Construction of a Cold Island Spatial Pattern from the Perspective of Landscape Connectivity to Alleviate the Urban Heat Island Effect

Abstract

This study presents an innovative approach to mitigating the urban heat island (UHI) effect by constructing a cold island spatial pattern (CSP) from the perspective of landscape connectivity, integrating three-dimensional (3D) urban morphology and meteorological factors for the first time. Unlike traditional studies that focus on isolated patches or single-city scales, we propose a hierarchical framework for urban agglomerations, combining morphological spatial pattern analysis (MSPA), landscape connectivity assessment, and circuit theory to a construct CSP at the scale of urban agglomeration. By incorporating wind environment data and 3D building features (e.g., height, density) into the resistance surface, we enhance the accuracy of cooling network identification, revealing 39 cold island sources, 89 cooling corridors, and optimal corridor widths (600 m) in the Changsha–Zhuzhou–Xiangtan urban agglomeration (CZXUA). Ultimately, a three-tiered heat island mitigation framework for urban agglomerations was established based on the CSP. This study offers an innovative perspective on urban climate adaptability planning within the context of contemporary urbanization. Our methodology and findings provide critical insights for future studies to integrate multiscale, multidimensional, and climate-adaptive approaches in urban thermal environment governance, fostering sustainable urbanization under escalating climate challenges.

1. Introduction

Rapid urbanization has significantly altered the composition of the underlying urban surface, with ecological areas such as woodlands and water bodies increasingly being supplanted by construction for housing and transportation [1]. The blue–green space (BGS), primarily comprising woodlands, water bodies, and wetlands, exerts a substantial cooling and humidifying effect [2,3,4]. On the one hand, the expansion of urbanized areas leads to heightened heat generation, and the thermal impact stemming from the metabolic processes of the substantially enlarged population merits consideration. On the other hand, the cooling and humidifying effect of encroached water bodies and vegetation is significantly reduced. Urbanization is leading to decreasing BGS in cities, while the urban heat island (UHI) effect is gradually increasing. Thus, anthropogenic heat emissions and subsurface alterations during urbanization are the primary drivers of the UHI effect [5,6,7]. The UHI effect denotes a condition where temperatures within urban areas exceed those of the surrounding rural regions [8,9,10,11,12]. British scholar Howard was the first to document this phenomenon, noting that temperatures in urban London were markedly [4] higher compared to those in rural areas. Later, in 1958, Manley formally characterized this phenomenon as the “urban heat island effect” [13]. The UHI effect can exert a variety of detrimental effects on both habitat and human health [8,14,15,16]. Therefore, the question of how to alleviate the urban heat island effect has become a significant subject of focus among scholars [17,18,19].

Currently, research on mitigating the UHI effect in urban planning primarily concentrates on the cooling and humidifying effects of BGS at the macroscale [20,21] and numerical simulations utilizing software like ENVI-met at the microscale [22,23]. The latter can effectively optimize green space layout and building configuration and reduce local air temperature at the community scale. However, computational complexity and resolution limit its application at larger scales (e.g., city) [24], though it has a minimal impact on the city-wide amelioration of the UHI effect. A multitude of studies attest to the significant role of urban BGS in mitigating the UHI effect [3,25,26,27,28]. Most of the BGSs in a city are “cold islands” [29,30]. Cold islands lower the temperature in the surrounding area [31]. Nonetheless, constrained by planning policies and other factors, the urban BGS cannot expand indefinitely. Accordingly, it is essential to maximize the cooling potential of existing urban cold islands [32].

However, current research on urban cold islands predominantly concentrates on statistical analyses at the macrolevel or at the simple patch level [33]. By overlooking the connectivity within the spatial configuration of cold islands and the significance of landscape networks, these research findings often struggle to achieve effective practical application. High-density urban development and spatial expansion contribute to the ongoing degradation of the urban thermal environment [34]. Particularly as urban agglomerations evolve, an increasing number of small urban heat islands coalesce into larger expanses, amplifying the UHI effect [35]. In this context, establishing a comprehensive and strategically designed cooling network can markedly diminish the persistent adverse effects of high-intensity urban development. Furthermore, interconnected cold islands possess superior cooling capacities compared to isolated patches [36], which not only enhance the quality of life for urban dwellers but also promote equitable allocation of ecological resources.

In recent years, scholars have begun to use landscape ecology methods such as MSPA and circuit theory to construct urban cold island networks from a landscape ecology perspective to alleviate the UHI effect [4,33,37]. Liu et al. used MSPA to extract high-temperature patches in Zhuzhou and formulated graded conservation measures for UHI patches according to their importance based on landscape connectivity [38]. Qian et al. developed a faceted-linear-point regional cooling network, considering the connectivity and network structure of cold islands in Wuhan [4]. Although previous research has made significant contributions to the construction of cooling networks, several shortcomings remain. Firstly, the cooling network is inevitably affected by wind direction and wind speed; however, none of the existing studies have considered the effect of the wind environment where the cooling network is located. Concurrently, in constructing the resistance surface, the majority of previous studies have focused solely on two-dimensional and natural factors, overlooking the impact of three-dimensional elements and socio-economic factors, such as building height. Secondly, the determination of resistance factor weights in existing studies has predominantly employed subjective methodologies, such as the analytic hierarchy process (AHP) [4,33]. Thirdly, there is a paucity of strategic measures proposed for the targeted integration of key regions into specific spatial contexts within the construction of cold island networks. Furthermore, existing studies predominantly focus on the scale of individual cities and small towns, with less emphasis on urban agglomerations. Unlike isolated cities, urban agglomerations exhibit interconnected heat island networks due to cross-boundary heat exchange (e.g., thermal radiation from transportation corridors, industrial clusters) and scaling effects where localized UHIs merge into regional thermal anomalies. This spatial complexity necessitates mitigation strategies beyond single-city approaches, addressing both intra-city hotspots and inter-city thermal interactions [39]. Therefore, constructing CSPs to mitigate the UHI effect at the urban agglomeration scale is significant for urban climate adaptation planning.

To address the aforementioned shortcomings, this study focused on the Changsha–Zhuzhou–Xiangtan urban agglomeration (CZXUA) as the subject of investigation. Based on assessments of the wind environment of urban agglomerations, we explored a framework for developing a cold island spatial pattern (CSP) within urban agglomerations, comprising cold islands, resistance surface, cooling corridor networks, and critical nodes. A CSP refers to a network framework designed to mitigate the UHI effect. It comprises interconnected cold island sources (large green/blue spaces with significant cooling potential), cooling corridors, cooling nodes (the area where cold air must pass through or frequently flow), and heating nodes (an area where cold airflow is impeded). The CZXUA is among China’s most prominent urban agglomerations [40,41], experiencing frequent episodes of hot weather and heatwaves during the summer months [42]. Hence, there is an imperative to establish a regional cooling network aimed at mitigating the UHI effect. The objectives of this research were to (1) determine the classification of UHI intensity through LST inversion derived from remote sensing imagery; (2) utilize MSPA–circuit theory and additional methodologies to identify and establish a cooling corridor network; and (3) propose a foundational framework for the development of CSPs in urban agglomerations. This study creatively constructs a CSP framework for urban agglomerations that integrates the wind environment and three-dimensional elements and proposes three-level governance strategies to provide systematic methods and practical paths for cross-regional climate adaptation planning.

2. Study Area and Data Sources

2.1. Study Area

The CZXUA, situated in the east-central region of Hunan Province, encompasses three prefecture-level cities, five county-level cities, five county towns, and 165 organic towns [43] covering a total area of 28,087 km², with geographic coordinates ranging from 26°03′ N to 28°41′ N, and 111°53′ E to 114°15′ E [41] (Figure 1). The CZXUA represents a critical component of the urban agglomerations within the middle reaches of the Yangtze River and serves as the central growth pole for economic development in Hunan Province. In 2022, occupying 13.3% of the province’s land area and housing 25.7% of its population, the CZXUA contributed 41.7% of the total GDP of Hunan Province. The CZXUA is situated in an area characterized by hot summers and cold winters and experiences a subtropical monsoon humid climate, marked by low temperatures and substantial rainfall in spring, high temperatures and intense sunlight in summer, drought conditions with scant rainfall in autumn, and cold, damp winters. Frequent occurrences of extreme heat are observed in summer, with temperatures occasionally exceeding 40 °C. Notably, Changsha is recognized as one of China’s four “furnace” cities [44]. In recent years, the rapid development of the CZXUA has led to expanding urban agglomeration, exacerbating the UHI effect, which significantly impacts the health of urban dwellers and the sustainable development of the urban agglomeration. An immediate construction of a CSP is imperative to alleviate the UHI effect.

2.2. Data Sources

The data utilized in this study primarily consisted of Landsat 8 OLI/TIRS satellite digital products, land use data, meteorological data, administrative district data, road network data, and DEM, among others, with specific data types, resolutions, sources, and applications detailed in

Table 1. Data sources and uses.

Data	Reference Year	Resolution	Sources	Data Uses
Land use data	2020	30 m × 30 m	GlobeLand30, http://www.globallandcover.com/ [45]	MSPA, resistance surface construction
Meteorological data	2020	-	Meteorological Bureau of Changsha, Zhuzhou, Xiangtan Municipality	Wind environment assessment
Landsat 8 OLI_TIRS	2020	30 m × 30 m	https://earthexplorer.usgs.gov/ [46]	Retrieval of land surface temperature
Road network data	2020	-	From Bureau of Natural Resources and Planning of Changsha, Zhuzhou and Xiangtan	Resistance surface construction
DEM	2020	30 m × 30 m	https://www.gscloud.cn/ [47]	Resistance surface construction
NDVI	2020	30 m × 30 m	MODIS13A3 (https://modis.gsfc.nasa.gov/) [48]	Resistance surface construction
Building vector data	2020	-	(Che et al., 2024) “3D-GloBFP: the first global three-dimensional building footprint dataset” [49]	Resistance surface construction

. The projection for all spatial data has been standardized to the equal-area projection of WGS 1984 UTM Zone 49N to ensure the spatial consistency across multi-source data. Raster data of varying resolutions were resampled to a uniform resolution of 30 m × 30 m.

3. Methodology

Based on assessing the wind environment of urban agglomerations, this study introduces a comprehensive framework for the construction of CSP within urban agglomerations: “identification of cold island sources–construction of resistance surfaces–extraction of cooling corridors networks–identification of critical nodes”. The methodologies employed in this research primarily encompass MSPA, landscape connectivity analysis, and circuit theory, among others (See

Table 2. Construction framework of CSP.

Items	Analysis Methods	Indicator Items
Identification of cold island sources	MSPA; landscape connectivity analysis	The cold area of LST
Construction of resistance surfaces	SPCA	Natural and socio-economic factors
Extraction of cooling corridors networks	Circuit theory	Cold island surfaces; resistance surface
Identification of critical nodes	Circuit theory	Cooling corridors; cold island surfaces; resistance surface

for each step’s research methodology and indicator items). The detailed research framework is depicted in Figure 2.

3.1. Retrieval of Land Surface Temperature

At present, remote sensing monitoring has emerged as the predominant method for UHI research, attributed to its facile data acquisition and labor-efficiency. The most extensively utilized approaches for land surface temperature (LST) inversion from remote sensing data include the atmospheric correction method (based on the radiative transfer equation), single-window algorithms, and split-window algorithms. The USGS has implemented updated calibration parameters for Landsat 8. However, due to TIRS11 band uncertainties [37], this study employed the atmospheric correction method. The underlying principle of the atmospheric correction method entails estimating the impact of atmospheric thermal radiation on the surface, subsequently deducting this atmospheric influence from the total thermal radiation detected by satellite sensors to derive the surface thermal radiation intensity, and then converting this to the corresponding surface temperature [50]. The UHI effect is particularly pronounced during periods of hot weather in the summer, hence our selection of Landsat 8 remote sensing data for the study area from July to September, with cloud cover below 10%, for LST inversion. For LST inversion, considering data availability, the remote sensing data from 28 August 2020 was ultimately chosen, utilizing ENVI 5.3 for pre-processing, which encompassed radiometric calibration, atmospheric correction, and image cropping. The procedural steps of the atmospheric correction method comprise image radiometric calibration, estimation of vegetation cover, determination of surface-specific emissivity, calculation of blackbody radiance, and calculation of LST.

(a) Vegetation cover fraction (FVC) calculation:

$$\mathrm{FVC}={\displaystyle \frac{NDVI-{NDVI}_{soil}}{{NDVI}_{veg}-{NDVI}_{soil}}}$$

(1)

where FVC is vegetation coverage; NDVI_soli is the NDVI value of areas with no vegetation cover or completely bare soil; and NDVI_veg is the NDVI value of areas with complete vegetation cover.

(b) Surface-specific emissivity calculation:

ε = 0.004 ∗ FVC + 0.986

(2)

where ε is the surface specific emissivity and FVC is the vegetation cover [51].

(c) Blackbody thermal radiation brightness calculation:

Lλ = [ε ∗ B(TS) + (1 − ε) ∗ L↓] ∗ τ + L↑

(3)

where Lλ is the value of thermal infrared radiation brightness received by the sensor, ε is the surface-specific emissivity, TS is the true LST, B(TS) is the blackbody thermal radiation brightness, and τ is the atmospheric transmittance in the thermal infrared band [52].

(d) LST calculation:

$$\mathrm{TS}={\displaystyle \frac{K2}{\mathit{ln}\left(1+\frac{K1}{D\left(t\right)}\right)}}$$

(4)

where TS is LST; in Landsat 8, K1 = 774.8853 W⋅m⁻²⋅sr⁻¹⋅μm⁻¹ and K2 = 1321.0789 k.

The derived LST showed high consistency with MODIS LST products, with sensitivity analysis confirming robustness to parameter variations (<0.5 °C error).

3.2. Classification of LST

To elucidate the characteristics of LST spatial distribution with greater clarity and precision and to mitigate the impact of extraneous variables, such as the timing of data acquisition and meteorological conditions, on the estimation of surface temperatures. It is imperative to implement a grading process for the surface temperature estimation results and to ascertain the UHI grade. Currently, UHI grading methodologies are primarily categorized into two approaches. The initial approach is the equidistant division method, characterized by a significant degree of subjectivity in determining division parameters, potentially leading to inaccurate categorization. The second method, which has garnered widespread adoption, is the mean–standard deviation approach, facilitating precise UHI classification through the judicious integration of LST’s mean and standard deviation within the study area [37,53]. Accordingly, this study employed the mean–standard deviation method for grading surface temperatures. It classifies the inverted surface temperatures into seven distinct grades, utilizing ±0.5 sd, ±1.5 sd, and ±2.5 sd as the demarcation points (

Table 3. UHI classification.

Classification	Grading Method	Value Range
Mean value	a	38.26
Standard deviation	std	3.73
Lower-temperature zone	T < a − 2.5 std	T < 28.94
Low-temperature zone	a − 2.5 std ≤ T < a − 1.5 std	28.94 < T < 32.67
Sub-low-temperature zone	a − 1.5 std ≤ T < a − 0.5 std	32.67 < T < 36.40
Medium-temperature zone	a − 0.5 std ≤ T≤ a + 0.5 std	36.40 < T < 40.13
Sub-high-temperature zone	a + 0.5 std < T ≤ a + 1.5 std	40.13 < T < 43.86
High-temperature zone	a + 1.5 std < T ≤ a + 2.5 std	43.86 < T < 47.59
Higher-temperature zone	T > a + 2.5 std	47.59 < T

).

3.3. Wind Environment Assessment in CZXUA

The purpose of the wind environment assessment is twofold. The first is to clarify the spatial distribution of wind speed and provide a natural driving force basis for the layout of the cold channel network. The second is to correct the resistance surface of the cold island due to it: adjust the resistance value through the wind field data to ensure that the cooling corridor direction is synergized with the ventilation potential to enhance the cooling efficiency.

According to the meteorological bureaux of Changsha, Zhuzhou, and Xiangtan, there are 611 meteorological automatic stations within the CZXUA. Of these, 293 are in Changsha, 192 are in Zhuzhou, and 131 are in Xiangtan (Figure 1). Automatic meteorological stations record wind speed and direction data on a daily, monthly, and yearly basis. Missing data stations and anomalous data values were excluded. Based on the perennial dominant wind direction of the urban agglomeration, we weighted the wind direction of each station by the annual average wind speed. Subsequently, the ventilation potential at each station location was determined. Kriging interpolation was employed to extrapolate data from known sites to unknown locations within the same region, thereby assessing the wind environment in the study area. The selection of the kriging interpolation method was mainly based on its adaptive advantages. The kriging method quantifies spatial autocorrelation through the semi-variance function, which more accurately reflects the spatial distribution law of the wind field than deterministic methods such as inverse distance weighting [54] and also provides an interpolation error surface (kriging variance), which identifies the reliability of the sparse region of the data.

$$\mathrm{Z}(\mathrm{s}0)={\sum }_{i=1}^N{\lambda }_{i}Z\left(s_i\right)$$

(5)

Here, Z(s0) is the predicted value at s0; Z(s_i) is the measured value at the ith position; λ_i is the unknown weight of the estimated value at the ith position; s0 is the predicted position; and N is the number of measurement sample points [54].

3.4. Cold Island Network Construction

3.4.1. Identification of Heat and Cold Island Sources

(1) MSPA

MSPA refers to the discrimination and identification of habitat patches within the study area that are crucial for landscape connectivity [55], highlighting the structural connectivity of the landscape. It categorizes the necessary image elements as foreground data based on spatial topological relationships, whereas the remaining image elements serve as background data. Subsequently, the foreground data are classified into seven distinct landscape types utilizing image processing techniques (

Table A2. MSPA types and implications.

Morphological Types	Implications
Core	Larger habitat patches in the foreground image, “sources” of multiple ecological processes.
Bridge	Narrow areas connecting adjacent core areas, characterized as corridors, represent corridors for energy exchange and material flow and are important for landscape connectivity.
Edge	A transition area is created between the core area and the peripheral non-study landscape areas to protect the ecological processes and natural succession in the core area and reduce the impact of anthropogenic disturbances in the external landscape.
Branch	Extended areas of the prospective study landscape with only one end connected to the edge zone. Bridges, traffic circles, or perforations serve as conduits for exchanging energy with the peripheral landscape.
Loop	Internal corridors connecting the same core area, also characterized as corridors, for energy exchange within the core area.
Islet	Isolated, fragmented, poorly connected patches that are not connected and have less potential for internal material and energy exchange and transfer.
Perforation	The transition area between the core area and its internal background data landscape, also with edge effects.

) [56]. MSPA is predominantly employed in landscape ecology research, focusing primarily on the identification of ecological sources [57,58]. In the present study, MSPA was utilized to delineate urban cold island (and heat island) areas [4,37,59]. The lower-temperature zone, low-temperature zone, and sub-low-temperature zone were designated as foreground data with an assigned value of 2, while other regions were classified as background data with a value of 1 (the converse applied for identifying heat island areas). Employing Guidos, seven spatial elements from the cold island (heat island) map were extracted. The cooling efficacy of an urban cold island is directly proportional to the area of the cold island [33], yet scholars have not established a definitive threshold for its size. Accordingly, this study referenced prior research [4] and considered the specific context of the study area to select cold and hot island patches with a core area exceeding 1 km² for subsequent landscape connectivity analysis.

(2) Landscape connectivity analysis

The cooling efficacy of urban cold islands is not only correlated with the area’s size but also directly proportional to the islands’ landscape connectivity [59]. Conefor 2.6 was utilized to assess the landscape connectivity. Upon comparison, optimal connectivity between patches was achieved when the distance threshold for connectivity was established at 10,000 m, with the connectivity probability threshold set at 0.5. Calculations were conducted for the index of possible connectivity (PC) and the index of connectivity importance (ICC) for each core area. The formulae for these calculations are as follows:

$$ICC=\left\{\sum _{i=1}^n\sum _{j=1}^n{\displaystyle \frac{a_i\times a_j}{1+{nl}_{ij}}}\right\}/A_L^2$$

(6)

$$PC=\left\{\sum _{i=1}^n\sum _{j=1}^n{p}_{ij}^{\mathrm{*}}{a}_i{a}_j\right\}/A_L^2$$

(7)

dPC = (PC − PCremove)/PC × 100%

(8)

where i ≠ j, n represents the total number of patches extracted from the ecological source, ai is the area of patch i, a_j is the area of patch j, nl_ij is the number of connections between patches i and j, $A_L^2$ represents the total area of the woodland landscape, $p_{ij}^{\mathrm{*}}$ is the maximum product probability of all paths between patches i and j, and PCremove is the overall index value of the remaining patches after removing a certain patch [60].

3.4.2. Construction of Resistance Surface

The cold island resistance surface reflects the difficulties faced by cold air in the process of landscape dispersion [61]. Impediments to cold air circulation within urban environments can be broadly classified into two categories: natural and socio-economic factors. Natural factors encompass DEM [33] and NDVI [62]. NDVI was selected primarily for its capacity to mitigate the impacts of radiative variations due to atmospheric conditions, including solar altitude, satellite viewing angles, topography, and cloud cover [62]. Socio-economic factors primarily comprise land use types [63], road densities [4], and other relevant variables. It is not difficult to find that current resistance factors are predominantly two-dimensional, neglecting the impact of three-dimensional elements [64]. Furthermore, the obstruction of urban airflow by dense high-rise structures in metropolitan areas merits consideration. Regions of heightened building density frequently serve as concentrated heat sources, thereby impeding the movement of cold air [59]. In addition, we used the MSPA type as an influence factor on the cold island resistance surface, as it accurately delineates the functional zones of the cold island (e.g., core, bridge, and loop) through the spatial topological relationship and quantifies the differential influence of the different zones on the cold air dispersion [59]. Based on the preceding analysis, we identified seven critical factors for constructing the resistance surface: land use types, MSPA types, DEM, NDVI, road densities, building densities, and building heights (

Table 4. Resistance surface factors.

Resistance Factors	Description	Grade	Value
Land use type	Natural, two-dimensional	Water bodies, wetlands	10
		Forest	30
		Grassland	50
		Farmland, bare land	70
		Construction land	90
DEM	Natural, three-dimensional	0–50	10
		50–100	30
		100–200	50
		200–500	70
		500–1000	90
NDVI	Natural, two-dimensional	>0.8	10
		0.6–0.8	30
		0.4–0.6	50
		0.2–0.4	70
		<0.2	90
Building height	Socio-economic, three-dimensional	0–15.96	10
		15.96–40.77	30
		40.77–79.76	50
		79.76–152.44	70
		152.44–452	90
Building density	Socio-economic, two-dimensional	0–0.039	10
		0.039–0.092	30
		0.092–0.174	50
		0.174–0.331	70
		0.331–1	90
Road density	Socio-economic, two-dimensional	0–0.0007	10
		0.0007–0.0027	30
		0.0027–0.0064	50
		0.0064–0.0123	70
		0.0123–0.027	90
MSPA type	Socio-economic, two-dimensional	Core	10
		Bridge	30
		Loop, branch	50
		Islet, edge, perforation	70
		background	90

). Spatial principal component analysis (SPCA) was employed to ascertain the weights of resistance factors. This approach mitigates the subjectivity inherent in prior weight assignment methodologies, such as the analytic hierarchy process (AHP) [4,33]. The SPCA calculation formula is as follows:

$${\mathrm{F}}_{\mathrm{i}}={\displaystyle \frac{X_i-{\mu }_i}{{\sigma }_i}}$$

(9)

$X_i$ is the original factor value; ${\mu }_i$ is the mean; and ${\sigma }_i$ is the standard deviation.

$${\mathrm{PC}}_{\mathrm{k}}={\sum }_{i=1}^P{w}_{ki}\times F_i$$

(10)

PC_k is the kth principal component score; w_ki is the loading of the ith resistance factor in the kth principal component; and F_i is the normalized ith resistance factor value.

$${\mathrm{W}}_{\mathrm{i}}={\displaystyle \frac{\sum _{k=1}^n\left|w_{ki}\right|\times {\lambda }_k}{\sum _{i=1}^7\sum _{k=1}^n\left|w_{ki}\times {\lambda }_k\right|}}$$

(11)

W_i is the final weight of the ith factor.

$$\mathrm{R}={\sum }^7{w}_i\times F_i$$

(12)

R is the integrated resistance surface before correction.

Existing studies have overlooked the impact of the wind environment. Consequently, this study corrected the resistance surface based on the ventilation potential:

$$\mathrm{R}’\mathrm{R},=\mathrm{R} \ast {\displaystyle \frac{S_a}{S_i}}$$

(13)

where R, is the corrected resistance surface; R is the resistance surface before correction; S_i is the ventilation potential of grid i; and S_a is the average ventilation potential of CZXUA.

3.4.3. Construction of Cold Island Spatial Pattern (CSP)

Circuit theory, a concept originating from physics, was initially introduced by the American scholar McRae for ecological studies and initially employed to forecast species inheritance [59]. It subsequently evolved to examine the random walk attributes of wildlife, utilizing the analogy of electrons’ random traversal within circuits to simulate species migration and dispersal processes. Consequently, circuit theory can be applied to pinpoint critical nodes within potential cooling corridor networks by modeling random flow dynamics [4,37,59]. Regions with higher current values are considered important cold and hot spaces (cooling nodes, heating nodes).

Linkage Mapper, utilizing CircuitScape and ArcGIS as its foundation, serves as the principal instrument for applying circuit theory. It is recognized for its efficiency, convenience, authenticity, and precision, hence its widespread adoption in recent years for studies pertaining to network construction [65,66]. The Linkage Mapper toolkit comprises Centrality Mapper, Pinchpoint Mapper, Barrier Mapper, and Linkage Priority Mapper. Centrality Mapper and Linkage Priority Mapper assess the aggregate current flow for each source and corridor junction. The greater the current centrality, the more pivotal the source and corridor are for the overall network connectivity. Empirical research [67,68] has validated the efficacy and accuracy of Centrality Mapper and Linkage Priority Mapper in grading the significance of sources and corridors. Pinchpoint Mapper is primarily utilized for the identification of ecological pinch points, which denote locations that are obligatory for organism migration. In circuit theory modeling, ecological pinch points are areas in ecological corridors with higher current density, representing a higher likelihood of passage at that location [69]. In the cooling corridor network, it is the area where cold air must pass through or frequently flow. These corridors’ key nodes, subject to concentrated airflow, can notably reduce the surrounding temperature such that we can define them as cooling nodes. Barrier Mapper is predominantly employed for the detection of ecological barrier points, which are regions that impede movement and communication among organisms, and their removal can enhance the connectivity between ecological sources [70]. Within the cooling corridors network, a heating node is an area where cold airflow is impeded. For example, a dense area of high-rise buildings or topographical barriers (e.g., mountain gaps) can significantly impede the flow of cold air (high-resistance areas), leading to localized heat buildup. Therefore, it can be defined as a heating node. The cold island sources, cooling corridors, cooling nodes, and heating nodes together constitute the CSP.

4. Results

4.1. UHI Classification

The LST inversion was conducted utilizing the atmospheric correction method. The analysis revealed that the LST of the CZXUA spanned from 8.73 to 53.54 °C. The inversion outcomes were subsequently employed to categorize the UHI effect, employing the mean-standard deviation approach. As depicted in Figure 3, the higher-temperature zones are predominantly situated within the central urban regions of Changsha, Zhuzhou, and Xiangtan, as well as within the core urban areas of Ningxiang, Liuyang, and Liling. The high-temperature and sub-high-temperature zones are primarily located in the periphery of the urban centers or in the BGS near the inner part of the urban centers. Moreover, there are also high-temperature zones (sub-high-temperature and higher-temperature zones) in the form of a “mesh.” This phenomenon is primarily attributed to the substantial thermal emissions from major regional traffic artery, including highways.

4.2. Assessment of the Wind Environment in CZXUA

Kriging interpolation was utilized to disseminate the data from 611 points across the study area. This ultimately yielded an assessment map depicting the CZXUA’s ventilation potential (Figure 4). Areas characterized by high ventilation potential are primarily concentrated within the eastern portion of the CZXUA, with a preponderance of these areas located at higher elevations, such as mountain ranges. Conversely, areas exhibiting low ventilation potential are predominantly situated in the low-hill regions of the central part of the CZXUA.

4.3. Construction of CSP

4.3.1. Construction of Cold Island Sources

Based on LST, MSPA and landscape connectivity analysis were conducted, resulting in the identification of 39 cold island sources, as depicted in Figure 5c. The cold island sources exhibit a broad spatial scale, ranging from 1 km² to 2019 km² in individual area, with a total aggregated coverage of 4356.73 km²—representing 15.5% of the entire study region (see

Table A3. Landscape connectivity analysis of the cold island sources.

ID	Area (km2)	dIIC	dPC	ID	Area (km2)	dIIC	dPC
1	1.9989	0.0001	0.0182	21	21.2364	0.0089	0.0224
2	1.0566	0.0423	0.0491	22	36.6705	0.0267	0.0525
3	3.3453	0.0002	0.0178	23	55.7964	0.0617	0.1206
4	11.8728	0.0028	0.1008	24	14.1048	0.0039	0.0447
5	14.0634	0.0072	0.0102	25	4.6170	0.0004	0.0122
6	1.3914	0.0000	0.0112	26	341.8010	2.3162	2.4245
7	25.5942	0.0152	0.0772	27	16.5186	0.0068	0.0215
8	47.2752	0.0515	0.1062	28	13.8375	0.0038	0.0125
9	5.7546	0.0046	0.0139	29	158.6910	0.4993	0.5453
10	2018.9500	80.8542	80.0947	30	16.4916	0.0054	0.0118
11	120.8350	0.2895	0.4207	31	7.8525	0.0012	0.0105
12	33.1992	0.0219	0.0730	32	24.6204	0.0120	0.0447
13	6.2415	0.0022	0.0197	33	7.1280	0.0020	0.0154
14	9.8433	0.0039	0.0210	34	3.2616	0.0005	0.0296
15	123.3870	0.3018	0.3988	35	3.9231	0.0006	0.0368
16	177.7980	0.6374	0.7393	36	6.0957	0.0015	0.0144
17	16.5159	0.0054	0.0175	37	42.7329	0.0362	0.1577
18	19.4778	0.0092	0.0309	38	13.2165	0.0042	0.0210
19	51.5538	0.0547	0.1153	39	860.5970	14.6833	14.7872
20	17.3853	0.0060	0.0163

for spatial distributions and area breakdowns). In terms of spatial distribution, the cold island sources are predominantly located in the eastern sector of Changsha, the central and southern regions, the southeastern area of Zhuzhou, and the northern region of Xiangtan (such as Jiuling Mountain Range and Lianyun Mountain Range). They are located away from urban built-up areas, exhibiting a distribution pattern of “peripheral concentration and core deficiency.” Most cold island sources are woodlands and mountain ranges, which overlap significantly with the large BGS of the CZXUA.

A total of 58 heat island sources were identified (Figure 5c). Regarding area, the heat island sources span an area ranging from 5.15 to 562.03 km². With respect to spatial distribution, the heat island sources exhibit uniform distribution across the CZXUA region, predominantly within urban and rural construction land, with a particular concentration in the metropolitan areas of Changsha, Zhuzhou, and Xiangtan. It exhibits a spatial distribution characterized by a “core cluster and network-like diffusion” pattern. This pattern suggests a pronounced UHI effect within the CZXUA.

The spatial distribution of cold and heat sources shows significant opposition, with the minimum boundary distance between them being just 300 m (e.g., between cold source 11 and heat source 13), highlighting the risk of urban expansion eroding natural cold sources.

4.3.2. Cold Island Resistance Surface

The magnitude of the resistance value quantifies the degree to which cold air circulation is facilitated or hindered within a given area, with lower values indicating preferential pathways for airflow and higher values reflecting thermal or structural barriers. Following the reclassification and valuation of the seven resistance factors, the cold island resistance surface was generated utilizing SPCA (individual resistance factor diagrams are shown in Figure A1, Figure A2, Figure A3, Figure A4, Figure A5, Figure A6 and Figure A7). Percentage and cumulative eigenvalues of the SPCA results (

Table A4. Percentage and cumulative eigenvalues.

PC Layer	Eigenvalue	Percentage Eigenvalues	Cumulative Eigenvalues
1	4.33 × 103	85.4373	85.4373
2	4.25 × 102	8.3833	93.8207
3	1.49 × 102	2.9359	96.7565
2	4.25 × 102	8.3833	93.8207
3	1.49 × 102	2.9359	96.7565
3	1.49 × 102	2.9359	96.7565
4	1.10 × 102	2.1715	98.928
7	1.34 × 10−6	0	100

) show that the cumulative variance contribution rate of the first three principal components reached 96.76%, indicating that these components encompass most of the spatial variability information from the original data. Therefore, we selected the first three principal components to construct the resistance surface and reduce redundancy while retaining key spatial characteristics. Finally, the ventilation potential data were used to correct the resistance surface (Figure 6). The resistance values span a range from 27.72 to 128.92. Regarding spatial distribution, regions exhibiting higher resistance values are predominantly located within urban and rural construction land, whereas the cold island sources exhibit lower resistance values.

4.3.3. Result of Cooling Corridor Identification

Based on the cold island sources and the resistance surface, 89 cooling corridors were delineated employing circuit theory and Linkage Mapper (Figure 7). The lengths of the cooling corridors extend from 72 to 188,316 m, exhibiting uniform distribution across the central and southern regions of Changsha and throughout Zhuzhou and Xiangtan. As indicated by the wind rose charts for Changsha, Zhuzhou, and Xiangtan, the prevailing wind direction within the CZXUA is from the northwest. A majority of the identified cooling corridors align parallel to the prevailing wind direction, thereby substantiating the precision of the identification process. This not only aids in mitigating the UHI effect but also enhances urban ventilation.

4.3.4. Cooling and Heating Node Identification

Employing Pinchpoint Mapper, a total of 42 cooling nodes were identified. These nodes span an area from 1.52 to 605.48 hectares, encompassing a collective area of 34.62 km² (Figure 8a). Cooling nodes are mostly distributed on the outskirts of cities or in large ecological spaces inside, such as the water bodies along the Xiangjiang River, Zhaoshan Scenic Area. These nodes form local cold sources through transpiration and surface radiation cooling effects, and are spatially distributed in a “patchy discrete” pattern. Barrier Mapper was utilized to identify heating nodes. A total of 63 heating nodes were discerned, spanning areas from 10.35 to 6028.42 hectares and aggregating to a total area of 314.7 km² (Figure 8b). Heating nodes are mainly concentrated in urban core areas, especially high-density built-up areas, transportation hubs and industrial clusters. Such areas have a high rate of surface hardening, significant anthropogenic heat emissions, forming a contiguous heat island effect area, and the spatial distribution is characterized by “multi-core agglomeration.”

Subsequently, we employed Centrality Mapper and Linkage Priority Mapper to categorize the cold island sources and cooling corridors, respectively. The CSP of the CZXUA was jointly constructed (Figure 8c). The CSP was stratified into three levels utilizing the natural breakpoint classification method. A total of six first-level corridors were discerned, amassing a cumulative length of 902.2 km; fifteen second-level corridors, aggregating to a total length of 845.3 km; and sixty-eight third-level corridors, summing to a total length of 479.6 km. A total of ten first-level cold sources, encompassing a collective area of 3912.4 km²; fourteen secondary cold island sources, aggregating to a total area of 296.5 km²; and thirteen tertiary cold island sources, encompassing a collective area of 147.82 km². The primary corridor runs through the study area in a southeast–northwest direction, parallel to the dominant wind direction. The secondary corridor is branched to the middle of the urban area, covering the transition area. The tertiary corridors are intensively embedded in urban built-up areas, especially in high-density heat island areas (such as the core area of the main city).

The LST of cooling corridors varies significantly across different widths. To analyze the heat dissipation efficiency of the cooling network, it is essential to delineate the thermal boundaries of corridors and rationally plan width thresholds to maximize their cooling effectiveness. In this study, we established 20 buffer zones centered on the corridors, with buffer distances ranging from 10 to 1000 m. We analyzed the cooling effects of corridors within different width thresholds and compiled LST data for ecological corridors under varying widths (Table A4).

Using average LST as the dependent variable and corridor width as the independent variable, curve estimation was performed to identify the best-fitting function. Results showed that quadratic and cubic functions achieved superior model fit (R² = 0.939 and 0.962, respectively) (

Table 5. The fitting results of different functions.

Equation	R2	F	Sig
Linear function	0.680	38.262	0.000
Logarithmic function	0.877	128.574	0.000
Quadratic function	0.939	130.694	0.000
Cubic function	0.962	136.564	0.000
Power function	0.877	128.645	0.000
Exponential function	0.680	38.206	0.000
Logistic	0.680	38.206	0.000

). Based on the polynomial functions illustrated in Figure 9 and considering constraints from planning policies and construction costs (as corridor width cannot expand indefinitely), the optimal threshold for cooling corridor width was determined to be 600 m.

4.4. Multiscale Analysis of CSP in Alleviating UHI Effect

To better reveal the differences in thermal environment responses at various spatial levels, provide a scientific basis for alleviating the UHI effect at different scales within the urban agglomeration, and avoid the limitations of single-scale strategies, we divided the CZXUA into three scales: the central urban area, the suburban scale, and the periphery of the urban agglomeration for comparative analysis (Figure A8). As shown in

Table 6. Multiscale analysis table.

Scale	Cold Island Source Area (km2)	Corridor Length (m)	Cold Island Source Density	Corridor Density (m/km2)	Source Centrality (Average)	Average LST (°C)
Urban scale	13.9	2528.17	1.43%	2.60	274.55	41.19
Suburban scale	538.22	969,319.68	4.86%	87.48	237.91	38.77
Urban agglomeration scale	4356.73	2,227,198.86	15.52%	79.31	152.42	36.25

, the corridor density peaks at the suburban scale (87.48 m/km²), reflecting the ecological buffering role of the suburban area. The average LST decreases gradually with scale expansion, which reflects the effect of regional collaborative cooling. The centrality of the cold island sources at the urban scale is much higher than that at other scales, highlighting the importance of the sources of the central urban area.

4.5. Relationship Between the CSP and the UHI

The CSP was analyzed by overlaying it with UHI patches (Figure 9). It was observed that certain cold island sources are near heat island patches. The main ones are cold island sources such as 11, 28, and 33. For these cold island sources, they should be prevented from being “eroded” by the heat island. Priority can be given to establishing ecological corridors between the two as a buffer zone. Concurrently, during the ecological planning of urban agglomerations, stringent regulations should be applied to cold island sources, with an emphasis on integrating key sources into the ecological protection red line for stringent preservation.

The “edge agglomeration” of the cold island source originates from the natural cold source advantage of the eastern mountains and the Xiangjiang River system, while the “core contiguity” of the heat island source is driven by high-density built-up areas and traffic heat emissions (Figure 10). This confrontation pattern shows that the trend of continuous expansion of heat islands needs to be broken through the cross-scale cooling corridor.

Additionally, it was observed that certain first- and second-level cooling corridors traverse UHI areas. These corridors exhibit considerable length. Consequently, the implementation of stringent conservation measures for these corridors is imperative for preserving the integrity of the urban agglomerations’ CSP. In tandem, enhanced protective strategies should be applied to corridors traversing densely populated urban locales. Measures such as river regulation, integration of shoreline resources, and widening of greenways can be taken.

There are also cooling nodes and heating nodes located within UHI patches. For these zones, strategies should be implemented, including converting farmland back to forests and lakes and enhancing urban parks, pocket parks, and other urban green infrastructure services. In instances where significant practical impediments or prohibitive economic costs for renovation are present, the promotion of green buildings to efficiently absorb solar radiation is advisable. Additionally, the architectural layout of the area should be strategically planned to diminish building density and augment green spaces and vegetation, thereby effectively mitigating the UHI effect.

5. Discussion

5.1. UHI Effect Mitigation Strategy Based on CSP

Based on the results of multiscale analysis (Section 4.4) and the overlay analysis of CSP and UHI patches (Section 4.5), we propose a hierarchical governance framework for the three-level urban heat island mitigation zones (Figure 11). The primary thermal mitigation zone corresponds to the urban scale, mainly covering the main urban areas of Changsha, Zhuzhou, and Xiangtan. This area also has the most significant heat island effect in the CZXUA (LST has the highest mean value, reaching 41.19 °C). The cold island sources within this zone (mainly including Xiangbiwo Forest Park and Maolichong Forest Park) have the highest centrality, with an average value of 274.55. Therefore, it is the most crucial and significant in mitigating the UHI effect of the CZXUA, and thus requires focused attention and strict protection. In addition, the density of cold island sources in the urban area is relatively low (only 1.43% of the area), so the density of cold island sources can be prioritized improved by constructing urban parks and other measures. The secondary thermal mitigation zone corresponds to the suburban scale, mainly covering the cold island sources on the city’s outskirts, including Xiangshan National Forest Park, Shaoshan Scenic Spot, etc. Suburban areas have the highest corridor density (87.48 m/km²), despite cold sources accounting for only 4.86% of the area such that the secondary thermal mitigation zone mainly plays a transitional role in alleviating the UHI effect within the urban agglomeration. The tertiary thermal mitigation zone corresponds to the urban agglomeration scale and is located mainly on the periphery of peri-urban agglomerations. This region features a relatively concentrated and substantial array of cold island sources, including the Jiuling Mountain System and the Lianyun Mountain System. The tertiary thermal mitigation zone primarily serves to prevent the proliferation of the UHI effect and to obstruct the connectivity of the urban heat island network both within and beyond the urban agglomeration. The construction of the three-tiered heat mitigation zones mirrors the distinct requirements for heat island management, the necessity for efficient resource allocation, and the spatial characteristics of heat islands. This hierarchical design aligns with the radial diffusion pattern of heat islands (“core–edge–periphery” structure). This structure is essential for effective urban planning and resource allocation to combat heat islands. Concurrently, the cooling and humidifying effects of existing large river systems, which are significant BGSs within urban agglomerations, should not be overlooked. Accordingly, we designate the existing large rivers, including the Xiangjiang River and its seven primary tributaries, potential cooling corridors.

In addition, while the 600 m threshold for width optimizes the cooling efficiency, its application in urban areas with dense development requires adaptive strategies. In high-density zones, strictly observing a 600 m corridor may conflict with urban development priorities. Therefore, we propose a hierarchical planning approach. In the periphery of urban agglomeration where land use intensity is low, a width of 600 m is feasible. In suburban transition zones, widths of 100–500 m can be used in combination with forest parks. In city centers, we stress micro-interventions that synergize with existing BGS. For example, widening riverbanks, upgrading urban parks, or integrating greenways into transport networks can cut land acquisition costs, and widths under 100 m are acceptable here.

5.2. Detailed Comparative Analysis with Existing Relevant Studies

In recent years, the more authoritative, downloaded, and cited literature on constructing urban cold island networks to mitigate the UHI effect mainly includes Peng et al. (2022) [59], Qian et al. (2023) [4], Qiu et al. (2023) [62], and Liu et al. (2024) [37]. The innovation of this study is reflected in the synthesis of methods, scale expansion, and the depth of practical strategies, which are important supplements and breakthroughs in existing research. Specific comparative analysis is shown in

Table 7. Detailed comparative analysis with existing relevant studies.

Research Sources	Research Scale	Methodological Differences	Improvement Point
Peng et al. (2022) [59]	Guangzhou–Foshan metropolitan area, heat island interactions at urban agglomeration scale not addressed.	1. The effect of wind environment is not considered. 2. Theoretical network construction only.	1. Taking urban agglomeration as the research object, considering the thermal influence between cities. 2. Using circuit theory to verify the efficiency of cooling corridor ventilation. 3. New cooling/heating node identification module (direct guidance for planning to land). 4. Wind environment assessment to correct the resistance surface. 5. Superimpose the CSP and heat island patch analysis, put forward more targeted strategies. 6. The optimal threshold for corridor width has been determined.
Qian et al. (2023) [4]	A single city (Wuhan) was used as the subject of the study, and heat island interactions at the scale of urban agglomerations were not addressed.	1. Only two-dimensional factors were considered, not three-dimensional factors 2. AHP subjective weighting method was used. 3. The influence of the wind environment is not considered.	1. Taking the urban agglomeration as the research object, considering the heat influence between cities. 2. Three-dimensional factors such as building height are considered. 3. SPCA is objectively empowered, overcoming the subjectivity of AHP. 4. Wind environment assessment to correct the resistance surface. 5. Superpositioning CSP with heat island patches to analyze and propose more targeted strategies. 6. The optimal threshold for corridor width has been determined.
Qiu et al. (2023) [62]	A single city (Nanjing) was used as the study object, and heat island interactions at the scale of urban agglomerations were not addressed.	1. Only two-dimensional factors were considered, not three-dimensional factors. 2. AHP subjective assignment method is used. 3. Only the dominant wind direction is considered to grade the corridor.
Liu et al. (2024) [37]	For example, the focus of Fuzhou’s central city is on the distribution of heat islands within the city, but it is not extended to the regional scale.	1. Only two-dimensional factors were considered, not three-dimensional factors. 2. The influence of the wind environment was not considered.	1. Taking the urban agglomeration as the research object, considering the heat influence between cities. 2. Three-dimensional factors such as building height are considered. 3. Wind environment assessment to correct the resistance surface. 4. Superpositioning CSP with heat island patches to analyze and propose more targeted strategies. 5. The optimal threshold for corridor width has been determined

.

5.3. Advantages of the Methodology

The CSP was constructed from the perspective of landscape connectivity to mitigate the UHI effect. It is more effective than previous studies that only focused on isolated landscape patches to mitigate the UHI effect over a larger regional scale. In terms of research scale, this study breaks the traditional single-city heat island research framework [4,62]. It proposes a regionalized heat mitigation paradigm on the scale of urban agglomeration [40], a typical form of urbanization in China. Taking the CZXUA as an example, its innovative value is reflected in: (1) cross-border heat management—constructing strategic cooling corridors across administrative boundaries to solve the problem of superposition of heat flow between cities; (2) hierarchical resource optimization—constructing three-level heat mitigation zones (city core–transition zone–periphery), differentiated allocation measures, and breaking through the single-city resource; and (3) policy scalability—complying with China’s “ecological civilization” regional collaboration policy and promoting climate resilience governance in synergy with sustainable development goals.

In the framework of this study, the paper assesses the wind environment, focusing on the impact of wind speeds. Building on this, this paper introduces a comprehensive paradigm for the construction of CSPs: “identification of cold island sources–construction of resistance surfaces–extraction of cooling corridor networks–identification of critical nodes.” We considered three-dimensional factors such as building height in constructing resistance surfaces, which is more scientific than most previous studies [4,37,64,71] that ignored the three-dimensional factor. Concurrently, SPCA was utilized to ascertain the weighting of each resistance factor, thereby addressing potential subjectivity inherent in prior methodologies [4,71]. Centrality Mapper and Linkage Priority Mapper were deployed to assess and rank the significance of cold island sources and the cooling corridor network, respectively, offering a more reliable approach than previous reliance solely on the natural breakpoint method [72]. Related studies [4,37] lack comparative analysis between cooling and heating spaces and urban heat island sources. This study compares the CSP’s cooling and heating nodes with urban heat islands within urban agglomerations, proposing more pragmatic and actionable strategic recommendations.

In summary, this study innovatively considered the effects of meteorological and three-dimensional factors such as wind speed and building height. We have established a comprehensive circular system that encompasses cold island sources, cooling corridor networks, and nodal spaces, thereby developing a CSP model that transitions from points to lines and from surfaces to rings. This design substantially augments the depth and scope of the research, concurrently enhancing its applicability in real-world contexts. This framework offers critical insights for mitigating the UHI effect amidst rapid urbanization and serves as a valuable reference for advancing sustainable urban development strategies and urban climate adaptation planning.

5.4. Limitations and Future Research Directions

The core methodology of the CSP construction framework in this study is universal and applicable to other urban agglomerations, such as the Yangtze River Delta (YRD) and Beijing–Tianjin–Hebei (BTH) urban agglomerations, which also face the challenge of UHIs. Firstly, regarding climate adaptation, key steps in the framework, like resistance surface construction, can be adjusted according to different climatic zones’ characteristics. For instance, in arid urban agglomerations, it is necessary to focus on the limitations of NDVI and water scarcity on the cold island network. The resistance surfaces can be modified accordingly by combining with the diffusion resistance of sand and dust. Secondly, the CSP framework can flexibly adapt to urban agglomerations at different spatial scales. The three-tiered heat island mitigation zones strategy (main city–suburban–suburban) proposed in this study can dynamically adjust the scope of the tiers. This dynamic adjustment allows the strategy to adapt to the expansion or contraction of urban agglomerations, making it a versatile tool for addressing UHI effects across various urban contexts. Nevertheless, the approach is not without its limitations. Primarily, while numerous studies depend on remote sensing technology to examine the urban thermal environment, such methodology can be compromised by adverse weather conditions, such as cloudiness or fog, thereby diminishing the image data quality. Additionally, we only used annual average wind speed data. Future studies could incorporate seasonal or even diurnal wind variations to optimize corridor designs for specific microclimates, particularly in high-density urban cores dominated by local airflow dynamics. Furthermore, our threshold setting of cold island areas greater than 1 km² overlooks the cumulative effects of small-scale cold islands (although individual small cold islands, such as community green spaces, have limited cooling capacity, their spatial aggregation may form a regional cooling network [21]) and the non-area-dependent ecological functions (some linear cold islands, such as river corridors, can play critical ventilation roles due to their continuity, despite their smaller areas [29], and the threshold setting may diminish their contributions). These limitations could result in a lower-than-optimal density of cooling corridor networks. Future studies could employ machine learning methods, such as random forests or convolutional neural networks (CNNs), to automatically identify the functional effectiveness of cold islands, thereby replacing the reliance on a single area threshold. Finally, the construction of cold island networks and CSP using circuit theory in this study also suffers from limitations such as vertical dimension simplification and insufficient 3D obstacle modeling, which can be followed up by coupling CFD models to improve the construction of 3D resistance surfaces.

6. Conclusions

The intensification of the UHI effect presents a substantial threat to urban environments and the well-being of their inhabitants. Accordingly, this study investigated strategic approaches for UHI mitigation by integrating urban cold islands and landscape connectivity to optimize the distribution of isolated cold islands. Concurrently, regarding the research scale, existing studies predominantly concentrate on urban or small-town scales, often overlooking the urban agglomeration scale. This study selected the CZXUA, a quintessential hot summer and cold winter region in China, as its research subject. We innovatively considered the effects of meteorological and three-dimensional factors such as wind speed and building height. Employing modeling techniques including MSPA, landscape connectivity analysis, SPCA, and circuit theory, this study proposes a comprehensive framework for CSP construction: identification of cold island sources, construction of resistance surfaces, extraction of cooling corridor networks, and identification of critical nodes. A total of 39 cold island sources, 89 cooling corridors, 42 cooling nodes, and 63 heating nodes were identified. We have calculated that the optimal threshold for the cooling corridor is 600 m. Based on the CSP, this study proposes a three-tiered urban agglomeration heat island mitigation framework. The primary thermal mitigation zone constitutes the most critical component of UHI effect mitigation, encompassing primarily the central urban areas of the urban agglomeration. The secondary thermal mitigation zone is predominantly situated in suburban areas and functions as a transitional zone. The tertiary thermal mitigation zone primarily serves to prevent the proliferation of the UHI effect and to obstruct the connectivity between the heat island networks within and beyond the urban agglomeration. The CSP developed in this study effectively mitigates the regional UHI effect at the macrolevel. Based on these findings, this study proposes the following policy enhancement pathways.

(1) Ecological Protection Policies: Incorporate core cold island sources into ecological protection redlines, legally prohibit encroachment, and establish ecological compensation mechanisms.

(2) Urban Planning Regulations: Formulate legal width standards for cooling corridors (e.g., 600 m), mandate ventilation spaces along new roads and riverbanks, and integrate 3D resistance factors into regulatory planning indicators.

(3) Regional Collaboration Mechanisms: Establish a cross-city thermal environment governance alliance under the CSP framework to jointly regulate cross-boundary heat flows (e.g., industrial heat sources, thermal radiation from transportation networks).

(4) Climate-Adaptive Investments: Prioritize the deployment of pocket parks and vertical greening systems in heating nodes and integrate these initiatives into special funding programs.