) level according to the MTSV thermal sensation vote. The average actual temperature (AT) was 31 °C. The green–blue–green–gray–green configuration (Category H) had a relatively low PET of 32.60 °C, while Category A had a lower AT of 31.34 °C. The AT of Category H (31.34 °C) was 0.34 °C higher than the minimum value of Category A but still 0.63 °C lower than the average. In contrast, the two subcategories of Category B had the highest values for both PET (39.48 °C) and AT (32.64 °C), suggesting that outdoor activities should be minimized in these configurations to avoid potential harm to the body. Regarding relative humidity (RH), previous studies indicate that the most comfortable humidity range for outdoor activities in summer is between 40% and 50% [69]. At this humidity level, people generally feel more comfortable. The relative humidity in the study area was 66.04%, with Category B having the lowest value of 62.27%. Both of these values exceeded the maximum limit of the optimal humidity range, making the outdoor environment less suitable for outdoor activities. The average wind speed (WS) was 0.61 m/s, with the gray–blue–gray composite type (Category B) having a wind speed of 0.69 m/s, which was slightly above the average. The higher wind speed contributed to a cooler and more comfortable sensation. However, because Category B had the highest values for both PET and AT, its overall performance was worse. In contrast, the green–blue–green–gray–green configuration (Category H) had a wind speed of 0.65 m/s, just 0.04 m/s lower than the maximum value, but still 0.04 m/s above the average. Therefore, Category H was considered the best configuration for achieving the river cooling island effect.

3.2. BGGI Results

Figure 8 illustrates the composite characteristics of blue–green–gray infrastructure across 77 study units. Overall, GN1, GN2, GN3, GY4, GY6, BUGN1, and GNGY1 exhibited a relatively high number of outliers, which largely depended on the construction environments of the respective sample areas. The fragmentation degree of green infrastructure was influenced by the land types on both sides of the riparian zones, with values ranging from a minimum of 0.0001 to a maximum of 0.0118, indicating significant feature variation with a difference value of 0.0117. GYGN1 performed better in capturing the interrelationships among infrastructures in urban riparian areas and thus best reflected their composite characteristics. However, it was also affected by the construction environment of the sample locations. Its feature variation was relatively high, with a difference value of 45.26, a maximum of 45.33, and a minimum as low as 0.07. Additionally, there were distinct differences among the feature values of BU3, GN4, GY2, GY5, GY8, and GY9. In contrast, BU1, BU2, GY1, GY4, and GY7 were influenced more by the river and its adjacent areas, resulting in relatively consistent construction conditions and, consequently, more concentrated data distributions.

3.3. Effect of BGGI on Thermal Comfort

3.3.1. Correlation Analysis: RDA

To ensure sufficient independence among the indicators for clear interpretation in subsequent analyses, this study evaluated the covariance among the 18 composite indicators of blue–green–gray infrastructure. The results show that all indicators passed the multicollinearity diagnostic, demonstrating good statistical independence (

Table 4. Covariance diagnostic results of blue–green–gray infrastructure composite characteristic indicators for 18 urban riparian zones.

Configuration Indicator		VIF Value	Tolerance Value
Blue infrastructure	BU1	1.917	0.522
	BU2	2.081	0.481
	BU3	9.997	0.183
Green infrastructure	GN1	4.191	0.239
	GN2	9.018	0.111
	GN3	9.116	0.11
	GN4	9.707	0.173
Gray infrastructure	GY1	8.925	0.112
	GY2	4.052	0.247
	GY3	1.916	0.522
	GY5	4.743	0.211
	GY6	2.483	0.403
	GY7	7.01	0.143
	GY8	2.255	0.443
	GY9	2.064	0.484
	GY4	4.195	0.238
Blue-green-gray infrastructure	BUGN1	1.529	0.654
	GNGY1	1.723	0.58

).

To explore the complex interactions between various indicators, this study employed redundancy analysis (RDA) to sequentially depict the relationships between thermal comfort indicators and the composite features of blue–green–gray infrastructure [70]. The results of the conditional effect analysis (see

Table 5. Statistical table of RDA conditional effects results.

Name	Explains	F	P
GY1	34.8	40	0.002
BU2	8.9	11.6	0.002
BU3	2.9	4.6	0.014
BUGN1	2.8	3.9	0.016
GY5	2.5	3.5	0.024
GY6	1.9	2.7	0.058
BU1	1.7	2.5	0.07
GY2	1.4	2.1	0.112
GY7	1.1	1.7	0.12
GY4	1	1.7	0.148
GNGY1	0.7	1.1	0.24
GY3	0.7	1.1	0.304
GN4	0.7	1.1	0.352
GN1	0.5	0.7	0.434
GY8	0.4	0.6	0.572
GN3	0.4	0.7	0.388
GY9	0.4	0.6	0.566
GN2	<0.1	<0.1	0.982

Note: Highly significant correlation factor (p < 0.01): dark gray background; significant correlation factor (p < 0.05): gray background; marginally significant correlation factor (p < 0.1): light gray background; other factors (p > 0.1): white background.

) show that the explanatory power of GY1 was 34.8%, which was highly significant (p < 0.01), indicating that the proportion of gray infrastructure is the primary environmental factor affecting summer thermal comfort. The explanatory power of flow direction (BU2) was 8.9%, also reaching a highly significant level (p < 0.01), making it another key influencing factor. Additionally, the explanatory power of BU3, BUGN1, and GY5 was 1.7%, all of which were statistically significant (p < 0.05), suggesting that these variables, while having weak explanatory power, still show a reliable correlation with thermal comfort. Furthermore, GY6 and BU1 had explanatory powers of 1.9% and 1.7%, respectively, with significance at a marginal level (p < 0.1), indicating that they may exhibit some influence trends, though the statistical support is weaker, and, thus, caution is needed in interpretation. The explanatory powers of the remaining variables ranged from 0.1% to 1.4%, and since they did not pass the significance test (p > 0.05), their explanatory effect on summer thermal comfort was statistically insignificant. To enhance the rigor and interpretability of the results, conventional significance standards were followed: p < 0.01 indicated high significance, p < 0.05 indicated significance, and p < 0.1 indicated marginal significance, with interpretation requiring a comprehensive judgment based on the actual context.

The ranking results of the composite features of blue–green–gray infrastructure and summer thermal comfort are shown in Figure 9, where longer ray lengths indicate higher explanatory power [71]. Specifically, the results show that GY1, BU2, BU3, and BUGN1 formed acute angles with PET and AT, indicating positive correlations between them. This means that a higher proportion of gray infrastructure, a flow direction closer to the northeast, a higher proportion of blue infrastructure, and greater distances between green infrastructure and water bodies result in higher air temperature (AT) values. Meanwhile, GY5, GY6, and BU1 formed obtuse angles with PET and AT, indicating negative correlations, meaning that higher average building height, greater building fluctuation, and wider river widths result in lower PET and AT values, making people feel more comfortable. Additionally, GY5, GY6, and BU1 formed acute angles with relative humidity (RH), indicating positive correlations, meaning that higher average building height, greater building fluctuation, and wider river widths result in higher RH values. Conversely, GY1, BU2, BU3, and BUGN1 formed obtuse angles with RH, indicating negative correlations. Furthermore, GY1, BU3, and GY5 formed acute angles with wind speed (WS), indicating positive correlations, meaning that a higher proportion of gray infrastructure, a greater proportion of blue infrastructure, a higher average building height, and better valley conditions result in higher wind speed (WS) values. In contrast, BU2, BUGN1, GY6, and BU1 formed obtuse angles with wind speed (WS), indicating negative correlations. Additionally, the angle relationships between the four dependent variables in the RDA diagram further suggest potential synergistic and trade-off mechanisms between them. The angle between PET and AT was small, showing a clear synergistic relationship. In contrast, relative humidity (RH) formed significant obtuse angles with both PET and AT, displaying a certain antagonistic relationship; wind speed (WS) formed near right angles with the other three dependent variables, suggesting its relatively independent role, potentially acting as one of the key factors regulating the thermal environment. This spatial distribution trend helps explain the interactions of different environmental factors in the complex regulation of thermal comfort.

3.3.2. Contribution and Threshold Analysis: BRT

To explore the mechanisms and strengths of the significant factors influencing the summer thermal comfort index, this study replaced seven significant indicators of blue–green–gray infrastructure composite features in the BRT model for analysis, with the contribution measures as follows: For Physiological Equivalent Temperature (PET) (see Figure 10a), the contribution order was GY1 > BU3 > GY5 > GY6 > BU1 > BU2 > BUGN1, where GY1 (33.24%) had the largest relative contribution and was the most important variable affecting the PET. For air temperature (see Figure 10b), the contribution order was GY1 > BU3 > BU1 > GY5 > GY6 > BU2 > BUGN1, with GY1 (36.51%) again being the largest relative contributor. For relative humidity (RH) (see Figure 10c), the contribution order was GY1 > BU3 > BUGN1 > BU1 > GY5 > GY6 > BU2, with GY1 (28.09%) and BU3 (20.84%) being the largest relative contributors. For average wind speed (WS) (see Figure 10d), the contribution order was GY5 > BU3 > BU2 > GY1 > BU1 > GY6 > BUGN1, with GY5 (25.62%) having the largest relative contribution, followed by BU3 (20.66%). It is worth noting that in the RDA results, GY1 (the proportion of gray infrastructure) played a dominant role in the comprehensive ranking of multiple thermal environment indicators as a structural factor. However, in the BRT model, the results for wind speed as a single dependent variable showed that GY5 (average building height) had the largest contribution. This difference mainly arose from the different mechanisms of the two analytical methods. RDA emphasizes the overall explanatory power of covariates under multiple dependent variables, while BRT focuses more on the response strength between a specific indicator and a single explanatory variable. In wind speed regulation, moderate building height could create a ventilation corridor effect in the riparian zone, effectively guiding and enhancing valley wind speed. As a result, GY5 became more prominent in the BRT model.

Using the boosted regression tree (BRT) model, this study further explored the marginal effects of key variables on summer thermal comfort. By identifying inflection points and response patterns from the marginal effect curves (Figure 11), optimal thresholds were determined for seven critical indicators during periods of high human activity. The results indicate that the optimal range for GY1 (proportion of gray infrastructure) was 0.10–0.22, with 0.10 being ideal for enhancing thermal comfort. BU3 (proportion of blue infrastructure) performed best at around 0.40, corresponding to the strongest cooling effect. For GY5 (average building height), the effective range was 20–37 m, with 37 m being optimal. GY6 (building height variation) showed the best performance at around 25 m. The optimal river width (BU1) was approximately 55 m. For BUGN1 (distance between green infrastructure and water bodies), the most effective range was 3–6 m, with 3 m yielding the strongest synergistic cooling effect. These identified thresholds provide empirical evidence to guide thermal environment optimization and infrastructure planning in urban riparian areas.

4. Discussion

4.1. Optimal Configuration

Urban spatial patterns encompass both artificial and natural systems. Land use types reflect the nature and distribution of land and form the basis of urban landscape design [72]. As rivers serve as key natural cooling islands in cities, their shape and orientation are typically constrained by urban planning and construction, making them difficult to alter. Therefore, optimizing land use and infrastructure along riverbanks is one of the most effective strategies for enhancing the river cooling island effect.

Physiological Equivalent Temperature (PET) is one of the most direct indicators of outdoor thermal comfort [73]. In this study, actual air temperature (AT), relative humidity (RH), and wind speed (WS) were also considered to more comprehensively explore optimal configurations for achieving cooling effects. Among the spatial layouts examined, the green–blue–green–gray–green (G-B-G-Gr-G) pattern performed best across multiple thermal indicators (see Figure 12), making it the preferred configuration for enhancing thermal comfort along urban rivers in summer. This pattern features a diversified landscape structure that maintains thermal balance through the synergy of vegetation, water bodies, and built elements. Vegetation cools the environment via evapotranspiration; water bodies, due to their high specific heat, absorb excess heat during the day and release it slowly at night; and the first row of riverside buildings helps shade hard surfaces and vertical façades. Together, these elements form an effective microclimate regulation system. Following the G-B-G-Gr-G type, the green–blue–green (G-B-G) and gray–green–blue–green (Gr-G-B-G) patterns also showed strong cooling performance. Generally, the proximity between green and blue infrastructure enhances cooling due to combined effects such as shading, evapotranspiration, airflow, and thermal storage [74,75]. Compared to G-B-G, the Gr-G-B-G-Gr configuration performed better in terms of thermal comfort, likely because low-rise gray infrastructure near the waterfront—such as roads and plazas—offers less surface roughness and better ventilation [76,77]. While higher wind speeds can improve outdoor comfort, evapotranspiration remains essential for lowering surface temperatures. However, overly dense vegetation may increase humidity and reduce comfort in hot conditions, indicating a trade-off between ventilation and evaporative cooling. To balance these factors, this study recommends a layered planting strategy, maintaining airflow corridors with open understory vegetation while placing high-evapotranspiration trees in lateral zones to enhance cooling without obstructing ventilation. Although the Gr-G-B-G-Gr pattern performed similarly to the G-B-G-Gr-G type, its green–gray synergy was slightly less effective. Prior research shows that gray infrastructure such as asphalt and concrete tends to store and re-radiate heat due to low albedo, exacerbating urban heat island effects [78]. However, when green and gray spaces are interwoven, vegetation can partially offset the heat storage of gray materials [79], effectively reducing local temperatures and improving comfort. Overall, the G-B-G-Gr-G layout offers the most balanced and efficient cooling configuration.

Previous studies on lakes, rivers, and parks show that buffer zones around water bodies exhibit increasing cooling effects with distance [80]. Using numerical simulation and remote sensing, this study confirms that closely integrated blue–green systems maximize cooling benefits [47]. Vegetation—particularly trees—plays a dominant role in temperature reduction via shading and evapotranspiration [81]. Notably, the four configurations with the highest PET values were gray–blue–gray (Gr-B-Gr), gray–blue–green–gray (Gr-B-G-Gr), gray–blue–gray–green (Gr-B-Gr-G), and gray–blue–green–gray–green (Gr-B-G-Gr-G) (see Figure 12). These layouts showed strong cooling potential when green infrastructure flanked both sides of water bodies, enhancing the ecological effect of river-based cooling corridors. Although this study focused on urban rivers rather than lakes or wetlands [82], its findings align with existing research, confirming that when considered independently, green infrastructure generally outperforms blue infrastructure in mitigating urban heat [29].

4.2. Optimal Location to Improve Thermal Comfort in Summer

In response to the challenges posed by global climate change, urban blue infrastructure has become a vital component of Nature-Based Solutions (NBSs) for mitigating the Urban Heat Island (UHI) effect. Originally implemented in water-sensitive urban design, blue infrastructure contributes to cooling through evapotranspiration, helping to reduce urban temperatures and enhance human thermal comfort [15]. However, rapid urbanization and land constraints in high-density cities have made it increasingly difficult to modify the width or alignment of existing rivers [83]. Moreover, the expansion of water bodies is often restricted by land use regulations. Despite these challenges, landscape designers can utilize thermal comfort modeling to identify optimal activity zones, creating more comfortable and accessible waterfront spaces. As shown in Figure 13, urban rivers not only regulate temperature and humidity but also enhance wind flow at confluences, thereby improving their microclimatic performance and functioning as crucial urban ventilation corridors [84]. In compact urban settings, excessive building scale, high density, surface roughness, or enclosed spatial forms can hinder airflow. Urban ventilation corridors typically include green corridors, road corridors, water corridors, and low-density urban areas [40]. Studies have shown that the angle between prevailing wind direction and river orientation significantly affects the efficiency of ventilation corridors. For optimal airflow, this angle should be minimized—main corridors should align within 30° of the prevailing wind and secondary corridors within 45° [85,86,87]. Previous research has demonstrated that rivers wider than 30 m exhibit significantly stronger cooling effects on land surface temperature (LST) [88,89]. Building on this, the present study recommends an average river width of 55 m and a blue infrastructure ratio of approximately 40% to achieve optimal thermal regulation. While these findings align with existing literature emphasizing the importance of sufficient river width for cooling, this study differs by focusing on a typical mid-sized urban river—Suzhou River—using high-resolution simulations and field data to propose precise, transferable design parameters. Suzhou River maintains a consistent width of 42–71 m and lacks narrow segments, making it ideal for defining applicable thresholds. These findings are particularly relevant for rivers of similar scale and spatial continuity. Furthermore, wider river sections not only improve local cooling but also facilitate the inflow of cool air from urban peripheries into central areas, enhancing both thermal comfort and ventilation efficiency [13]. A key contribution of this study lies in the systematic quantification of climate regulation in relation to river spatial dimensions, offering scalable and actionable guidelines for medium-scale waterfront design and climate-adaptive planning. However, this study found no significant correlation between canopy depression, tree cover, and summer thermal comfort in riparian zones. This may be attributed to limitations in study area size or data resolution. It is also possible that overly dense vegetation—particularly tall trees—can obstruct longwave radiation and upper-level air convection [90]. Therefore, tree planting density should be carefully managed to promote ventilation and improve thermal comfort.

4.3. Optimization Recommendations of Shoreline Green Infrastructure

During hot seasons, green infrastructure reduces ambient temperatures through evapotranspiration, shading, and the interception of solar radiation by plant canopies [81]. As key ecological components in cities, blue–green infrastructure represents a typical Nature-Based Solution (NBS) for addressing climate change [91], playing a vital role in enhancing outdoor thermal comfort. While blue strategies often outperform green strategies when applied independently, their combination yields complementary environmental benefits, including synergistic cooling effects and enhanced ecosystem services [92]. Studies have shown that the cooling influence of rivers on adjacent green spaces diminishes with distance from the water’s edge [93]. Based on this, the present study defined an optimal distance between green infrastructure and the riverbank within the riparian zone. In line with previous findings (Figure 14), this study identified 3 m as the optimal minimum distance, which maximizes the clustering effect of small and micro-scale blue–green patches and significantly enhances thermal comfort in riparian areas. This 3 m threshold applies to medium-sized urban rivers with widths between 40 and 60 m, such as Suzhou River, and offers concrete design guidance for waterfront spatial planning. Our findings align with earlier studies: for rivers 30–50 m wide, optimal cooling occurs when green spaces are located within 20–50 m of the riverbank; for water bodies over 100 m wide, a significant synergistic cooling effect can be observed when the distance between water and greenery is less than 250 m [88,94]. These studies collectively suggest the existence of an optimal range for blue–green spatial proximity. However, unlike previous research that focused primarily on large rivers or regional scales, this study narrowed its focus to medium-width rivers and quantified the minimum effective blue–green spacing. This fills a key gap in fine-grained parameterization for urban-scale waterfront renewal and provides actionable and transferable design references for riparian zones in cities. In addition, prior studies established a correlation between surface temperature and vegetation cover, indicating that areas with higher vegetation density tend to be cooler [95]. Chinese scholars also analyzed 20 samples of tree, shrub, and herbaceous communities in Fuzhou’s Minjiang Park, finding a negative correlation between vegetation biomass and physiological equivalent temperature (PET) [96]. However, this study did not find a significant relationship between canopy structure or tree coverage and summer thermal comfort in riparian areas. This may be due to the limited spatial scope and resolution of the data. For instance, the ENVI-met simulation in this study used only one tree species (Cinnamomum camphora), whereas, in reality, a variety of tree species collectively influence microclimatic effects [97]. Differences in leaf reflectance, leaf area index (LAI), and transpiration rates across species can significantly affect microclimate variables such as temperature and humidity [98], which, in turn, influence outdoor thermal comfort. These aspects will be further explored in future studies. Moreover, while dense vegetation—especially tall trees—provides many benefits, careful planning is required. Overly dense tree planting may hinder longwave radiation release and upward air convection [99]. Therefore, managing tree density is essential to improve wind flow and enhance thermal comfort.

4.4. Optimization Recommendations of Shoreline Gray Infrastructure

Compared to other infrastructure types, gray infrastructure—such as roofs, pavements, walls, and other built surfaces—tends to accumulate reflected solar heat, exhibiting significant heat-retention characteristics [100]. As shown in Figure 15, reducing the proportion of gray infrastructure to approximately 10% is effective in improving thermal comfort. However, maintaining such a low proportion in high-density urban environments, particularly in waterfront zones dominated by commercial or transport functions, is challenging. In this study, only two residential waterfront areas achieved this target by minimizing unnecessary paving, implementing simplified pathways, and limiting site fixtures. For areas where gray infrastructure cannot be significantly reduced, thermal comfort can be improved by planting large-canopy trees and introducing high-shade species [101]. Research indicates that broad-canopy trees can lower the Physiological Equivalent Temperature (PET) by 2–4 °C, enhancing thermal adaptability during outdoor summer activities [102]. Additionally, a network of tree shade helps regulate microclimate distribution and mitigates heat accumulation along pedestrian paths and water edges. Therefore, even when gray surface coverage cannot be drastically reduced, optimizing shading strategies remains essential for achieving thermal comfort. Some studies have found that increasing building height around lakes diminishes their cooling effects, both horizontally and vertically [103]. Other research has highlighted the importance of vertical factors affecting the cooling performance of ground materials [104]. Specifically, stronger ventilation above surfaces enhances downstream cooling, and reducing surrounding building heights improves heat dissipation. The layout and form of first-row buildings along riverbanks significantly influence outdoor thermal comfort. Our findings suggest that when the first row of buildings is kept between 20 and 37 m in height, they provide sufficient shading without significantly obstructing local air circulation between land and water. This aligns with previous studies, which note that building height is crucial to maintaining urban ventilation corridors [105]: buildings that are too low cannot capture upper-level airflow, while excessively tall structures increase surface roughness, reduce ventilation efficiency, and may create airflow bottlenecks [106]. Through wind field simulations, this study further refines these insights, showing that building heights close to 37 m can optimize airflow guidance within river valley spaces, thereby enhancing microclimate regulation. Unlike prior research that typically focuses on theoretical or macro-scale relationships between building height and ventilation, our results provide actionable design parameters for waterfront development, supporting both ecological thermal comfort and urban ventilation goals.

In existing riparian zones, constraints such as the waterfront skyline and natural river channel width make it difficult to modify overall building height or spacing. However, certain spatial elements remain adjustable to improve summer thermal conditions [107]. Specifically, lowering the height of continuous buildings and maintaining an average first-row elevation around 25 m can increase visual openness and boost average wind speed. This, in turn, enhances the river’s cooling effect and improves overall thermal comfort during hot weather.

4.5. Discussion on Design Application

In high-density urban environments, riparian zones are often narrow and subject to stringent anthropogenic constraints, making large-scale interventions particularly challenging [98]. Therefore, priority in optimizing design for these areas must be given to configuration selection and proportion control based on specific contextual demands. When spatial scale is flexible or linear features are prominent, configuration design should be prioritized. This study proposes an optimal green–blue–green–gray–green configuration that alternates blue, green, and gray infrastructures to maximize the river cooling island effect. When spatial layout is constrained or precise thermal comfort regulation is needed, proportion control should take precedence. For undeveloped areas with undefined surrounding land use planning, a combined approach is recommended. By applying the green–blue–green–gray–green configuration, positioning activity zones within a 30° angle relative to the river direction and prevailing wind direction can reduce the Physiological Equivalent Temperature (PET) by 0.7 °C. Furthermore, a 55 m cross-section design, with blue infrastructure occupying 40%, can lower the PET by 1.5 °C. Placing green spaces within 3 m of blue infrastructure enhances synergistic evapotranspiration and integrated shading-reflectance effects [108], effectively reducing the PET by 0.6 °C. Crucially, controlling the gray infrastructure proportion below 10% or employing functionally composite and material-optimized gray–green overlay designs disrupts continuous heat surfaces and reduces the PET by approximately 3 °C. For the surrounding environment, an average first-row building height of 37 m ensures sufficient shading without excessively obstructing sunlight to water bodies and vegetation. Simultaneously, building layouts can guide airflow, lowering the PET by about 2.2 °C. Maintaining building height variations around 25 m maximizes ventilation efficiency and further decreases the PET by roughly 1.6 °C. Coupled with blue–green synergistic effects, cold air transport efficiency can increase by 30%. Shanghai serves as a typical example of a high-density city, with its central districts predominantly composed of urban built-up land and characterized by high population and building densities [109]. In contrast, low-density cities or suburbs contain more ecological land, where green infrastructure plays a more significant role. Therefore, the blue–green–gray infrastructure optimization thresholds proposed here generally apply to urban spatial planning with floor area ratios (FARs) between 2.0 and 4.0.

It should be noted that the thermal comfort thresholds developed in this study were based on Shanghai’s typical monsoon climate, and their applicability may be significantly limited in other climatic zones. In tropical regions, strong solar radiation and high background temperatures mean that taller building facades and dense street interfaces facilitate continuous shading, substantially lowering surface temperatures and reducing water evaporation rates. Thus, optimal design thresholds in these areas may favor higher building densities and shading ratios to minimize solar radiation load. Conversely, in arid and water-scarce regions, limited water availability reduces the contribution of water body evaporation to microclimate regulation and raises the cost of evaporative loss. Blue–green infrastructure effectiveness significantly declines, and thermal comfort optimization relies more on surface material albedo, vegetation coverage, and evapotranspiration efficiency. Building and block layouts in these regions typically adopt a “high shading–low evaporation loss” composite regulation strategy, balancing shading and moisture retention. In cold climate zones during winter, enhancing heat accumulation is the primary goal. Compared with high-density shading configurations, lower building heights and open street valleys improve solar radiation penetration and surface heat storage, thereby increasing winter thermal comfort. Threshold sensitivities in these regions focus more on solar energy harvesting efficiency, block orientation, and wind chill mitigation, emphasizing openness and sunlight accessibility. Additionally, in cold, dry, and low-precipitation boreal areas, reduced evaporative potential further diminishes blue–green infrastructure’s role in microclimate regulation, necessitating systematic cross-climate validation of adaptability and effectiveness [110]. Particularly in winter, river cooling mechanisms differ from the evaporation- and convection-dominated pathways of summer, relying more on sensible heat exchange between water and air [111]. Previous studies indicate a significant positive correlation between river width and spatial scale with surface temperature in winter, contrasting markedly with summer findings [112]. Winter thermal comfort prioritizes heat retention rather than heat reduction. Consequently, riparian spatial design strategies should be seasonally adjusted—for example, by increasing riparian vegetation density, installing adjustable windbreaks, and optimizing building orientation to create sunlit retention zones—to enhance winter thermal comfort. In summary, future research should adopt a climate-responsive urban design perspective, integrating key parameters such as radiation load, water resource availability, diurnal temperature variation, and wind environment across different climate zones. The aim is to establish a climate-adaptive spatial optimization framework that maximizes thermal environment regulation under diverse climatic conditions.

4.6. Limitations of This Study and Suggested Future Research

This study has several limitations, and future research directions deserve further exploration. First, this study used data from a representative extreme hot day in Shanghai on 29 July 2019 (maximum air temperature ATmax = 36.9 °C, minimum relative humidity RHmin = 50%, maximum wind speed WSmax = 1.7 m/s) for model validation and thermal scenario simulations, effectively validating the river cooling effect in summer. However, due to the weakening of solar radiation and lower environmental temperatures in winter, the river cooling island effect is weaker in winter, and the cooling efficiency may be affected due to the reduced temperature difference between water and air [113]. Therefore, based on a single summer weather sample, this study may have overestimated the intensity of the river cooling island effect at the annual scale, with certain seasonal applicability limitations. Future studies should incorporate meteorological data from multiple seasons and construct dynamic time-series models to simulate and analyze the annual fluctuations of the cooling island effect. Secondly, this study focused on a typical area along the Suzhou River in Shanghai, characterized by high urban density and compact building layouts, within a hot-summer and cold-winter cold climate zone. Therefore, the results of this study may not be directly generalizable to other climatic conditions or urban forms. Future research could involve comparative studies in typical tropical and cold climate zones, using remote sensing thermal measurements, meteorological observations, and field surveys to identify the climate-dependent and topographic response characteristics of river cooling island effects, thereby enhancing the generalizability of the findings. Third, due to the limitations in landscape pattern scale, this study primarily used ENVI-met for microclimate simulations at the block scale and did not cover macro elements such as wind corridors, hydrological paths, and activity flow lines within the urban space. Future research is recommended to incorporate multi-scale modeling approaches, continue using high-precision simulation tools (such as ENVI-met) at the micro scale (<500 m), and employ CFD and urban-scale microclimate simulations (such as PALM, UrbClim) at the meso scale (>1 km) for combined simulations. This would help assess the cooling island linkage mechanisms of blue–green–gray infrastructure at different scales and their coupling regulation effects on human activity patterns. Additionally, future studies will integrate perspectives from landscape design, urban sociology, transportation studies, and other interdisciplinary fields, while also considering social, economic, and cultural factors, to further explore how to optimize river cooling island effects. Such research will aid urban planners and designers in better understanding citizen experiences when designing outdoor spaces. Ultimately, these studies will contribute to developing more comprehensive and effective strategies to create comfortable and attractive riparian environments, thereby improving the overall quality of urban life.

5. Conclusions

This study presented several novel findings. First, the green–blue–green–gray–green composite feature type is the most conducive to creating river cooling islands to improve summer thermal comfort. This supports the notion of a synergistic effect between urban green spaces and water bodies in terms of cooling. Additionally, the study demonstrated that extending the river cooling island to enhance thermal comfort in the riparian zone is more effective when green spaces are adjacent to water bodies. Furthermore, our results highlight the importance of minimizing the presence of gray infrastructure near urban rivers. Activity areas should be strategically located at an angle of 30° between the river’s direction and prevailing winds (southeast winds), with an average river cross-section width of 55 m and the optimal proportion of blue infrastructure at 40%. Crucially, the distance between green infrastructure and water bodies should be reduced to within 3 m, as this enhances the aggregation effect of blue–green spaces and maximizes the synergistic effect of small and micro blue–green patches. Moreover, the proportion of gray infrastructure should be kept at around 10%. Finally, the average height of the first-row buildings on both sides of the river should be maintained between 20 and 37 m, with an overall building height fluctuation of approximately 25 m being optimal.

This study also makes several innovative contributions. First, this study integrated the configuration of green space infrastructure in the riparian zone by increasing the area of green space and enhancing connectivity between green spaces. Then, the potential design interventions’ impacts on thermal comfort in the riparian zone were evaluated through a control group experiment. Additionally, unlike previous studies on extreme heat that typically focus only on daytime periods or peak temperature periods [112], this study, based on the principle of maximizing landscape service efficiency, chose the most active period in riparian zones and analyzed the average value for that period. This approach provides a more comprehensive and valuable reference for the development of climate-adaptive urban riparian zones. Furthermore, using Shanghai’s Suzhou River as the study subject, this study employed redundancy analysis (RDA) and boosted regression trees (BRTs), utilizing their unique advantages in handling complex multivariate datasets [64,67]. This methodology revealed the interrelationships between multiple variables and identified key factors affecting thermal comfort and their threshold mechanisms. Based on these findings, this study proposes optimal configurations and recommendations, offering valuable insights for the construction of climate-adaptive riparian zones in southeastern coastal areas and high-density urban environments. Simultaneously, this study provides a solid scientific foundation for urban infrastructure planning practices.