The Cooling Effects of Landscape Configurations of Green–Blue Spaces in Urban Waterfront Community

Abstract

Optimizing the configuration of green–blue spaces is crucial in mitigating the urban heat island effect. However, many existing studies neglect to consider the synergistic cooling effect of green–blue space and its spatial comparison, focusing instead on individual ecological elements of green space or water bodies. Additionally, the relative importance of different configuration indicators and their marginal effects on the cooling effect of green–blue space remain unclear, with an identified need for the quantification of indicator thresholds for maximizing the cooling effect. To address these gaps, this study investigated green–blue spaces in 30 urban waterfront communities located in Kunshan City, Jiangsu Province, China, and measured the scale, distribution, morphology, green–blue relationship, and built environment of these spaces. To determine the cooling effect, maximum air temperature and mean cold island intensity were measured using ENVI-met simulations. Correlation analyses and boosted regression trees (BRT) were utilized to identify the configuration indicators that affect the cooling effect and their action threshold. The results show that green space distribution and water body shape are the most important features affecting the maximum air temperature, with green space patch density (PD) and water landscape shape index (LSI) contributing 21.3% and 20.9% to the reduction in temperature, while the thresholds are 550 and 4.2, respectively. The contribution of green–blue space percentage is critical in raising urban cold island intensity, with threshold effects at 43%. These findings provide practical guidance for the efficient exploitation of the synergistic cooling effects of green–blue space and enhancement of climate resilience in coastal cities.

1. Introduction

According to the “World Cities Report 2022: Envisaging the Future of Cities” [1] released by UN-Habitat in June 2022, the urbanization rate will increase from 56% in 2021 to 68%, and the global urban population will expand by 2.2 billion people by 2050. In such a continuous and rapid urbanization process, physical landscapes, such as water bodies and vegetation, are replaced with a large number of artificial impervious surfaces with low albedo, while the urban heat island (UHI) effect is becoming an ever-more serious environmental problem in cities all over the world [2,3,4,5,6,7]. The UHI effect is a phenomenon in which the temperature in urban areas is higher than that in the surrounding countryside due to local heat accumulation [8,9]. The range and intensity of extremely high temperatures in cities are increasing. These trends not only impair thermal comfort and cause respiratory diseases in the population, but also lead to deterioration of air quality, energy loss and other urban problems [10,11]. It is urgent to alleviate the UHI effect, and research on the cooling effects of green–blue spaces is growing [12].

Urban green–blue landscape as a term refers to a combination of two types of landscape spaces, namely blue space and green space. It mainly includes parks, roadside green spaces, residential green spaces, rivers, lakes, reservoirs, ponds, and other artificial, semi-artificial, or natural spaces in cities [13]. The presence of such green–blue landscape elements provides a significant cooling effect on the environment in urban areas [11,14], primarily through shading and evaporation by vegetation [15,16]. Additionally, water bodies with high specific heat capacity and evaporation capabilities contribute to the cooling effect through reducing the air temperature in their vicinity [10,17]. Optimizing the configuration of the green–blue landscape is more ecologically beneficial, cost-effective, and operationally feasible compared to other measures, such as increasing energy efficiency and lowering greenhouse gas emissions [18]. Prior research has established that the spatial makeup of these spaces, including size, shape, and distribution patterns, is instrumental in determining the cooling effect [19,20]. Moreover, different action thresholds for these influence processes have been reported in a significant body of the literature [21,22,23,24], while a more con-dispersed distribution of green areas tends to increase cooling intensity [25]. A negative correlation between the shape index and cooling effect intensity of green areas was confirmed [3]. Furthermore, water surface ratio [4,26], river width [27], and the size of a water body [10] were also identified as key components in decreasing the urban heat island effect. Moreover, the efficiency of cooling islands is directly proportional to the built-up land proportion around a water body.

Conclusively, a significant body of the literature extensively and quantitatively evaluated the impact characteristics of urban green spaces and water bodies on urban microclimate. Nonetheless, limited research was previously directed toward the investigation of green–blue aggregated spaces. The available evidence indicates that the synergistic cooling effect attributed to green–blue spaces is superior compared to that of single, isolated ecological spaces [28,29,30]. Typically, waterfront communities constitute green–blue spaces, and their thermal environmental issues are closely related to residents’ living, working and public leisure activities. This study accentuates that waterfront vegetation can regulate the water temperature via shading, hence affecting the radiation balance of the water body. This process significantly lowers the high temperature of the waterfront space [31]. Moreover, it is noteworthy that vegetated riverbanks undergo a more effective cooling process compared to hard-surface riverbanks [32]. Green–blue spaces’ cooling effect is positively associated with the width of the green corridor, and a width of 20–25 m is ideal for waterfront green corridors [27]. The network pattern of blue and green landscape spaces are shown to have a positive impact on the diffusion of cold air along the waterfront space [18,33]. Therefore, it is necessary to explore the synergistic cooling effect of green–blue spaces from a holistic and systematic perspective, and investigate the marginal effects of their influencing factors. Marginal effect refers to the increased efficiency resulting from the increase in the supply of one factor within a certain range, while upon overcoming the threshold range, additional input can result in reduced efficiency [34,35]. Therefore, marginal effects can effectively clarify the extent of the impact of various factors in the urban thermal environment, particularly when considering the interaction between these factors [36,37]. Furthermore, through identifying key values, i.e., the action thresholds at which different configurations of indicators exert their maximum cooling effect [38], this study can proceed to adjust the configuration of blue and green spaces under the limited urban construction space, thereby achieving the maximum cooling effect.

To attain the aforementioned objective, this research examined 30 waterfront communities located in Kunshan City, Jiangsu Province, and leveraged ENVI-met, a dynamic microclimate simulation software, to measure surface thermal environment attributes. Utilizing correlation analysis and the boosted regression trees (BRT) model, the contribution rate and relative importance of each spatial feature (e.g., area, distribution, shape, green–blue relationship, and built environment attributes) towards the cooling effect under the interaction of diverse green–blue space factors were determined. Additionally, this study examined the marginal effects of each spatial factor on the thermal environment of waterfront communities. With a focus on climate adaptation, this research aims to provide the criteria and methodology for enhancing green–blue spaces in waterfront communities.

2. Materials and Methods

2.1. Study Area

Kunshan City is located in Jiangsu Province, eastern China, between longitude 120°48′ E–121°09′ E and latitude 31°06′ N–31°32′ N. It belongs to the northern subtropical southern monsoon climate zone, with a total area of 931.5 km². Kunshan City is a typical water network city, being rich in blue and green ecological resources. There are 2815 rivers and 38 lakes in the city, with a total length of 2820 km. The rich waterfront green areas are concentrated in the central area of the city. In recent years, with the expansion of urban construction, the heat island effect in Kunshan widened year by year, and the heat island intensity increased year by year. The central city repeatedly experiences extreme summer heat and warm winter weather.

This study focused on the central urban area of Kunshan City, and 30 typical urban waterfront communities were selected as the research sites (Figure 1). These waterfront communities are located in the central construction area of the city, and are mostly surrounded by commercial and residential land uses. They are distributed along major urban rivers, such as Zhangjiagang River, Xiaoyu River, Hanputang River, Qingyanggang River, and Lou River, mostly facing water from the east, south, and north, and fully exposed to prevailing winds in summer. At the same time, the neighborhoods are surrounded by linear waterfront green corridors, city parks, and green spots with various patterns of blue and green spatial configurations. Studies have pointed out that neighborhood and building form can affect temperature distribution [39,40,41,42]. To eliminate the influence of building environment on the thermal environment evaluation, the waterfront communities selected for this study have regular boundaries with similar internal building distribution patterns. They are predominantly multi-story (4–6 stores) and mid-rise (7–9 stores) buildings, with a controlled range of 15–25 ha and an average area of 20.27 ha.

2.2. Research Framework

The research framework of this study is shown in Figure 2, and consisted of four main steps: data acquisition, variable extraction, model construction, and results analysis. The first step was to collect and process the necessary spatial and meteorological data; the next step was to carry out the quantitative characterization of the thermal environment and green–blue spatial configuration of the waterfront space, with the assistance of Fragstats, ArcGIS, and ENVI-met software. In the third step, this study constructed correlation and BRT models between the urban green–blue landscape configuration indicators and the thermal environment characteristics; in the results analysis section, this study sorted the key allocation indicators that affect the cooling effect, and ranked their importance based on the model results, calculating the action thresholds. This approach provided a scientific basis for the optimization of green–blue spatial configurations under the guidance of climate suitability in the future.

2.2.1. Data Source and Description

The data used in this study were derived from official government, online open source, and field research data sources, including the following: (1) Landsat8 OLI_TIRS remote sensing imagery on 26 August 2021, with an image accuracy of 30 m. Data was sourced from the Geospatial Data Cloud website (https://www.gscloud.cn (accessed on 7 October 2021)), an open source data site developed by the Computer Network Information Center of the Chinese Academy of Sciences, which was mainly used to record and re-check the basic conditions of this study; (2) meteorological data for the central area of Kunshan City, collected on 26 August 2021, to support validation of the simulation results of the ENVI-met model. Data were obtained from China National Meteorological Science Data Center (http://data.cma.cn (accessed on 26 August 2021)), a platform directly managed by China’s ministry of science and technology, while actual observations were provided by the Kunshan Meteorological Station, which is located in the eastern part of this study area, near the Lou River (Figure 1). Hour-by-hour data on temperature, humidity, mean wind speed, and wind direction were collected throughout the day of this study. The actual temperature measured on that day was the highest average daily temperature in Kunshan in 2021, reaching 29.9 °C. The wind direction was northeast to east to southeast, with an average wind speed of 0–5.3 m/s and no unusual meteorological phenomena; and (3) data on urban green areas, water bodies, buildings, and subsurface, mainly sourced from the Kunshan Bureau of Natural Resources and Planning. Refined corrections were made via combining remote sensing images and field observation data.

2.2.2. Variable Extraction

• Variables for characterizing landscape configurations of green–blue spaces

Through a combination of literature review and field research, a total of five distinct categories comprising ten spatial indicators were selected to effectively characterize the green–blue landscape configurations of waterfront communities. These categories include scale, distribution, morphology, green–blue relationships, and built-up area (

Table 1. Variables for characterizing landscape configurations of green–blue spaces.

Spatial Characteristics	Variable	Description	Calculation Formula	References
Scale	Percentage of Landscape (PLAND)	The proportion of green space or water patch types in a spatial unit, which is the most common indicator used to evaluate urban green–blue space.	Equation (1) Where aij is the area of patch ij in study unit i (m2), and Ai is the area of study unit i (m2). $(0\le {\mathrm{PLAND}}_{\mathrm{i}}\le 100)$	[43,44]
	Largest Patch Index (LPI)	The size of the largest patch of green space or water body in a spatial unit, with a higher value indicating a larger size of the largest patch in the spatial unit.	Equation (2) Where Max(aij) is the area of the largest patch ij in unit i (m2), and Ai is the area of study unit i (m2) $\left(0 < {\mathrm{LPI}}_{\mathrm{i}}\le 100\right)$.	[45,46]
	Mean Patch Index (MPI)	The average size of green space or water patches in a spatial unit, with a higher value indicating a larger average size of patches in a spatial unit.	Equation (3) Where n is the number of patches in study unit I, aij is the area of patch ij in study unit i (m2), and Ai is the area of study unit i (m2)$\left(0<{\mathrm{MPI}}_{\mathrm{i}}\le 100\right)$.	[43,44]
Distribution	Patch Density (PD)	The number of green space or water patches in a spatial unit, reflecting the density of distribution of certain patches in the built environment. A Larger value indicates a denser distribution of patches.	Equation (4) Where nj iis the number of patches of a certain type j in spatial unit i, and Ai is the area of study unit i (m2)$\left({\mathrm{PD}}_{\mathrm{i}}\ge 0\right)$.	[25,44]
	Aggregation Index (AI)	The degree of connectivity and aggregation between patches of a particular type of green space or water body in a spatial unit, with smaller values indicating a more dispersed spatial distribution of this type of landscape.	Equation (5) Where aij is the common sides of patch i, and j, max→aij is the maximal common sides of patch type $\left(0\le {\mathrm{AI}}_{\mathrm{i}}\le 100\right).$	[47,48]
Morphology	Related Circumscribing Circle (RCC)	The linearity of the shape of a particular type of green space or water patch in a spatial unit. When the value is close to 0, the patch shape is closer to circular; the larger the value, the closer the patch shape is to linear.	Equation (6) Where aij is the area of patch ij in study unit i (m2), and ${\mathrm{a}}_{\mathrm{ij}}^{\mathrm{S}}$ is the area of the minimum circumscribing circle of patch ij in study unit i (m2) $(0<{\mathrm{RCC}}_{\mathrm{i}}<1).$	[49]
	Landscape Shape Index (LSI)	The complexity of the patch shape of a green space or water body in a spatial unit, which is found via calculating the deviation of a patch shape from a circle or square of the same area.	Equation (7) Where Eij is the total length of patch j (m) in study unit i, and Ai is the area of study unit i(m2) $\left({\mathrm{LSI}}_{\mathrm{i}}\ge 1\right).$	[26,50]
Green–blue relationship	Percentage of green space and water bodies (P_GB)	The total proportion of green–blue ecological space in a spatial unit.	Equation (8) Where aig_j is the area of green space patch j in study unit i (m2), aiw_k is the area of water patch k in study unit i (m2), and Ai is the area of study unit i (m2)$\left(\mathrm{P}\_\mathrm{GW}\ge 0\right).$	[46,51,52]
	Average width of water green space (AW_GB)	The width of the adjacent green space around a water body in a spatial unit.	Equation (9) Where Ai_gb is the area of green space adjacent to the water body in study unit i (m2), and Lw is the total length of the water in study unit i (m)$\left(\mathrm{AW}\_{\mathrm{GB}}_{\mathrm{i}}\ge 0\right).$	[27]
Built-up environment	Floor area ratio (FAR)	The intensity of built development within a spatial unit, with an unlimited range of values.	Equation (10) Where Ai_b is the area of the building in study unit i (m2), and Ai is the area of study unit i (m2) $\left({\mathrm{FAR}}_{\mathrm{i}}\ge 0\right).$	[51]

). The scale indicators describe the proportion of urban green space or water patches within a given area, thus serving as the most direct indicators for measuring the cooling effect of green–blue space. The distribution characteristics, on the other hand, highlight the structural configuration of different types of green space or water patches; morphological indicators characterize the geometry of green space or water patches; and green–blue relationship captures the coupling and relational characteristics of water bodies and green patches in waterfront communities. Finally, built-up environment characteristics provide a spatial description of the relationship between the green–blue space and other built-up spaces.

In sum, each of these categories of indicators plays a critical role in influencing the diffusion and heat exchange processes of cold air in waterfront spaces, ultimately impacting the cooling effect of green–blue landscapes within these communities. However, differences exist regarding the relative importance and marginal effects characteristics within each of these categories. The measurement of each of these types of variables was referred to existing studies, and was performed in Fragstats 4.2 and ArcGIS 10.6 software platforms.

$${\mathrm{PLAND}}_{\mathrm{i}}=\frac{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{a}}_{\mathrm{ij}}}{{\mathrm{A}}_{\mathrm{i}}}\times 100$$

(1)

$${\mathrm{LPI}}_{\mathrm{i}}=\frac{\mathrm{Max}\left({\mathrm{a}}_{\mathrm{ij}}\right)}{{\mathrm{A}}_{\mathrm{i}}}\times 100$$

(2)

$${\mathrm{MPI}}_{\mathrm{i}}=\frac{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{a}}_{\mathrm{ij}}}{{\mathrm{nA}}_{\mathrm{i}}}\times 100$$

(3)

$${\mathrm{PD}}_{\mathrm{i}}=\frac{{\mathrm{n}}_{\mathrm{i}}}{{\mathrm{A}}_{\mathrm{i}}}\times 100$$

(4)

$${\mathrm{AI}}_{\mathrm{i}}=\frac{{\mathrm{a}}_{\mathrm{ij}}}{\max \to {\mathrm{a}}_{\mathrm{ij}}}\times 100$$

(5)

$${\mathrm{RCC}}_{\mathrm{i}}=1-\frac{{\mathrm{a}}_{\mathrm{ij}}}{{\mathrm{a}}_{\mathrm{ij}}^{\mathrm{S}}}$$

(6)

$${\mathrm{LSI}}_{\mathrm{i}}=\frac{0.25{\mathrm{E}}_{\mathrm{ij}}}{\sqrt{{\mathrm{A}}_{\mathrm{i}}}}$$

(7)

$$\mathrm{P}\_{\mathrm{GW}}_{\mathrm{i}}=\frac{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{a}}_{\mathrm{ig}\_\mathrm{j}}+{\sum }_{\mathrm{k}=1}^{\mathrm{n}}{\mathrm{a}}_{\mathrm{iw}\_\mathrm{k}}}{{\mathrm{A}}_{\mathrm{i}}}\times 100$$

(8)

$$\mathrm{AW}\_{\mathrm{GB}}_{\mathrm{i}}=\frac{{\mathrm{A}}_{\mathrm{i}\_\mathrm{gb}}}{{\mathrm{L}}_{\mathrm{w}}}$$

(9)

$${\mathrm{FAR}}_{\mathrm{i}}=\frac{{\mathrm{A}}_{\mathrm{i}\_\mathrm{b}}}{{\mathrm{A}}_{\mathrm{i}}}$$

(10)

2.

Variables for characterizing the cooling effect of green–blue space

The UHI phenomenon contributes significantly to the frequent occurrence of extreme heat events and long-term warming trends in central urban areas during summer. This phenomenon manifests primarily in the surface temperature of the city. Consequently, this study adopted two indices, namely the maximum air temperature at pedestrian level (1.5 m above ground level) and the mean intensity of urban cooling islands (UCImean), to exemplify the cooling impact of green–blue spaces. Compared to average temperatures, the maximum air temperature served as a more comprehensible indicator of the capacity of green–blue spaces to mitigate extremely high temperatures across the region [53]. This measure was a critical aspect of comprehending the phenomenon of surface temperature change and improving the habitat. In contrast, mean urban cold island intensity was better suited to quantifying the magnitude and intensity of cooling effect within each spatial unit of green–blue landscapes. This index is commonly employed in related studies [26], being calculated as.

$${\mathrm{UCI}}_{\mathrm{mean}}=∆{\mathrm{T}}_{\mathrm{mean}}=\overline{ \mathrm{T}}-\overline{{ \mathrm{T}}_{\mathrm{i}}}$$

(11)

where $\bar{\mathrm{T}}=\frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{t}}_{\mathrm{i}}}{\mathrm{n}}$ is the average air temperature in the study area, i.e., the average daily air temperature of a total of 30 communities with n rasters during the simulation period,$\overline{{ \mathrm{T}}_{\mathrm{i}}}$ is the average daily air temperature between simulation periods in study community i. A positive value means that community is relatively a cold island area, and a negative value means a relatively hot region. i.e., the higher the value, the stronger the cooling effect.

2.3. Study Methods

2.3.1. Measuring the Characteristics of the Urban Thermal Environment Based on ENVI-Met

The thermal environment of waterfront communities was studied through the utilization of ENVI-met 5.0.1 Student Edition for simulation and measurement. ENVI-met is a 3D urban microclimate simulation software developed by Michael Bruse and Heribert Fleer of the Research Group Climatology, Institute for Geography, Ruhr-University, Bochum, in 1998 for the dynamic simulation of the interaction of surface–vegetation–air in small-scale urban three-dimensional spaces [54]. ENVI-met consists of a one-dimensional boundary model, three-dimensional core model, soil and surface model, and nested grids, the latter of which are designed to minimize the interferences resulting from boundary effects. The three-dimensional core model includes a building model, vegetation model, atmospheric model, radiation model, etc.

Considering that the modeling area had a large influence on the model results, this study further defined the modeling area when simulating the 30 waterfront communities in 3D. The horizontal modeling area was required to include at minimum one complete river or road surrounding the original waterfront community boundary, extending outwards for a distance of at least one building, while the extension distance was at least greater than the height of the tallest building in each unit. The vertical modeling height was set at twice the tallest building within the modeling area. Once the modeling area was determined, information on the distribution and types of water bodies, green spaces, roads, buildings, and other sub-surfaces were vectorized within the software, with information on the types of trees and building heights input at a finer level of detail.

Subsequently, the ENVI-guide module was employed to input the simulation parameters, along with the measured meteorological data. The steady-state wind boundary conditions for the model input were selected, with wind speed and direction assessed at the meteorological station in Kunshan City on 26 August 2021. Specifically, a constant wind speed measured at 10 m height as 1.43 m/s, with a constant wind direction of 149.23°. Temperature and humidity data were input in the module as hourly temperature and relative humidity data throughout the day on 26 August.

To ensure the accuracy of the simulated data, a representative sample of a waterfront community—R1—was selected for validation. The simulation start time was set to two hours before the lowest temperature point of the day, at 4:00 a.m. on 26 August 2021, and the simulation duration was set to 12 and 24 h, respectively, with the remaining meteorological parameters maintained as aforementioned. The results indicated that the mean air temperature measured for the community on that day was 31.055 °C (Figure 3a), and the simulated mean temperature for both simulation periods was 31.02 °C, which differed only by 0.035 °C from the measured data. Hence, it was concluded that the ENVI-met software simulation results for the average daily temperature of the waterfront community at this scale closely approximated the true values.

To streamline the simulation process, numerical simulations for the 30 waterfront communities were unified, and the maximum air temperatures at pedestrian height and mean cold island intensity were calculated for a total of 12 h from 4:00 a.m. to 4:00 p.m. on 26 August 2021 (Figure 3b).

2.3.2. Correlation and BRT Analysis

BRT is a versatile regression modeling method designed to improve the performance of a single model through fitting multiple models and combining their predictions [55]. Through leveraging the strengths of both regression trees and boosting models, BRT generates multiple regression trees while utilizing a randomly selected subset of data to fit each new tree, thus reducing the variance of the final model and significantly enhancing the accuracy of predictions [55]. Due to its capacity to handle various types of response variables, such as normal, count, etc., using appropriate and robust loss functions, BRT has been widely used in studies of urban development and its influencing factors, as well as studies of the relationship between UHI effect and configuration indicators [56,57].

To investigate the link between urban green–blue landscape configurations and cooling effects, this study selected 30 waterfront communities as the sample. Through applying professional statistical analysis software SPSS Statistics 26, this study conducted correlation and collinearity analysis to identify key configuration indicators affecting summer cooling effects. Subsequently, using the dismo package of R programming language, this study constructed BRT, with maximum air temperature and mean urban cold island intensity in waterfront communities being the dependent variable and five categories of indicators serving as the independent variables. Furthermore, this study utilized this model to disentangle the effects of each indicator on the characteristics of the urban thermal environment under the combined impact of complex spatial factors. The cooling contribution and marginal effects of spatial indicators were quantified and analyzed, and the action threshold was clarified. The BRT model was set up with the following parameters: learning rate of 0.01, tree complexity of 0.5, and split ratio of 0.5. Each time, 50% of the data was taken for training and 50% of the data was taken for analysis, and cross-validation was performed 10 times. Ultimately, this study obtained the relative contribution and marginal effects of configuration indicators on the intensity of summer cooling effect.

3. Results

3.1. Spatial Characteristics of the Thermal Environment Distribution

• Maximum air temperature at pedestrian height

Formulated on the outcomes yielded via the ENVI-met simulations, the highest recorded surface temperature at pedestrian altitude was assessed for each waterfront community (Figure 4). The recorded figures ranged from 30.94 °C to 31.83 °C, with an average value of 31.76 °C and a standard deviation of 0.23. The maximum temperatures for the 30 communities were categorized into five levels and plotted using the natural breakpoint method. The diagrammatical depiction reveals that the general trend of the highest temperature values in the waterfront communities decreases from west to east. The western waterfront districts are surrounded by a large area of parkland; however, the linear riverfront green and road green spaces within the communities are narrower, with the highest value located in Unit R15 on the north bank of the Lou River. The central and eastern waterfront communities, located within the city’s central development area, are mostly adjacent to the city’s first-level river, possessing a wider river space and abundant riverfront green space. The lowest temperature measurement is ascertained in Unit R11 on the east bank of Hanpu Pond, which is edged by water on two sides and endowed with incessant road green space and waterfront green space.

2.

Mean urban cold island intensity

Using the outcomes generated using the ENVI-met simulation, the mean urban cold island intensity was ascertained for each waterfront community (Figure 5). The resulting values ranged from −0.47 °C to 0.43 °C, with a mean value of 0 °C and a standard deviation of 0.24. The examination of the data allocated the mean urban cold island intensity of the 30 communities into 5 classes via the natural breakpoint method, which was then graphically presented. The display revealed that certain western waterfront communities exhibited lower values, while their eastern counterparts demonstrated significantly higher values. Notably, the maximum value was recorded within Unit R3, situated along the west bank of the Yehe River, which was recognized for its extensive internal green space, spacious riverfront green space, and tree-lined roads and riversides. Comparatively, the least value was observed in the westernmost Unit R18, which was marked by a narrower river adjacent, tall buildings, sparse internal green space, and limited tree coverage along the riverbank.

3.2. Characteristics of Green–Blue Landscape Configurations in Waterfront Communities

Table 2. Results of green–blue landscape configurations in 30 waterfront communities.

Configuration Indicator			Range	Maximum	Minimum	Mean	Median	C·V (Coefficient of Variation)
Scale	Green space	PLAND_G	31.2	41.2	10.0	25.9	25.0	0.33
		LPI_G	35.4	36.5	1.1	8.7	6.9	0.80
		MPI_G	0.2	0.2	0.0	0.1	0.1	0.63
	Blue space	PLAND_B	31.6	32.9	1.3	10.5	9.0	0.66
		LPI_B	31.6	32.9	1.3	9.6	7.0	0.76
		MPI_B	5.1	5.2	0.2	1.4	1.2	0.76
Distribution	Green space	PD_G	545.7	731.2	185.4	425.8	409.6	0.36
		AI_G	23.5	95.0	71.5	88.9	89.6	0.05
	Blue space	PD_B	19.5	22.9	3.3	8.9	7.5	0.53
		AI_B	10.8	99.4	88.6	95.6	96.2	0.03
Morphology	Green space	RCC_G	0.3	0.7	0.4	0.6	0.6	0.10
		LSI_G	0.3	1.0	0.6	0.9	0.9	0.08
	Blue space	RCC_B	17.4	25.1	7.6	13.8	13.6	0.29
		LSI_B	5.6	7.2	1.6	3.7	3.5	0.35
Green–blue relationship		P_GB	33.4	55.0	21.6	36.4	35.9	0.24
		AW_GB	65.4	65.4	0.0	22.1	17.5	0.77
Built-up environment		FAR	2.0	2.9	0.9	1.3	1.1	0.5

Note: $\mathrm{C}·\mathrm{V}=\mathrm{SD}÷\mathrm{MEAN},$ where SD is standard deviation of these variables. $\mathrm{C}·\mathrm{V}$ is used to characterize degree of dispersion of values of variables.

shows the characteristics of the green–blue space configuration of the 30 waterfront communities. In general, all communities have some area of green space and water bodies, with P_GB ranging from 21.6 to 55.0% and a mean value of 36.4%. However, the pattern characteristics of green and blue spaces vary considerably. Influenced by the width of the surrounding rivers and construction of waterfront green spaces, the scale indicators and distribution indicators, such as LPI_G and MPI_G (SD = 0.80, 0.63) for green spaces, PLANG_B, LPI_B, and MPI_B (SD = 0.66, 0.76, 0.76) for water bodies, and PD_B (SD = 0.53) for water bodies, vary greatly from community to community. Moreover, there is some data duplication for PLANG_B and LPI_B. In contrast, there is little difference between the RCC of green space and water body space, with green space being mostly clustered and water bodies being distributed in bands. The LSI of water bodies is influenced by urban texture with some differences. Analysis of AW_GB shows that there are some communities that have not built waterfront green spaces, and the values of the variables vary too much between communities. The FAR, on the other hand, is mostly influenced by the height of the floors, and there is some fluctuation.

3.3. Influencing Characteristics of Green–Blue Landscape Configuration Indicators on the Cooling Effect

3.3.1. Correlation Analysis

This study correlated the maximum air temperature at pedestrian height and mean urban cold island intensity with the five types of spatial configuration indicators of the waterfront community respectively, and the results are shown in

Table 3. Results of correlation analysis between green–blue landscape configuration indicators and maximum air temperature at pedestrian height and mean urban cold island intensity in 30 waterfront communities.

Configuration Indicator			Correlation Analysis Methods	Maximum Air Temperature	Sig Value	Mean Cold Island Intensity	Sig Value
Scale	Green space	PLAND_G	Pearson’s correlation	−0.16	0.399	0.507 **	0.004
		LPI_G	Spearman’s correlation	0.055	0.775	0.377 *	0.04
		MPI_G	Spearman’s correlation	−0.188	0.321	0.546 **	0.002
	Blue space	PLAND_B	Pearson’s correlation	−0.435 *	0.016	0.434 *	0.016
		LPI_B	Spearman’s correlation	−0.405 *	0.026	0.333	0.073
		MPI_B	Pearson’s correlation	−0.375 *	0.041	0.480 **	0.007
Distribution	Green space	PD_G	Pearson’s correlation	0.524 **	0.003	−0.345	0.062
		AI_G	Pearson’s correlation	−0.361 *	0.031	0.537 **	0.002
	Blue space	PD_B	Spearman’s correlation	−0.009	0.962	−0.199	0.291
		AI_B	Pearson’s correlation	−0.431 *	0.017	0.386 *	0.035
Morphology	Green space	RCC_G	Spearman’s correlation	−0.132	0.486	0.063	0.739
		LSI_G	Spearman’s correlation	0.269	0.15	−0.221	0.241
	Blue space	RCC_B	Spearman’s correlation	0.466 **	0.01	−0.112	0.556
		LSI_B	Pearson’s correlation	0.473 **	0.008	−0.387 *	0.034
Green–blue relationship		P_GB	Pearson’s correlation	−0.515 **	0.004	0.864 **	0
		AW_GB	Spearman’s correlation	−0.112	−0.112	0.378 *	0.039
Built-up environment		FAR	Spearman’s correlation	0.453 *	0.012	−0.431 *	0.017

Note: ** and means correlation is significant at the 0.01 level (2-tailed); * and means correlation is significant at the 0.05 level (2-tailed). _G is green space configuration indicator, _B is blue space configuration indicator, and _GB is green–blue space configuration indicator.

. Pearson’s correlation analysis was conducted for normally distributed indicators and Spearman’s correlation analysis was conducted for non-normally distributed indicators.

The results showed that the maximum air temperature at pedestrian height was significantly correlated with 10 configuration indicators, with a stronger correlation observed for water bodies configuration indicators and a weaker correlation for built-up land characteristics. Similarly, the average cold island intensity was significantly correlated with 11 configuration indicators, with stronger correlation observed for green–blue relationship and weaker correlation for the morphological feature of green spaces.

Further analysis revealed that patch density of green space (PD_G) exhibited the strongest correlation with maximum air temperature at pedestrian height in 30 waterfront communities (Table 3). Overall, the correlations are from strong to weak: PD_G > P_GB > LSI_B > RCC_B > FAR > PLAND_B > AI_B > LPI_B > MPI_B > AI_G. The strongest correlation with mean urban cold island intensity is percentage of green and blue spaces(P_GB), with an absolute value of 0.864. Overall, the correlations are from strong to weak: P_GB > MPI_G > AI_G > PLAND_G > MPI_B > PLAND_B > FAR > LSI_B > AI_B > AW_G B> LPI_G.

To ensure that each indicator has a sufficiently independent explanatory meaning when regression analysis is performed, we performed an assessment of the collinearity among the green–blue spatial configuration indicators, which revealed that percentage of blue space and largest water bodies patch Index (PLAND_B and LPI_B) exhibited a VIF value >10 and had collinearity. Considering the non-substitutable characterisation significance of PLAND_B, this study excluded the indicator of LPI_B, and the remaining 13 indicators all passed the collinearity analysis (

Table 4. Results of collinearity analysis of green–blue landscape configuration indicators in 30 waterfront communities.

Configuration Indicator			VIF Value	Tolerance Value
Green space	Scale	PLAND_G	4.235	0.236
		LPI_G	2.029	0.493
		MPI_G	4.552	0.22
	Distribution	PD_G	3.764	0.266
		AI_G	3.144	0.318
Blue space	Scale	PLAND_B	2.791	0.358
		MPI_B	2.555	0.391
	Distribution	AI_B	5.215	0.192
	Morphology	RCC_B	1.559	0.642
		LSI_B	3.023	0.331
Green–blue relationship		P_GW	1.089	0.918
		AW_GW	1.148	0.871
Built-up area		FAR	1.156	0.865

Note: _G is green space configuration indicator, _B is blue space configuration indicator, and _GB is green–blue space configuration indicator.

).

3.3.2. BRT Analysis

The present study employed BRT analysis to evaluate the individual contributions of landscape configuration indicators to the cooling effect of the waterfront community, including two variables, maximum air temperature at pedestrian height, and mean urban cold island intensity.

• Maximum air temperature at pedestrian height

Among the eight configuration indicators influencing maximum air temperature at pedestrian height in waterfront communities (Figure 6a), the relative contributions of PD of green space (PD_G) and LSI of water bodies (LSI_B) are higher, being 21.298% and 20.927%, respectively. The relative contribution of three factors—P_GW, RCC_B and FAR—exceeded 10%.

Overall, the combined relative contributions of blue, green, and green–blue spaces and built-up environment indicators were found to be 50.457%, 25.458%, 13.783%, and 10.302%, respectively. Specifically, the distribution of green space and morphological characteristics of blue space exerted a greater influence on the maximum summer temperature at the community scale in Kunshan City.

2.

Mean urban cold island intensity

Among the 11 spatial characteristics that affect the mean urban cold island intensity of the waterfront community (Figure 6b), P_GW exhibited the highest relative contribution, being 32.264%, followed by MPI of green space (MPI_G) at 17.827% and AI of water bodies (AI_B) at 14.435%. Values of the remaining configuration indicators did not exceed 10%, and the least impact that a configuration indicator had on the mean urban cold island intensity pertained to LPI of green space (LPI_G), which was a mere 0.867%.

Overall, the combined relative contributions of green–blue relationship, blue space, and green space constituted 38.266%, 29.758%, and 25.049%, respectively. Conversely, the combined relative contribution of built-up land indicators was low, being 6.927%. Under this framework, among the indicator types, the green–blue relationship, scale of green space, and distribution characteristics of blue space exerted greater influence on the mean urban cold island intensity at the community scale in Kunshan City.

3.4. Marginal Effects of Green–Blue Landscape Configuration Indicators on the Cooling Effect

This study conducted an analysis of marginal effects of spatial configurations on the cooling effect in 30 waterfront communities based on BRT analysis (above 10%). Based on the inflection point and trend analysis, thresholds for the effect of each factor on summer cooling effect in waterfront communities were determined.

• Maximum air temperature at pedestrian height

The results of the analysis of the maximum temperature of pedestrian height are shown in the Figure 7a, with five important configuration indicators. All four indicators are positively correlated with the maximum air temperature at pedestrian height. The two factors with the highest contribution, namely PD_G and LSI_B, have threshold values of 550 and 4.2, respectively. RCC_B has an effect threshold of 0.945, and FAR has an effect of 1.5. Conversely, P_GW is negatively correlated with the maximum air temperature at pedestrian height. When the total proportion of green–blue space is between 36% and 41.5%, the effect of mitigating local extreme high temperature is significant.

2.

Mean urban cold island intensity

The results of the analysis of the marginal effects of green–blue landscape configurations characteristics on the mean urban cold island intensity in 30 waterfront communities are shown in the Figure 7b, with 3 important configuration indicators. Among them, two factors with high contribution rates are positively correlated with mean urban cold island intensity. P_GW between 32.5% and 43% and MPI_G between 3% and 8% have a greater impact on the mean urban cold island intensity. Conversely, the value of AI_B fluctuates with the mean urban cold island intensity, with no significant positive or negative trend.

4. Discussion

4.1. The Influence of Green–Blue Landscape Configurations on the Cooling Effect

As urban development intensifies, surface temperatures tend to increase, making green–blue spaces an effective mitigating solution. Waterfront communities near urban rivers that have more riverfront green space and woodland tend to have more favorable microclimate conditions. The characteristics of green–blue space patterns that have a significant impact on urban summer cooling include the scale and distribution of green spaces, distribution and morphological characteristics of water bodies, relationships of green–blue spaces, and the surrounding built-up land ratio.

Notably, the distribution of green spaces and morphology of water bodies have a more significant impact on the maximum air temperature, followed by the proportion of green–blue spaces. More dispersed distribution of green spaces and more regular water body shapes result in a more effective reduction in localized high temperatures in the community, whereas communities with a plot ratio of less than 1.5 and increased blue and green spaces contribute to reducing the urban heat island effect.

Moreover, the green–blue relationship is found to have the greatest influence on the average cold island intensity, followed by the area of green space. Increasing the scale of green and blue space is effective in improving the cooling effect of waterfront communities, while maximizing green space where the expansion of blue space is limited is equally effective in reducing average temperatures.

These findings are consistent with the previous literature review, which found that green and blue spaces directly mitigate the thermal environment of communities. In general, dispersed green spaces are better at blocking heat and promoting evaporation in a greater number of areas. Conversely, concentrated bodies of water generally represent communities close to wider and more intact urban waterways, allowing for more complete heat exchange with the assistance of wind. Both of these factors can be effective in mitigating localized heat problems in communities. In contrast, the scale of green spaces and water bodies does not have a significant effect in reducing the localized temperature compared to mitigating the average daily temperature. This finding shows that we are no longer limited by the issue of land resources when facing climate problems, and can target the spatial relationship of green and blue spaces according to the problems in different places, effectively exploiting the synergistic cooling effect of blue and green spaces. This approach is not only limited to adjusting the area, with the direction of the waterfront, wind direction, and distribution and connectivity of water bodies and green spaces all being important factors. The cooling air delivery role of the green–blue space network pattern is affirmed. For example, the layout of an east-west-oriented waterfront green belt can enhance circulation of cold air between the river and waterfront green space, which can effectively increase the distance and intensity of the cooling effect. Without taking up a large amount of built-up area and adding substantial water bodies or green spaces, small areas of spatial renewal and adaptive ecological measures can achieve considerable thermal environment optimization.

4.2. Synergistic Strategies for Optimizing the Landscape Configurations of Green–Blue Spaces

Therefore, to address the urban heat island problem in waterfront communities, synergistic strategies are recommended that focus on multiple dimensions, such as scale, distribution, morphology, green–blue relationship, and cityscape relationship. Based on the thermal environment of waterfront communities, specific measures can be categorized into three types: (1) for waterfront communities with localized heat problems, optimizing green–blue space should emphasize the density of big green patches in hot areas and forms of water features, and control the proportion of hard paving and natural landscapes; (2) for waterfront communities requiring overall cooling, optimization should focus on the total proportion of green–blue space, concentration of large green areas, and proportion of ecological space within the community; (3) within waterfront communities with a primary objective of thermal environment improvement, optimizing green–blue space should consider total green–blue control, green space layout adjustment, green space patch area control, and water body morphology design to ensure a sufficient scale of waterfront green spaces and regular water bodies in high-temperature gathering areas.

4.3. Nature-Based Solutions for Enhancing Climate Resilience in Global Cities

Climate change is profoundly affecting every city around the globe and impacting people’s lives. In this study, the cooling effect of urban natural spaces was examined from a microscopic perspective in one of China’s coastal cities—Kunshan City—and the key role of urban green–blue ecosystems in addressing climate issues was identified. As stated in “Climate Change 2022: Impacts, Adaptation and Vulnerability”, published by the Intergovernmental Panel on Climate Change (IPCC), it is crucial to protect urban ecosystems and undertake related adaptation actions. Specifically, it is not enough to adjust the scale and form of green spaces and water bodies. Adaptation actions must be carried out in the context of nature, institutions, and social culture, expanding from the single management of space to the integrated configurations of green–blue spaces and their surrounding built-up land, and thinking about how waterfront green–blue spaces can efficiently deliver their ecological and social benefits. Of course, thinking at community level does not fully support our exploration of nature-based solutions. We should continue to combine multi-disciplinary and multi-sectoral cooperation to create more value from the limited natural space in cities, effectively exploit the climate resilience of green–blue spaces, and enhance urban climate adaptation.

4.4. Limitations and Future Research

The present study investigates 30 waterfront communities in Kunshan, a region characterized by a high level of urban, economic, and social development, located in the water network area adjacent to Shanghai. Notwithstanding the representativeness of the selected sample, it is important to acknowledge certain limitations associated with the context of this study area, where several factors, such as the density of the river network, the intensity of site development, and the distribution of high-rise buildings, intersect to shape the thermal environment characteristics of the investigated communities. Consequently, the findings derived from this empirical study should be cautiously extrapolated to cities situated in non-water network areas or with a lower level of development. More generalized correlation results are needed to advance future typological studies.

To further refine the framework adopted for the analysis of the studied communities, the system of indicators for green–blue space allocation needs to be expanded and improved. In particular, additional indicators should be included to capture differences in the cooling effect of different types of green spaces. This approach would require the development of an extended set of three-dimensional configuration indicators aimed at exploring their respective influence on the cooling effect in a more detailed way. The measurement of the cooling effect of green–blue spaces also requires further investigation, specifically concerning the gradient of the effect on thermal comfort. Furthermore, to comprehensively assess the thermal comfort of residents in urban environments, it is essential to consider the combined impact of wind, temperature, and humidity. In this regard, future studies should strive to explore the combined effects of green–blue space configuration characteristics on these variables to inform more scientifically grounded and effective optimization of green–blue space configuration.

5. Conclusions

Through a combination of model simulations and mathematical analysis, this study provides an initial examination of the impact patterns of green–blue spatial configuration characteristics on the summer cooling effect at a community scale. The findings demonstrate that several green–blue spatial configuration indicators significantly contribute to the urban cooling effect in 30 typical water-front communities located in Kunshan City. These indicators include green space scale and distribution, water bodies distribution and morphology, green–blue relationship, and city–landscape relationship. Utilizing BRT analysis, the research outcomes establish visual quantification of the impact characteristics of green–blue spatial configuration features on the cooling effect, and determine the action thresholds of distinct spatial indicators. These results offer valuable guidance for the optimization of green–blue spatial configurations in cities.

More importantly, in a time of increasing climate problems, this study leads us to a comprehensive understanding of the cooling effect and other ecological effects of urban ecological spaces. It is necessary to focus on nature-based solutions that emphasize the key role of urban ecosystems, and overcome the shortcomings of a single infrastructure through the integration of green, blue, and grey spaces. In particular, coastal cities such as Kunshan City require the active promotion of the further integration of blue and green spaces, together with their hard spaces, from a socio-ecological perspective, for the purpose of harnessing the climate resilience of said spaces, thus facilitating sustainable urban development.