Constructing an Ecological Network Based on Heat Environment Risk Assessment: An Optimisation Strategy for Thermal Comfort Coupling Society and Ecology

Abstract

Urbanisation is a significant driver of global climate change. It increases global temperatures, impacting the health of residents. To date, research on urban heat environments has focused on society or ecology, overlooking the value of integrating the two factors. The research objective is to integrate socio-ecological benefits, explore the construction methods of ecological-cooling networks, and provide reasonable guidance for urban climate planning, thus contributing to the alleviation of urban heat risks and improving thermal comfort. Using Fuzhou as an example, an environmental risk framework was used to construct an urban heat environment risk assessment strategy based on hazards (thermal comfort), exposure (human-development footprint), and vulnerability (natural geographic conditions). The source area was identified based on evaluation results, an ecological network was constructed using circuit theory, and key nodes were identified. Results showed that in 2005 and 2020, 3% and 12% of areas in Fuzhou had higher thermal environmental risks, the proportion of low-risk areas was 43% and 28%, respectively. In sum, 54 ecological source locations, 124 ecological corridors, 76 ecological pinch points, 110 obstacle points, and 12 stepping stones were identified during the construction of corridors in 2020. Compared with 2005, the source area has decreased by 1622.46 km² and the average length of the corridor has also decreased by 4.69 km.

1. Introduction

Since the beginning of the 21st century, global urbanisation has accelerated, reaching a rate of 55% by 2018, and it is expected to reach 68% by 2050 [1]. These trends, in conjunction with the associated increases in population density, exacerbate global climate change. The Intergovernmental Panel on Climate Change (IPCC) reported that the current global temperature has significantly risen compared with that of the last century and predicted an increase of 1–3.7 °C by the end of the 21st century. However, the increase in impervious surfaces within cities that reduce heat dissipation has contributed to the urban heat island (UHI) effect. Urban temperatures are typically higher than suburban temperatures [2], increasing the risk of extremely high temperatures [3,4]. Recent natural disasters caused by persistently high temperatures, such as volcanic eruptions, forest fires, and glacier melting, have resulted in irreversible damage in the context of the environment [5,6].

With an increased research focus on thermal environmental challenges, scholars have studied the causes and discussed the impact mechanisms of urbanisation-induced land-use changes and landscape functional and morphological effects [7,8]. Recent studies have assessed the influence of natural geographic conditions on climate change vulnerability [9,10]. In terms of optimising thermal environmental measures, constructing ecological cities has emerged as a means of maintaining ecosystem functions. However, while urban ecosystems are complex natural–social–economic composite systems [11], most research has focused exclusively on one aspect of heat environment risk assessment, neglecting the overall urban ecosystem [12]. This causes an imbalance in ecological and economic development, impeding sustainable societal development [13]. Therefore, in 2014, the IPCC proposed an environmental risk framework driven by climatic systems and socioeconomic development, primarily including hazards, exposure, and vulnerability (IPCC 2014). These three aspects better reflect the complex climate change risks, focusing on the interactions between risk drivers and multiple risks [14]. However, existing research often considers the urban thermal environment formed by the aggregation of high-surface-temperature patches as hazards of urban thermal environment risk [15] and ignores the impact of disasters on people. Heat exposure poses a significant threat to human health with the increasing intensity and frequency of extremely high temperatures [16]. The human body senses extreme temperature changes, which cause comfort or discomfort, referred to as the human discomfort index (DI) [17]. However, human perceptions of thermal comfort are influenced by temperature as well as humidity, wind speed, and solar radiation [18].

To mitigate urban thermal environmental issues and control UHI, slowing population growth and urban development has been suggested [19]. This approach may exacerbate societal challenges without aptly addressing the root problem. Increasing blue–green spaces is a better solution to UHI [20]. Indeed, urban blue–green spaces can create a ‘cool island effect’ through evapotranspiration, shading, and heat absorption, mitigating thermal environmental effects [21]. However, certain blue–green spaces fail to provide effective cooling conditions due to various factors, such as climate conditions and terrain, among others [22], an issue often overlooked in research. Moreover, owing to limited land resources, cities are unable to solve the urban heat island issue by markedly increasing blue–green spaces [23]. Nevertheless, by changing the urban development model, restructuring urban land-use patterns, and adopting a socio-ecological strategy, cities can achieve long-term sustainable development [24].

Urban thermal environment risk assessment requires the analysis of interactions and cold–heat relationships between thermal risk patches to alleviate the urban heat island effect. Lin et al. revealed the influence of patch morphology on UHI, revealing that blue–green spaces with better connectivity have a stronger cooling capacity [25]. Ecological networks, a scientific spatial planning method, can achieve connectivity between urban spatial structure and function [26]. To solve the urban heat island challenge, a spatial structure model has been developed, comprising the source, corridor, and node. These networks are categorised as either heat island networks constructed out of areas with higher thermal environmental risks as source sites or cold island networks constructed by assessing ecosystem services, ecological sensitivity, and landscape connectivity [27]. Qian and Li showed that constructing networks by connecting fragmented cool islands can effectively cool the entire area [28], while poorly connected cool islands can reduce overall network functionality [29]. However, the strategy of constructing cold island networks based on thermal environmental risk assessment can better address the negative impacts attributed to high urban development [30]. Meanwhile, few studies have constructed cooling networks with the assessment of integrated socio-ecological factors, potentially causing one-sided functioning of cooling networks.

Additionally, the construction of a cold island network to optimise the ecological spatial pattern has been widely discussed [31]. Currently, the minimum cumulative resistance model (MCR) is employed to primarily construct cold conduction networks, while circuit theory is rarely applied in this area. Compared with MCR, circuit theory can more accurately determine corridors’ width and range while offering a certain advantage in identifying key nodes [32]. The constructed cold island network is more stable, which will be more conducive to alleviating urban thermal environmental risks and gradually achieving comprehensive social and ecological development [33].

The developmental benefits of socio-ecology were coupled using DI instead of traditional surface temperature indices, thereby incorporating human perceptions into thermal environmental risk assessments. Ultimately, spatial optimisation of thermal comfort is achieved by constructing thermal environmental risk ecological networks. Additionally, the aim was to balance the contradictions between human activities and ecological risk in urban development, considering humans as the carriers of risk effects to optimise the constructed ecological network, achieve sustainable development of the urban thermal environment, and improve the well-being of residents. To this end, a case study in Fuzhou, the capital city of Fujian Province in China, which has a pronounced heat island effect and rapid urban development, was conducted (Figure 1).

2. Data Sources and Overview of Research Area

2.1. Study Area

Fuzhou, Fujian Province, China, is in the south-eastern coastal region of China and downstream of the Minjiang River Basin (Figure 2), a typical estuarine basin terrain [34] with an overall landscape pattern of ‘eight mountains, one water, one field’. Fuzhou is rich in natural resources and is a mountainous city, with mountainous terrain covering 72.68% of its total land area. The plain area is only 489 km², accounting for 4% of the total land area. It is precisely due to these unique geographical and climatic conditions that the dissipation of heat is challenging in Fuzhou, resulting in a pronounced UHI effect. Fuzhou has a subtropical monsoon climate with hot and humid summers and warm and dry winters [35]. The average temperature in 2021 was 21.8 °C, 1.5 °C higher than in 2005. Moreover, as the capital city of Fujian Province, Fuzhou has experienced a population growth of 2.65 million people in the past 20 years, with an increase in the built-up area from 102 km² to 305.3 km² [36]. With the rapid urban development and population boom, the temperature of the built-up area of Fuzhou became 4 °C higher than that of the green areas [37], further exacerbating the discrepancy between social development and the thermal environment.

2.2. Data

The data used in this study include land use, elevation, normalised difference vegetation index (NDVI), meteorological, potential evapotranspiration, traffic network, demographic, economic, and night-time lighting data. Supplementary Table S1 lists all data sources used.

3. Methods

3.1. Thermal Environmental Risk Assessment

To address the primary research objective, urban thermal environmental risks were evaluated using an environmental risk framework, considering hazards, exposure, and vulnerability. This study incorporated the DI into the risk assessment, allowing hazards to not only represent environmental heat risk but also include human thermal comfort perception. Research has demonstrated that thermal exposure is closely related to rapid urban development [38]. Therefore, the human footprint index was used to represent the exposure to heat risk, thereby reflecting the impact of socioeconomic factors on the heat environment risk. Natural factors, such as climate and mountainous terrain, are used to assess the vulnerability of heat risk, representing the ecological resilience of the natural environment when faced with external disturbances [9] and reflecting the influence of natural ecological factors on heat environment risk. The combined effect of these three factors determines the intensity of urban heat environmental risk (Figure 3) [39].

3.1.1. Hazards—Environmental Thermal Comfort

The thermal comfort of the environment is jointly determined by temperature and relative humidity. Land-surface temperature can be calculated using bands 3, 4, and 6 of Landsat 5 and bands 4, 5, and 10 of Landsat 8 [40]. The present study used the GEE platform to invert the land-surface temperature using the single-window algorithm [41]:

$$LST=\frac{T_B}{1+\left(\frac{\lambda T_B}{\rho }\right)ln\epsilon }-273.15$$

(1)

where $LST$ is the surface temperature (°C), $T_B$ is the radiant brightness temperature $\left(\mathrm{K}\right)$, $\rho =hc/\delta =1.438\times {10}^{-2}\mathrm{mK}$, $\delta$ is Boltzmann’s constant, $\delta =1.38\times {10}^{-23} \mathrm{J}/\mathrm{K}$, $h$ is Planck’s constant, $h=6.626\times {10}^{-34}\mathrm{Js}$, $c$ is the speed of light, $c=2.998\times {10}^8 \mathrm{m}/\mathrm{s}$, $\epsilon$ is the surface specific emissivity, and 273.15 is the Kelvin and Celsius temperature-conversion constant. The wavelength of the emitted radiation $\lambda$ (11.45 μm for band 6 of Landsat 5, 10.90 μm for band 10 of Landsat 8), the radiant brightness temperature $T_B \left(°\mathrm{C}\right)$, and the surface specific emissivity $\epsilon$ are calculated as follows.

In both Landsat 8 and Landsat 5, $DN$ should be converted into spectral radiance, while the brightness temperature is obtained using the inverse function of Planck’s law:

$$L_{\lambda }=\mathrm{gain}\ast DN+\mathrm{bias}$$

(2)

$$T_B=\frac{K_2}{ln\left(1+K_1/L_{\lambda }\right)}$$

(3)

where $L_{\lambda }$ is the radiant brightness of the thermal infrared band (W·m⁻²·sr⁻¹·μm⁻¹), $T_B$ is the brightness temperature (°C), gain and bias are the gain and bias values of the thermal infrared band, respectively, and $K_1$ (W·m⁻²·sr⁻¹·μm⁻¹) and $K_2$ (K) are the calibration constants of the thermal infrared band. The parameters are listed in Supplementary Table S2 [42].

Surface emissivity is the ratio of radiation emitted by an object to that of a black body at the same temperature and wavelength. Its value is closely related to the material structure of the surface. The surface emissivity is calculated using the NDVI threshold method proposed by Sobrino et al. [43]. First, the NDVI value was calculated, and based on the NDVI threshold values of different land surfaces, the surfaces were classified and assigned emissivity values. NDVI < 0.2 indicates bare land and emissivity of 0.973, while 0.2 ≤ NDVI ≤ 0.5 indicates a mixed pixel of vegetation and bare land. The emissivity formula is as follows:

$$\epsilon =0.004p_z+0.986$$

(4)

$$p_z=\left(NDVI-NDV{I}_{min}\right)/\left(NDV{I}_{max}-NDV{I}_{min}\right)$$

(5)

where the $NDVI$ is calculated from the surface reflectance ρ of Landsat 8 bands 4 and 5 and Landsat 5 bands 3 and 4.

Thom’s DI was used to calculate environmental thermal comfort [44]:

$$DI=T_a-\left(0.55-0.0055R{H}_a\right)\left(T_a-14.5\right)$$

(6)

where $DI\left(°\mathrm{C}\right)$ is the thermal discomfort index, $T_a$ is the ambient dry bulb temperature $\left(°\mathrm{C}\right)$, which was replaced by the surface temperature in this study [40], and $R{H}_a$ is the relative humidity (%).

Table 1. Classification of human thermal comfort levels.

Range (°C)	Discomfort Index Classification
>32	State of medical emergency
29–32	Everyone feels severe stress
27–29	Most of population suffers discomfort
24–27	Over 50% population feels discomfort
21–24	Under 50% population feels discomfort
<21	No discomfort

lists the response of thermal environment to human health risk under different DI values [40].

3.1.2. Exposure—Human-Development Footprint

The human footprint model constructed by Huang et al. [45] was used to reflect the risk index of urban development on the thermal environment. Considering the rapid and concentrated urbanisation of Fuzhou, road network density, land development intensity, night-time light index, and population density were selected to calculate the human footprint index for Fuzhou. Data from 2005 and 2020 were used to reflect the development of the index:

$$\mathrm{HFI}=A+LULC+PD+NTL$$

(7)

where HFI is the human footprint index, A is the road density obtained by calculating the weighted sum of Euclidean distances for different road grades (highways, railways, and national roads: 0.87, provincial roads: 0.8, urban primary roads: 0.53, urban secondary and county roads: 0.2, other roads: 0.1) [46], LULC is the land-use development intensity (building: 1, farmland and unused land: 0.8, other ecological lands: 0), PD is the population density, and NTL is the night-time light index. Each indicator was normalised using the interquartile range normalisation method to calculate the human footprint index, eliminating differences in dimensions between variables.

3.1.3. Vulnerability—Natural Geographic Risk

Natural ecological conditions are the fundamental contributors to thermal environmental risk [9]. Based on existing research findings and in conjunction with the actual situation in Fuzhou, we focused on assessing the vulnerability of the thermal environmental risk in terms of topographical conditions, land-use attributes, and natural climate. Elevation and slope were selected as evaluation indicators for topographical conditions, vegetation coverage and land use were selected as the indicators for land-use attributes, and rainfall, wind speed, and the dryness index were selected as the indicators for natural climate. After normalisation, each indicator was weighted by expert scoring and calculated (

Table 2. Evaluation system of thermal environmental vulnerability indicators.

Goal Layer	Criterion Layer	Indicator Layer	Indicator Weight	Unit	Elaboration of Indicator	Trend
Vulnerability to thermal environmental risks	Topographical conditions	Elevation	0.1209	m	Temperature decreases with increasing altitude; the higher the altitude, the lower the thermal risk vulnerability index.	−
		Slope	0.1991	°	The greater the slope, the more prone it is to various types of geological hazards and the greater the thermal risk vulnerability index.	+
	Site attributes	Land use	0.0771	—	The greater the intensity of land-use development, the greater the thermal environmental vulnerability.	+
		NDVI	0.1211	%	Vegetation cover can mitigate thermal risk; the higher the vegetation cover, the lower the thermal risk vulnerability.	−
	Natural climate	Rainfall	0.1899	mm	Rainwater volatilisation and heat absorption can provide cooling effect, the higher the rainfall, the lower the vulnerability.	−
		Wind speed	0.1381	m/s	High wind speed is conducive to the realisation of hot airflow, thus reducing thermal environmental vulnerability.	−
		Dryness index	0.1537	—	A high dryness index increases environmental comfort, thus reducing thermal vulnerability.	−

). Several methods were used, including inviting 19 experts from related fields, such as climatology, landscape geography, and ecology, to rate the importance of each indicator, conducting group strategy calculations in YAAHP 10.3 software to determine indicator weights and passing consistency testing.

3.1.4. Assessment Methodology

Based on the evaluation results of the sources of danger, exposure, and vulnerability, these were divided into six levels and assigned values (

Table 3. Evaluation hierarchy of thermal environmental risk indicators.

Indicators	Weight	Grade
		10	8	6	4	2	0
Hazards	0.363	>32	29–32	27–29	24–27	21–24	<21
Exposure	0.304	>1.5	1–1.5	0.5–1	0–0.5	−0.5–0	<−0.5
Vulnerability	0.243	>−0.3	−0.35–−0.3	−0.4–0.35	−0.45–0.4	−0.5–0.45	<−0.5

). Using ArcGIS 10.5, the weights of each indicator were overlaid and calculated, resulting in the thermal environmental risk assessment for Fuzhou. To facilitate the proposal of thermal environmental risk optimisation strategies, we divided the risk assessment results into five levels: low, relatively low, medium, relatively high, and high. Standard deviational ellipse (SDE) analysis was used to study the spatiotemporal evolution characteristics and direction of the three assessment dimensions and the final thermal environmental risk. The calculation methods are shown in Supplementary Note S1.

3.2. Thermal Environment Network Construction

3.2.1. Screening of Ecological Sources

Ecological sources play a crucial role in maintaining regional ecological security [47] and can serve the function of cooling and temperature reduction [28]. In the present study, to fully exploit the function of the thermal risk sources, areas with low risk were extracted as ecological source areas. Moreover, considering the perspectives of the International Union for Conservation of Nature, ecological source areas were selected based on area and connectivity founded upon the thermal environmental risk assessment. The scale of the source area is a necessary condition to maintain its ecological function. However, landscape connectivity is an indicator of the degree to which energy movement is promoted or hindered between patches in source areas, and its level reflects the significance of the patches in the regional ecological network. This study employed Conefor Sensinode 2.6 software to calculate the potential connectivity (PC, Equation (8)) and overall connectivity (ICC, Equation (9)) [48] of each source area. With a connectivity probability of 0.5 and connectivity distance threshold of 1000 m and considering the basic situation of Fuzhou, patches > 0.5 km² with connectivity > 0.1 were selected as the ultimate ecological source areas.

$$PC=\frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{a}_i\times a_j\times p_{ij}^{\ast }}{A_L^2}.$$

(8)

$$IIC=\frac{{\sum }_{i=1}^n{\sum }_{j=1}^n\frac{a_i\times a_j}{1+n{l}_{ij}}}{A_L^2}n$$

(9)

where $n$ is the number of patches, ${\mathrm{a}}_{\mathrm{i}}, {\mathrm{a}}_{\mathrm{j}}$ is the area of patches $\mathrm{i}, \mathrm{j}$, respectively, $p_{ij}^{*}$ is the maximum likelihood of a species moving between patches $i, j$ (one-way only), $n{l}_{ij}$ is the number of nearest connections between patches $i, j$, and $A_L$ is the overall area of the landscape.

3.2.2. MCR-Based Network Construction

Ecological corridors represent paths of least resistance connecting various source areas, playing a crucial role in material circulation and energy flow. In the heat risk network, corridors can simulate the potential transmission of cold air. By enhancing the connectivity between source areas through these networks, the heat risks in specific areas can be effectively mitigated [49]. In this study, land use, elevation, slope, evaporation, NDVI, precipitation, population density, and road network density were selected to construct a comprehensive resistance surface [50] (

Table 4. Integrated resistance surface evaluation indicator system.

Resistance Factor	Weight	Grade	Resistance Value	Trend	Resistance Factor	Weight	Grade	Resistance Value	Trend
Elevation (m)	0.11	<200	1	+	Land use	0.12	Water	1	+
		200–400	3				Forests	3
		400–800	5				Farmland	5
		800–1000	7				Unused land	7
		>1000	9				Building	9
Slope (°)	0.15	<8°	1	+	Population density	0.17	>50	1	+
		8°–15°	3				10–50	3
		15°–25°	5				5–10	5
		25°–35°	7				1–5	7
		>35°	9				<1	9
Evaporation (mm)	0.12	>1400	1	+	Precipitation (mm)	0.11	>2000	1	+
		1300–1400	3				1900–2000	3
		1200–1300	5				1800–1900	5
		1100–1200	7				1700–1800	7
		<1100	9				<1700	9
NDVI	0.11	>0.9	1	+	Road network density	0.11	<50	1	+
		0.75–0.9	3				50–100	3
		0.5–0.75	5				100–150	5
		0.25–0.5	7				150–200	7
		<0.25	9				>200	9

). The Linkage Mapper tool was used to extract both the ecological and potential ecological corridors. The Centrality Mapper module was used to identify centrality, quantifying the importance of different corridors, which are further classified into levels 1, 2, and 3 using the natural breakpoint method [51].

3.2.3. Ecological Node Extraction

When constructing regional heat risk networks, in addition to the planar source areas and linear corridors, it is necessary to identify important ecological nodes in the network to help maintain its stability. Additionally, by optimising these nodes, the function of the ecological network can be better realised, reducing the urban heat environmental risk.

Ecological pinch points represent areas in the corridors with the highest current density and are important nodes in ecological grid construction. In this study, the corridor weighted cost distance was set to 2000 m [52], and the current density distribution map was obtained by iteratively calculating the input currents to the source in a many-to-one mode sequence. Using the natural breakpoint method, the current intensity was divided into three levels and categorised from high to low as Level 1, 2, and 3 protection areas. The Level 1 protection area was extracted as the pinch-point area.

Ecological barrier points refer to areas in the corridors with significant resistance to material and energy flow, hindering connectivity between source areas. These points typically represent areas significantly disturbed by human activities, such as disaster spots, factories, and roads. We used Linkage Mapper to identify barrier points [53] and obtain improvement coefficients. The higher the improvement coefficient, the greater the degree to which network connectivity can be restored following remediation. Using the natural breakpoint method, the improvement coefficient was divided into three levels, categorised from high to low as an extremely important improvement area, important improvement area, and general improvement area. The extremely important improvement area was extracted as the choke-point area.

Stepping stones effectively increase the connectivity between source area patches and enhance the energy flow between sources. In this study, the intersections between ecological corridors were considered stepping stones.

4. Results

4.1. Thermal Environmental Hazards, Exposure, and Vulnerability

In 2020, the overall thermal comfort risk level in Fuzhou was relatively low (Figure 4a,b). Most areas of the central city were in the Level 3 risk zone, with only assigned Level 4 or Level 5 risks. There was virtually no risk of thermal discomfort in the forested areas on the outskirts of the city. However, considering the change in the DI from 2005 to 2020, there is a clear trend of increasing thermal discomfort risk in Fuzhou (Figure 4c), with an average increase of 0.55 °C. With urbanisation, the DI in the areas surrounding the central city and some coastal regions increased by 4–6 °C, significantly impacting on the health of local residents. Moreover, although the upper reaches of the Min River currently have a relatively low thermal comfort risk, the change in the DI over these 15 years should also be considered. The standardised ellipse results of the UTC over the last two years (Figure 4c) reveal that the thermal comfort risk in Fuzhou is gradually converging toward the central city and extending in both directions along the Min River.

From the perspective of the human-development footprint, Fuzhou has experienced rapid development in the past 15 years (Figure 4d,e). The scale of the central city of Fuzhou continues to expand and gradually develop toward the coastal areas. As the western region is dominated by mountainous woodlands, the overall elevation is high, and the terrain is undulating, with an overall slow development trend. For the old city areas that were already at a high level of development in 2005, the development space is limited, with a relatively consistent overall development level (Figure 4f). Meanwhile, owing to the influence of the ‘Park City’ policy, government renovations and updates in the old city areas have reduced the intensity of land development and increased the comprehensive value of urban sustainable development. Furthermore, with the development of road traffic, the mountainous counties in the west have certain development potential. Overall, although the eastern coastal areas of Fuzhou are developing rapidly, they rely on the development of the central city. Therefore, the overall development trend continues to shift towards the central city (Figure 4f); however, the direction has remained consistently in the north-west–south-east direction.

Changes in the thermal environmental vulnerability of Fuzhou have primarily occurred in the central city and coastal areas (Figure 4g,h). The overall vulnerability shows an upward trend, with the south-eastern coastal region exhibiting the most noticeable increase. In contrast, the vulnerability in the northwestern mountainous areas remains stable (Figure 4i). Notably, the Min River—the most crucial river resource in Fuzhou—has been affected by urban development. The thermal environmental vulnerability of the entire riverbank has been damaged to varying degrees. Based on the standardised ellipse results from 2005 and 2020, the trend and direction of vulnerability changes in Fuzhou have not significantly changed, maintaining overall coordinated development.

4.2. Comprehensive Thermal Environment Risk

The results of the thermal risk assessment of Fuzhou, which synthesised hazards, exposure, and vulnerability, revealed a current prominent risk in Fuzhou (Figure 5a,b). Areas with high thermal environmental risk values account for 12.26%, concentrated primarily in the main city and coastal areas. In contrast, low-risk areas make up 28.02% and are primarily located in the western forested regions where human activity is limited. This portion of the forested areas, despite having a low risk value, was similarly disturbed to some extent by urban development and will require focused protection in subsequent development. Considering the changes in the thermal environmental risk levels over 15 years, the overall risk in Fuzhou in 2020 has significantly increased compared with 2005, with the average risk value increasing by 0.70 (Figure 5c). In 2020, there were 3262.54 km² areas in Fuzhou classified as low risk, a decrease of 1712.92 km² compared with 2005, while high-risk areas increased by 1044.29 km² compared with 2005 (Figure 5d,e). All areas under other risk levels in 2020 increased compared with 2005: areas with relatively low, medium, and relatively high risks increased by 110.27, 218.76, and 342.19 km², respectively. Furthermore, based on the results of the transition matrix, most of the changes in thermal environmental risk trends were from low to high risk (Figure 5f).

4.3. Screening of Thermal Risk Source Areas

After filtering by area and connectivity, 30 ecological source areas were identified in 2005, covering an area of 4642.73 km² (

Table 5. Statistical results of ecological source areas in Fuzhou in 2005 and 2020 based on risk assessments.

Year	Level 1 Ecological Source Area			Level 1 Ecological Source Area			Level 1 Ecological Source Area
	Number	Area (km2)	Percentage (%)	Number	Area (km2)	Percentage (%)	Number	Area (km2)	Percentage (%)
2005	5	3853.59	83.00%	9	616.97	13.29%	16	172.17	3.71%
2020	3	1710.75	56.64%	18	894.21	29.61%	33	415.32	13.75%

): five Level 1 ecological source areas, nine Level 2 ecological source areas, and sixteen Level 3 ecological source areas. In 2020, although the number of ecological source areas increased to 54 (Table 5), the area decreased by 1622.46 km². Notably, the area of the Level 1 ecological source areas significantly decreased, totalling 2142.84 km², and their proportion declined substantially.

From the perspective of the distribution of source levels, in 2005, Level 1 ecological source areas are distributed primarily in the western mountainous region (Figure 6a). Owing to their excellent ecological conditions, they are crucial in regulating the thermal environmental risks in Fuzhou. In 2020, some of the Level 1 source areas in the western mountains changed to Level 2 or Level 3 source areas. All source areas in the eastern coastal region became Level 3 (Figure 6b), indicating that the connectivity of ecological source areas was disrupted. Furthermore, owing to urban development, the number of ecological source areas in the coastal region decreased in 2020. By 2020, the ecological source areas located in the southern coastal region of Fuzhou and the Pingtan Island area no longer functioned as source areas.

4.4. Thermal Environment Ecological Network Construction

In 2005, 64 ecological corridors were extracted with a total length of 597.83 km. After eliminating redundant and excessively short corridors, 45 were retained (Figure 7a): 6 Level 1 corridors, 16 Level 2 corridors, and 23 Level 3 corridors (Figure 7a). In 2020, owing to the reduction in source areas, more corridors were required to construct the entire ecological network. As a result, there were 124 corridors in 2020 with a total length of 762.56 km. After screening, 54 Level 3 corridors, 26 Level 2 corridors, and 4 Level 1 corridors were retained (Figure 7d). Although the number and length of corridors increased in 2020, the proportions of Level 1 and Level 2 corridors decreased. Moreover, the average length of the 2020 corridors was 4.69 km shorter than the 2005 corridors.

Our identification of ecological pinch points in the corridor yielded a maximum current value of 0.192 in 2005, which had decreased to 0.156 in 2020 (Figure 7b). In 2020, the Level 1, Level 2, and Level 3 protected areas were 139.93 km² (9.90%), 643.59 km² (45.52%), and 630.23 km² (44.58%), respectively (Figure 7e). The proportions of the Level 1 and Level 2 protection areas in 2020 had decreased compared with those in 2005. The results of the corridor improvement coefficients show a total improvement zone area of 1969.59 km² in 2020, representing a 583.31 km² increase from 2005 (Figure 7c). In 2020, the proportion of the extremely important improvement area had decreased by 3.87% compared with that in 2005; however, the area increased by 85.66 km². The area and proportions of the important improvement area and general improvement area in 2020 were greater than those in 2005 (Figure 7f).

4.5. Node Identification for Ecological Networks

Pinch points are key nodes with high current values in the heat flow process and are primary areas for ecological protection; therefore, damaging pinch points may significantly impact the regional thermal environmental risk. The first-level ecological protection areas can be regarded as ecological pinch points, which are crucial regions in the thermal risk network pattern that require priority protection to maintain the connectivity of the ecological corridors. The results show that pinch points were primarily located at both ends and in the middle of the corridors. In 2005, there were 34 pinch points (Figure 8a), which had increased to 76 by 2020 (Figure 8d). The land types occupied by pinch points mainly included forests, water, and grassland.

Barrier points are areas that hinder energy flow. Extremely important improvement areas were identified as ecological barrier points. In 2005, 45 barrier points were identified (Figure 8b), increasing to 110 by 2020 (Figure 8e). The primary land types occupied by barrier points were buildings and farmlands. Lastly, by identifying the intersection points of the corridors, 12 stepping stones in 2005 and 2020 were found (Figure 8c,f).

5. Discussion

This study refined several commonly used methods to assess the thermal environment, integrating socio-ecological benefits to conduct risk assessments and relying on the results to construct a thermal risk ecological network. However, with the progression of urbanisation, the thermal environmental risk in Fuzhou continues to escalate, necessitating further optimisation of the ecological network’s functionality and structure. Therefore, in the context of existing research, the optimisation strategies for thermal comfort in Fuzhou in two respects were explored further.

5.1. Thermal Environmental Risk Optimisation under Socio-Ecological Trade-Offs

5.1.1. Spatial Correlation Analysis of Thermal Environmental Risk

Urbanisation leads to the formation of UHI, increasing thermal environmental risks [54], and natural geographical conditions are the fundamental causes of climatic differences [55]. Therefore, the MGWR method was used to explore the impact of urban socio-ecological factors on environmental thermal comfort [56]. Research methods are shown in Supplementary Note S2. The spatial distribution characteristics of the regression coefficients of each explanatory variable calculated using the MGWR model are illustrated in Figure 9.

Social development and thermal comfort were positively correlated overall. The regression coefficient for 2005 and 2020 ranged between −3.855 and 3.538 (Figure 9a) and −2.525 and 2.699 (Figure 9b), respectively. Moreover, the R² in 2020 had increased by 0.07 compared to 2005 (

Table 6. Parameters related to the results of MGWR calculations.

Year	Indicators	R2	Adj.R2	AICc	AIC	RSS
2005	Exposure	0.693	0.671	21,753.488	21,638.337	3709.958
	Vulnerability	0.705	0.685	21,192.971	21,089.927	3570.227
2020	Exposure	0.763	0.747	18,564.149	18,455.667	2862.112
	Vulnerability	0.806	0.793	16,133.122	16,030.760	2350.376

). This indicates that with rapid urbanisation, the risk of the urban thermal environment increases. Zhang et al. [57] have shown an increasing trend in the correlation between population and UHI. The authors posited that population density is an indirect factor for UHI, while the direct factor might be the comprehensive development of society, as confirmed in this study. Based on the spatial distribution characteristics, the impact of socioeconomic factors on thermal comfort in 2005 was unstable, with a clear two-tiered spatial pattern. By 2020, areas with a positive correlation between socioeconomic factors and thermal comfort increased, while the correlation tended to stabilise. This is primarily because in 2005, Fuzhou was in its early stages of development, with significant urban–rural differences, and the thermal comfort changes in remote areas were largely influenced by the development of the central city [58]. However, with societal development, some of the forested areas were also developed, directly impacting environmental thermal comfort [59]. In addition, unlike the study by Zhao et al., this study suggests a stronger correlation between the comfort of the thermal environment in suburban areas and the intensity of social development. The thermal environment in suburban areas is more susceptible to damage from urban development. This may be due to the particularity of mountainous coastal areas, where the development of suburban areas can cause damage to the surrounding ecological space, leading to a decrease in thermal comfort [60].

Unlike in the case of social explosion, the correlation between natural geography and thermal comfort is strong. This is consistent with most current research findings, e.g., Hou ranked the determining factors of urban thermal environment and found that the NDVI has a greater impact on thermal environment than urban morphology and human activity levels [61]. The regression coefficient for 2005 ranged from −1.120 to 6.819 (Figure 9c). The regression coefficients in the south-east and certain eastern coastal areas were higher than in the western mountainous regions due to the more variable climates in coastal areas [62], resulting in weaker correlations with natural geographical factors. The R2 for 2020 increased to 0.806, with an overall strong correlation. Although the coastal areas retained a relatively weak correlation, the two-tiered phenomenon observed in 2005 was mitigated. The regression coefficients of 2020 were higher than those of 2005 (Figure 9d), indicating that natural geographical conditions have a more pronounced impact on comfort. This phenomenon will further exacerbate the instability of environmental thermal comfort, affecting the health of urban residents [63].

5.1.2. Strategies for Optimising Thermal Environmental Risks

Overall, environmental thermal comfort is influenced by social and ecological factors, which showed an increasing trend over time, consistent with previous findings [55]. While Zheng et al. [64] reported that large-scale urbanisation exacerbates thermal environmental risks, we did not observe significant changes in the thermal environmental risks in the central urban area from 2005 to 2020; however, the surrounding development areas and some rural areas had higher thermal environmental risks. Hence, when urban development reaches a certain stage, the impact of urbanisation on thermal environmental risks is reduced. Therefore, when facing a thermal environment risk, it is necessary to integrate the dual factors of social development and environmental protection. This approach can prevent urban development from falling into ecological protection or extreme economic development situations [24], achieving sustainable development of the urban climate environment. However, social development does not only directly impact thermal comfort but also the natural environment, which indirectly exacerbates thermal risks [65]. Accordingly, we propose thermal environmental risk optimisation strategies for three areas: ‘central city’, ‘suburban areas’, and ‘forest under-space’.

The central city area has the highest thermal environmental risk, with high natural vulnerability and social exposure. However, the current development of this area is relatively stable. Hence, subsequent development in the central city area should focus on ecological restoration and environmental micro-renovation, e.g., increasing greenery through community gardens, rooftop gardens, and urban green walls, which can create cooling ‘green islands’ within the main city area [66]. Urban green spaces of moderate size and relatively complex shapes have better cooling effects [67]. In addition, for high-density urban areas in the centre, it is also necessary to focus on the comfort of the internal thermal environment of buildings. In this process, attention should be paid to energy consumption and its impact on the external environment [68].

Suburban areas around the main city are high value in terms of economic development, and the ecological conditions of this region have not been excessively damaged. These areas are therefore optimal for scientific management and development. However, urban development boundaries must be tightly regulated. Otherwise, land overuse and disorderly expansion will further intensify thermal environmental risks. Additionally, in areas with good habitat quality, ecological corridors can be established to strengthen the connection between external source areas and the central city, maintaining energy transmission inside and out to achieve cooling effects [69].

Wong et al. [70] purported that expanding green infrastructure is the most effective measure to address urban heat risk and that most of the forest under-space in Fuzhou constitutes an important cooling source on the periphery of its main urban area. The development cost of this region is relatively high and its benefits are low, making it more suitable for maximising ecological values.

5.2. Optimisation Strategies for Thermal Environmental Ecological Networks

5.2.1. Network Pattern Optimisation Based on Urban Green Infrastructure

Based on the results of the heat risk network construction in Fuzhou in 2020, we found that owing to urban development, the central urban area and the south-eastern coastal region lack network connectivity, potentially hindering the ecological network from fully functioning, particularly in the central urban area with severe heat risk. Compared with the existing cold conduction network, the network constructed in this study lacks associations with the internal sources of the central urban area [33]. In a study by Lin et al. [71], a green infrastructure network for the main urban area of Fuzhou was constructed, with source areas primarily comprising urban parks within the main city, meeting residents’ living, leisure, and recreational needs. Hence, our research references their methods and uses MSPA to supplement ecological source areas in Fuzhou. Zhao et al. [72] demonstrated the important relationship between core areas and urban thermal environments. Therefore, based on comprehensive social and ecological analysis, we extracted the core area as a source area supplement, which will benefit the stability of the overall cooling network structure. This network construction method that combines social, ecological, and spatial forms will further enrich cooling networks. After screening for area and connectivity, we added five ecological source areas to the central urban region and four to the south-eastern coastal region, increasing the area by 49.67 km² (Figure 10a). Using the optimised source areas, we identified 51 Level 3 corridors, 50 Level 2 corridors, and 9 Level 1 corridors (Figure 10b–d). Meanwhile, through ecological node identification, we found 85 ecological pinch points, 101 barriers, and 24 stepping stones (Figure 10e).

After optimisation, the distribution of the ecological network became more uniform, with improved integrity and connectivity. This optimisation can better alleviate heat environmental risks in rapidly developing areas, such as the central urban area and south-eastern coastal region [73]. Additionally, the proportion of secondary corridors in the optimised network has significantly increased and is primarily distributed in the central region, indicating that the added source areas in the central city can enhance the stability of the overall network. However, the land in the central urban area has high economic value, and excessive ecological construction might intensify social issues [13]. The increased number of barriers in the network directly reflects this phenomenon. Therefore, corridor construction in this region requires more comprehensive assessment and management, utilising compound structures or fragmented land spaces to establish links between the main urban area and external ecological source areas. Corridors passing through urban areas have a more pronounced improvement effect on the overall thermal environment than those on the urban periphery [74]. Compared to the network constructed by Liu et al., the network in this study does not include the Minjiang River as a source area, primarily due to the influence of social development [33]. The Minjiang River has certain thermal environmental risks that prevent it from serving as a source area. Compared with previous studies, the network in this study combines ecological and social benefits, balances the contradiction between their development, and improves the reliability and ease of implementation of the network. Finally, although nine ecological source areas were added to the network, their functionality was relatively lacking. The ecological source areas added within the main city have a high heat risk owing to surrounding construction [75], and their ecological functions can be improved in future developments through optimising plant diversity, increasing green coverage and ecological engineering design [8]. The source areas added in the coastal region are constrained primarily by connectivity, which can be optimised by increasing their scale and setting stepping stones [73].

5.2.2. Network Structure Optimisation Based on Node Characteristics

For ecological pinch points, the strategy should prioritise ecological conservation and use human intervention as a supplementary method. If the current land-use type of the pinch point is for ecological purposes, stricter controls should be in place, establishing ecological ‘red lines’ to avoid artificial exploitation. If the main coverage type is farmland, ecological protection forest belts can be established around the farmland to protect the agricultural ecosystem chain, enhancing the ecological function of the pinch point. For ecological obstacle points, the strategy should prioritise human intervention, with natural recovery as a supplementary approach. The restoration of these obstacle points requires a balance between ecological and economic values. Planting economically valuable crops, such as lychee and longan, can reduce ecological resistance while maximising social and ecological benefits [76]. In addition, small-scale green spaces offer the same cooling effect [77]. Initiating afforestation in areas with high resistance in the main urban area, which has high heat risks and low green coverage, is suggested. Lastly, stepping stones play an essential role as transit points in the network. By attracting ecological flows and material exchanges, a potential stepping-stone network can be formed [78]. Stepping stones can also provide additional pathways for heat flow, resulting in improved cooling effects [73]. During the construction of stepping stones, the node regions can be expanded, prioritising the preservation of key stepping-stone areas’ original ecology. Given the rich natural resources in Fuzhou, some stepping-stone areas can be transformed into multifunctional regions such as forest farms or forest parks.

5.3. Limitations

The study model was designed to provide a comprehensive assessment of heat environmental risks based on socio-ecological benefits. However, there are certain limitations. First, in our model, hazards, exposure, and vulnerability were treated as parallel relationships, possibly overlooking trade-offs and synergies between ecology and society. Second, this study only considered the supply effects of the heat environment. As different regions have varying population needs, spatial differences in supply and demand could emerge. Therefore, future studies should incorporate human needs. Third, the corridors constructed in this study used areas with low heat environmental risks as sources, resulting in a network integrating regional cool island spaces. Future studies should combine high-risk and low-risk areas to construct different spatial network patterns using the source–sink theory. Finally, the ecological network constructed by this research institute can only address the thermal environmental risks in Fuzhou. In the future, strategies for addressing the thermal environmental risks in Fuzhou can be proposed on a larger scale, taking into account the surrounding areas.

6. Conclusions

This study evaluated the environmental risk caused by heat in Fuzhou on three dimensions: hazards, exposure, and vulnerability. Temperature and humidity were combined to calculate environmental thermal comfort and the human footprint index was applied for exposure and natural geographical conditions for vulnerability. Following risk categorisation, ecological source areas were extracted, and an ecological network was established using a linkage map. The analysis included ecological pinch points, barriers, and stepping stones within the network. Finally, optimisation strategies for thermal comfort in Fuzhou were proposed from two angles. The primary outcomes of this study were threefold. (1) The central urban area of Fuzhou and other study areas have prominent heat environmental risks, with annual increases. The area with a high risk level in 2020 had exhibited a substantial increase from that in 2005. Thus, to optimise thermal environmental risks, the central urban area should focus on ecological restoration and urban micro-renewal. Moreover, suburban areas must control excessive urban expansion, while understory areas should prioritise ecological protection. (2) Influenced by the increased thermal environment risk, the source area in 2020 was reduced compared with that in 2005, and source areas were unevenly distributed, while the south-eastern coastal area and central city lacked vital cooling sources. Following optimisation, nine ecological source areas that were more evenly distributed were added. This improvement in integrity and connectivity is more conducive to mitigating the thermal environmental risks in Fuzhou. (3) The stability and functionality of the constructed cooling network for Fuzhou City in 2020 was reduced compared with that in 2005, primarily reflected by the shortened average corridor length, reduced ecological pinch points, and increased choke points. Hence, intervention in pinch points must be reduced, the repair of choke points strengthened, and the protection of the stepping-stone area expanded. This study evaluates urban thermal environmental risks from a social ecological perspective and identifies corresponding cooling networks. This method has certain universality and can provide reference for the formulation of urban climate development policies.