Urban Thermal Regulation Through Cold Island Network Evolution: Patterns, Drivers, and Scenario-Based Planning Insights from Southwest China

Abstract

With the dual pressures of accelerating urbanization and global climate warming, understanding the evolution and connectivity of cold island networks has become crucial for managing urban thermal risks. This study explores the spatiotemporal dynamics, driving mechanisms, and scenario-based projections of cold island networks in a rapidly urbanizing region of Southwest China. Using multi-temporal Landsat imagery (2000–2024), ecological resistance surface modeling, and least-cost path analysis, the study evaluated historical changes and simulated future scenarios for 2035 and 2050 under both Natural Development (ND) and Park City (PC) planning interventions. The findings reveal that: (1) Between 2000 and 2024, rapid urbanization significantly expanded high-temperature areas, fragmented cooling sources, and reshaped cold island connectivity into a hierarchical corridor network centered on a dominant ventilation axis; (2) Since 2019, ecological restoration measures have notably enhanced the structural cohesion and connectivity of cooling corridors, partially mitigating previous fragmentation; (3) Scenario simulations indicate that proactive ecological planning could reduce the extent of high-temperature zones by approximately 20% by 2050, demonstrating strong potential for mitigating future thermal risks. Overall, the results emphasize the necessity of incorporating continuous cold island corridors and connectivity principles into urban spatial planning to enhance climate resilience and support sustainable development.

1. Introduction

Amid accelerating global climate change and rapid urbanization [1], the urban heat island (UHI) effect has emerged as a pressing environmental concern [2,3,4], posing significant threats to urban ecological security, public health, and sustainable development. Originally introduced by Manley in 1958, the UHI phenomenon refers to the markedly higher temperatures observed in densely built urban areas with extensive impervious surfaces, in contrast to their rural surroundings [5]. The intensification of UHIs not only increases energy demand but also disrupts the ecological regulation functions of cities. As urban areas expand, natural land surfaces are replaced with roads, buildings, and paved spaces, which alter surface energy exchanges and reduce nighttime cooling, thereby generating abnormal thermal environments [6,7,8].

Under global warming, the effects of UHIs are compounded and amplified. The frequency, duration, and intensity of extreme heat events have significantly increased, contributing to higher incidences of heat-related illnesses, elderly mortality, and energy consumption [9,10,11]. UHI effects also destabilize ecosystems, alter species distributions, and increase surface runoff, reducing a city’s environmental carrying capacity and adaptability—particularly in high-density urban agglomerations and newly developing areas [12,13].

Land Surface Temperature (LST), a widely used remote sensing metric, has proven effective for assessing and mitigating UHIs due to its spatial continuity, temporal consistency, and ability to capture surface thermal variations [14,15]. LST enables fine-scale identification of thermal responses across land cover types, locates heat hotspots and cooling sources, and supports early-warning systems and mitigation planning for heat risks [16]. It also provides a scientific basis for embedding heat risk control into urban planning, architectural design, and land use management [14]. As cities confront rigid construction land growth and shrinking ecological space, the traditional reliance on new blue-green infrastructure (e.g., parks, waterbodies, or green corridors) faces implementation challenges. Thus, incorporating LST spatial patterns into early-stage planning and optimizing existing cooling sources and ventilation pathways in situ has become a key strategy for enhancing thermal comfort and urban climate resilience [17].

The connectivity of blue-green infrastructure is widely recognized as critical to regional heat mitigation. Unlike isolated green patches, continuous blue-green corridors enhance ventilation, facilitate cool air transmission, and support ecological flows—offering stronger synergistic cooling effects and greater resilience potential [18]. In response, recent studies have adopted various methods to optimize blue-green networks based on thermal patterns, including wind-tunnel experiments, Weather Research and Forecasting (WRF) model simulations [19,20], computational fluid dynamics (CFD) [21,22], and least-cost path (LCP) modeling in GIS environments [23]. However, a critical methodological gap persists. Most approaches either implicitly assume that structurally defined green spaces are functionally effective “cold sources” or remain confined to descriptive analytics without providing forward-looking, policy-relevant insights.

To address this gap, this study introduces the “Cold Island Network” framework [24], which represents a conceptual and methodological advancement over traditional ecological network design. The paradigm shifts from the conventional emphasis on biological flows to a focus on thermodynamic processes. Crucially, our framework defines network sources based on their thermal performance—i.e., empirically measured low Land Surface Temperature (LST)—rather than merely on their structural land-cover type. This performance-based definition ensures that the network is anchored in proven cooling assets. Consequently, the resistance surface is modeled not as an impediment to species movement, but specifically to quantify the impedance to thermal energy exchange and cold-air advection.

To operationalize this framework, we integrate Morphological Spatial Pattern Analysis (MSPA) with circuit theory and the Linkage Mapper tool. This integration transcends the limitations of single-path models such as the LCP, capturing the multi-directional diffusion of thermal flows analogous to electrical currents. Here, MSPA not only identifies habitat cores but also delineates the spatial morphology of functionally significant cold sources, while circuit theory reveals multiple potential ventilation pathways. Circuit theory, originally developed in landscape ecology, provides a more physically grounded alternative by simulating ecological flows through analogies to voltage, current, and resistance. It effectively reveals multi-path diffusion channels and critical connectivity nodes [25,26], and has been widely applied in corridor identification and ecological network design [27]. To further enhance robustness and practical applicability, we integrate MSPA with Linkage Mapper: MSPA identifies core, edge, bridge, and other structural elements within green-blue spatial configurations [28,29], which are then used by Linkage Mapper to generate resistance surfaces and perform connectivity analysis—mapping ecological and ventilation corridors with spatial precision [30].

Furthermore, while LST research has advanced from single-time analyses to long-term trend detection and increasingly incorporates machine learning and spatial regression models for improved prediction, most existing work remains limited to static imagery correlations. There is still a lack of integrated approaches that couple dynamic LST simulations with future scenario forecasting. Especially scarce are studies on the spatial evolution of cold island systems under alternative development trajectories [24], leaving planners without quantitative, scenario-based insights for proactive thermal management. Meanwhile, the “Park City” model, a holistic strategy that embeds nature throughout the urban fabric, remains theoretically promising [24,26,27,28,29,30] but largely unexamined in terms of its quantifiable impact on urban thermal resilience.

As a strategic new district at the heart of the Chengdu–Chongqing Twin-City Economic Circle, this area exemplifies the challenges and opportunities faced by rapidly urbanizing regions in China. Since 2000, accelerated construction land expansion and large-scale land-cover transformations have substantially intensified urban heat islands (UHIs). At the same time, policy shifts emphasizing “ecological priority” and “green, low-carbon transformation” have driven the implementation of blue–green infrastructure at unprecedented scales, particularly under the national “Park City” framework [31,32]. These conflicting yet concurrent forces—urban thermal intensification and ecological restoration—make this district a representative case for investigating the spatial dynamics of urban thermal environments in the context of policy-led transformation.

Against this backdrop, this study sets out to explore the evolution and planning implications of cold-island ecological networks in a rapidly developing new district in Southwest China. Specifically, it aims to: (1) retrieve land surface temperature (LST) from multi-temporal remote-sensing imagery (2000–2024) to detect changes in cold-source patterns and identify thermally vulnerable areas; (2) integrate circuit theory and resistance-surface modeling to delineate potential diffusion corridors and key structural nodes within the cold-island network; (3) apply PLUS model under two contrasting development pathways—business-as-usual and Park City scenarios—to simulate urban thermal conditions for 2035 and 2050, thereby assessing the effectiveness of ecological interventions in mitigating future heat risks.

2. Materials and Methods

2.1. Study Area

The study area is located in a rapidly urbanizing new district in Southwest China, spanning approximately 103°47′59″ E–104°15′34″ E and 30°13′38″ N–30°40′23″ N (Figure 1). Established in 2014 as one of China’s national-level strategic development zones, this district plays a critical role in promoting regional economic integration and urban innovation, being deeply embedded within national strategic frameworks such as the Belt and Road Initiative and the Yangtze River Economic Belt. Initially dominated by agriculture and ecological land use, the area has experienced rapid transformations driven by intensive urbanization, industrial clustering, and comprehensive urban function development [33]. The total planned area covers around 1578 km², with an urbanized core region of approximately 564 km². Its resident population is projected to exceed 5 million by 2035.

Located at the southern edge of the Chengdu Plain—historically known as the “Land of Abundance”—the district is characterized by predominantly flat plains interspersed with gentle hills, providing favorable geographic conditions for urban expansion. The Jinjiang River traverses the district, forming a dense network of water bodies that support regional ecological stability. The area experiences distinct seasons, a subtropical humid monsoon climate, with an average annual temperature of 16.4 °C and 1300 mm of rainfall [34]. Such mild and humid climatic conditions underpin the district’s attractiveness for residential settlement.

Since its establishment, the district has faced increasing ecological pressure and thermal risks due to rapid construction land expansion. In response, local policymakers have proactively pursued ecological-oriented urban planning strategies. This policy-led transformation provides a unique context for studying the evolution of urban thermal environments. The district’s development can be characterized by distinct phases.

Pre-development Phase (before 2014): Prior to its formal establishment, the area was largely dominated by agriculture and natural landscapes, representing a crucial baseline.

Accelerated Urbanization Phase (2014–2019): Following its official designation as a national-level new district in 2014, the area underwent intensive construction, leading to significant thermal environmental changes.

Ecological Priority Phase (2019–Present): With the launch of China’s pioneering “Park City” initiative in 2019 [31], a major policy shift occurred, emphasizing deep integration of ecological elements—such as mountains, rivers, forests, farmland, and wetlands—into urban spaces. Key projects under this policy include major ecological restorations, such as the creation of urban lakes, large-scale forest parks, and interconnected green corridors. By significantly expanding blue–green infrastructure, including the addition of around 233 hectares of urban forests since 2017 and a targeted forest coverage rate of 35% by 2035, the district has positioned itself as a national model for sustainable and resilient urbanization. This recent phase makes it an ideal case for assessing the effectiveness of proactive ecological interventions.

Data Sources

The data used in this study primarily include remote sensing imagery, topographic variables, socioeconomic indicators, and spatial distance factors. The satellite imagery for Land Surface Temperature (LST) retrieval was sourced from the United States Geological Survey (USGS), covering five time points: 2000, 2009, 2014, 2019, and 2024. The imagery includes data from Landsat 5 TM, Landsat 8 OLI/TIRS, and Landsat 9 OLI-2/TIRS-2. LST was retrieved using the single-window algorithm applied to the thermal infrared band (Band 10), with surface emissivity estimated based on NDVI. Temperature values were then classified and normalized using the standard deviation method.

Digital Elevation Model (DEM) and slope data were obtained from the Shuttle Radar Topography Mission (SRTM) dataset, with a spatial resolution of 30 m. Slope was derived from the DEM and used as a topographic resistance factor influencing cold island diffusion.

To capture the influence of urban development on the connectivity of cold corridors, several Euclidean distance rasters were generated, including distance to water bodies, residential areas, railways, and major roads. Road network data were sourced from OpenStreetMap (OSM) and processed in QGIS to generate corresponding distance layers.

Gridded GDP data were obtained from the China GDP Spatial Distribution Dataset provided by the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, with a resolution of 1 km. Population density data were derived from the WorldPop global population dataset, with a spatial resolution of 100 m, representing regional population distribution and anthropogenic activity intensity.

All datasets (

Table 1. Data sources used in this study.

Data Name	Year(s)	Resolution	Resolution
Land Surface Temperature	2000, 2009, 2014, 2019, 2024	30 m	USGS (Landsat 5 TM, Landsat 8/9 OLI-TIRS)
NDVI	2000, 2009, 2014, 2019, 2024	30 m	Calculated from Landsat (Band 4 and 5), via USGS
DEM	2020	30 m	Geospatial Data Cloud
GDP	2020	1 km	China GDP Spatial Dataset (IGSNRR, Chinese Academy of Sciences)
Population Density	2020	100 m	WorldPop
Road Data	2020	—	Open Street Map

) were reprojected to WGS_1984_UTM_Zone_48N and clipped to the boundary of Tianfu New District. Selected variables, such as GDP and population density, were further normalized for use in constructing the ecological resistance surface.

2.2. Research Methods

2.2.1. Workflow

As shown in Figure 2, the workflow of this study consists of six main steps. First, multi-temporal Land Surface Temperature (LST) and Relative LST (RLST) were derived from remote sensing data to analyze their spatiotemporal evolution. Next, cold island patches with significantly lower temperatures were identified, and an ecological resistance surface was constructed based on factors such as land use, NDVI, and topography. Based on this, core cold sources were extracted, and a multi-level cold island corridor network was established using the Minimum Cumulative Resistance (MCR) model to evaluate its spatial structure. Finally, the PLUS model was applied to simulate land surface temperature changes under two scenarios—Natural Development and Park City—for the years 2035 and 2050, providing scientific support for optimizing the urban thermal environment.

2.2.2. Land Surface Temperature (LST) and Relative LST (RLST) Retrieval Method

To investigate the urban thermal environment and its temporal variation, both Land Surface Temperature (LST) and Relative Land Surface Temperature (RLST) were calculated based on multispectral Landsat imagery (Landsat 5 TM, Landsat 8/9 OLI-TIRS) acquired in summer (July to September) with less than 5% cloud cover.

All satellite images used in this study were sourced from the USGS Collection 2 Level-1 Terrain Precision (L1TP) products. To facilitate regional analysis and computation, all images were uniformly reprojected to the WGS_1984_UTM_Zone_48N coordinate system. Radiometric calibration and atmospheric correction were performed using the Radiometric Calibration and FLAASH modules in ENVI (v5.3), whereby DN values were converted into radiance and subsequently into surface reflectance. To ensure temporal consistency across different sensor, data quality control was implemented in two ways. First, all images were obtained from the USGS Collection 2 products, which have undergone rigorous cross-calibration, thereby providing official guarantees of radiometric consistency between sensors. Second, in LST retrieval, sensor-specific thermal infrared band configurations were addressed by applying well-established algorithms tailored to each sensor, ensuring both the accuracy and comparability of the results.

First, the digital numbers (DN) from the thermal infrared band (Band 10) were converted into Top-of-Atmosphere (TOA) spectral radiance using metadata-provided scaling factors. This was performed by applying the formula:

$$L_{\lambda }=M_L\times Q_{cal}+A_L-O_i$$

(1)

where $M_L$ and $A_L$ are radiance rescaling coefficients, and Oi is a correction value (typically 0.29 for Band 10).

Then, the TOA radiance values were transformed into brightness temperature (BT) using the Planck-based conversion with band-specific constants $K_1$ and $K_2$ extracted from the image metadata:

$$BT={\displaystyle \frac{K_2}{\mathit{ln}\left(\frac{K_1}{L_{\lambda }}+1\right)}}-273.5$$

(2)

Next, the Normalized Difference Vegetation Index (NDVI) was computed using the near-infrared (Band 5) and red (Band 4) reflectance values:

$$NDVI={\displaystyle \frac{Band5-Band4}{Band5+Band4}}$$

(3)

Based on the NDVI values, the vegetation proportion $P_v$ was estimated using the normalized NDVI method:

$$P_v={\displaystyle \frac{NDVI-{NDVI}_{min}}{{NDVI}_{max}-{NDVI}_{min}}}$$

(4)

After that, land surface emissivity $\epsilon$ was calculated using an empirical relationship that incorporates vegetation proportion:

$$\epsilon =0.004\times P_v-0.986$$

(5)

With emissivity estimated, the actual Land Surface Temperature (LST) was calculated by correcting brightness temperature with emissivity and atmospheric effects:

$$LST={\displaystyle \frac{BT}{1+\left(\frac{\lambda ·BT}{\rho }\right)\mathit{ln}\epsilon }}$$

(6)

where $\lambda$ =10.8 μm is the wavelength of emitted radiance and $\rho$ = $hc$ /$\sigma$ = 14380, a constant derived from Planck’s law.

To reduce inter-annual baseline differences and enhance temporal comparability, the LST values were then normalized using a min–max rescaling approach to produce the Relative LST (RLST):

$$RLST={\displaystyle \frac{{{LST}_i-LST}_{min}}{{{LST}_{max}-LST}_{min}}}$$

(7)

Finally, the RLST values [35,36] were classified into five thermal zones using the mean-standard deviation method. These zones include Very Low, Low, Moderate, High, and Very High temperature areas, allowing for spatial pattern interpretation and change identifications across years.

This sequential procedure ensures consistency and comparability of thermal landscape dynamics over time and forms the basis for subsequent cold source identification and corridor modeling.

2.2.3. Cold Island Network Construction and Key Node Identification

To better analyze the spatial pathways and regulatory mechanisms for urban thermal environment mitigation, this study constructed a cold island ecological network for Tianfu New District. This process was based on three sequential steps [37]: the identification of Cold Island Core Sources (CICS) using MSPA, the construction of an ecological resistance surface, and the extraction of corridors and key nodes.

(1)

Identification of CICS with MSPA

To quantitatively identify the spatial morphology and structural connectivity of cold source patches, Morphological Spatial Pattern Analysis (MSPA) was employed using the Guidos Toolbox software (GTB v3.3 Rev.6). MSPA is a raster-based sequence of mathematical morphological operators that classifies the spatial pattern of foreground and background pixels, providing a detailed segmentation of landscape structure.

The MSPA module in Guidos Toolbox was used to classify the structure of cold-source patches into seven morphological types (

Table 2. Seven MSPA Structural Types.

Number	Type	Description
1	Core	Large, stable patches with strong connectivity; key cold source units
2	Edge	Outer boundary of core patches; often affected by external conditions
3	Bridge	Links between core areas; important corridors for thermal flow
4	Loop	Closed shapes around cores; form local buffer structures
5	Branch	Narrow extensions from cores; weakly connected
6	Islet	Small, isolated patches with limited ecological function
7	Background	Non-source areas, often urban or developed land

): Core, Edge, Bridge, Loop, Branch, Islet, and Background. Among them, Core and Bridge types were considered to be stable and functionally significant cold sources, and served as key nodes for subsequent connectivity analysis. By comparing MSPA outputs across multiple years, we examined the spatiotemporal evolution and fragmentation of the cold-source network.

In the MSPA, a binary raster map was required. Grid cells that met the cold-island criteria were classified as foreground (value = 2), representing potential cold-source patches, while all other areas were designated as background (value = 1). An eight-neighbor connectivity rule was applied to define patch adjacency. The edge width was set to 1 pixel (equivalent to 30 m), consistent with the spatial resolution of the Landsat data. This parameter choice effectively captures the direct boundary effects between cold sources and the surrounding warm matrix, without excessively fragmenting the core areas.The resulting Core patches represent zones with strong cooling potential, identified through two conditions [12]: (1) Areas with relative land surface temperature (RLST) values below −1 °C, using the natural breaks method; (2) Land use types including forest, grassland, water, and wetland. Forests and grasslands contribute to cooling through evapotranspiration, while water bodies and wetlands store heat and reduce surface temperature indirectly [38]. These four land types were therefore considered key ecological patches in the blue–green spatial system.

After constructing the core patch network, the study evaluated the overall connectivity level of the cold island system under different patch configurations. Conefor 2.6 was used to quantitatively assess structural connectivity between core cold patches, using two metrics: IIC (Integral Index of Connectivity), and PC (Probability of Connectivity) [39]. Based on this evaluation, the final set of Cold Island Core Sources (CICS) was selected [24]. After the output, core patches were selected as the primary cold sources, as they represent large, contiguous, and relatively stable areas capable of providing sustained and significant cooling services. Bridge patches serve as critical structural linkages between core patches, functioning as potential channels for cool-air flow.

(2)

Construction of the Ecological Resistance Surface

The ecological resistance surface serves as a fundamental basis for simulating cold-air diffusion paths and constructing the cold island network [40]. Its values reflect the degree to which various geographic and anthropogenic factors impede ecological flows. To model the spatial impedance to cool air diffusion, a thermal resistance surface was constructed, serving as a fundamental basis for simulating ventilation paths within the cold island network. The surface values do not represent abstract “ecological” resistance, but rather quantify the degree to which various geographic and anthropogenic factors impede thermodynamic processes, specifically the advection of cool air. Based on urban climate literature and regional characteristics, this study selected four key variables as proxies for components of the surface energy balance: land use type, DEM, road density, and NDVI. The CRITIC method [41] was applied to standardize these factors and determine their objective weights.

Specifically, land use resistance values were assigned according to the thermo-physical properties and ecological functions of each land category: impervious surfaces, with high thermal admittance and storage and a predominance of sensible heat flux, were given high resistance, whereas vegetated areas and water bodies, which promote latent heat flux through evapotranspiration and evaporation, were assigned low resistance. DEM captured the influence of topography on gravity-driven cold-air drainage, with lower elevations and gentle slopes offering less resistance to air movement and pooling. Road density represented a dual thermal penalty, acting both as a proxy for anthropogenic heat emissions from traffic and as an indicator of increased sensible heat flux from asphalt, while also enhancing surface roughness and impeding near-surface airflow. NDVI served as a proxy for vegetation density and health, reflecting the potential for evapotranspirative cooling, a critical mechanism for heat dissipation. All raster data were normalized to a 0–1 scale, and the final weights were calculated based on each factor’s standard deviation and conflict intensity using the CRITIC method (

Table 3. Weight assignment for ecological resistance factors.

Factor	Effect Direction	Effect Direction	Explanation
Land Use	+	0.586	Reflects differences in ecological substrate types
DEM	+	0.166	Indicates terrain variation and energy exchange potential
Road Density	+	0.073	Represents urban disturbance intensity
NDVI	−	0.175	Indicates vegetation coverage, negatively correlated with temperature

).

(3)

Ecological Corridor and Node Identification

Ecological corridors are physical or energy flow pathways that connect ecological source areas within a region. Typically based on linear or strip-shaped landscape units, they support species migration, resource flows, and environmental regulation [42]. In this study, to simulate the dispersal potential of cool air between cold-source patches, we first identified core sources using the MSPA model, then constructed an ecological resistance surface, and applied the Minimum Cumulative Resistance (MCR) model to delineate the spatial structure of cold-air corridors, key connectivity nodes, and potential bottleneck areas. Corridor path computation and visualization were implemented using the Linkage Mapper toolbox, forming the foundational framework of the cold island ecological network in Tianfu New District.

Linkage Mapper produces two core outputs: (1) Least-Cost Corridor maps, presented as raster data, which depict optimal connection paths between all pairs of core cold-source patches and represent the main structure of the thermal flow network; (2) A LinkTable attribute file, which records detailed information on corridor cost, connectivity weight, path length, and source-destination node IDs, providing a quantitative basis for prioritizing corridors and optimizing the ecological network.

Based on spatial scale, structural significance, and ecological function, the cold-air corridors were classified into three levels. Primary corridors: Backbone routes that connect major core cold sources, typically spanning large-scale blue–green infrastructure; they serve as the main pathways for cool-air delivery and ecological stability. Secondary corridors: Links between medium-sized cold patches, enhancing cooperation among local ecosystems and maintaining regional cooling continuity. Tertiary corridors: Connections among small remnant green spaces or isolated patches, which enhance microclimate regulation and increase the spatial diversity of heat dissipation pathways.

This multi-level corridor system constitutes the structural backbone of the cold island network and offers critical spatial insights for identifying key bottlenecks and guiding blue–green infrastructure planning.

2.2.4. Land Surface Temperature Simulation and Scenario Design

To simulate the future evolution of the urban thermal environment, this study employed the Plus model to dynamically predict the spatial distribution of Land Surface Temperature (LST) in Tianfu New District for the years 2035 and 2050. Two development scenarios were established for comparative analysis: the Natural Development scenario (ND) and the Carbon Peak scenario (CP). The objective was to identify spatial trends in the expansion of high-temperature zones and the contraction of cooler areas under different conditions, thereby assessing the regulatory potential of carbon peak policies on the urban thermal environment.

(1)

Plus Model Construction and Driving Factor Selection

The Plus model integrates the temporal transition probabilities of Markov chains with the spatial self-organization capability of cellular automata, making it suitable for simulating the dynamic evolution of spatial phenomena. In this study, LST classification maps from 2000 to 2024 were used as historical inputs to build a transition probability matrix for five temperature categories: low, sub-low, medium, sub-high, and high temperature zones. These matrices were then used to simulate the future distribution of LST categories based on existing spatial patterns.

To enhance the spatial guidance capability of the model, eight driving factors closely related to LST variation were selected as input variables. These include natural topography, ecological conditions, human activities, and transportation accessibility. Specifically, elevation (DEM) and slope reflect terrain variability and heat accumulation characteristics; distance to water bodies indicates the cooling influence of surface water; distance to residential areas represents development intensity and anthropogenic disturbances; distance to railways and main roads reflect transportation network density; GDP indicates regional economic activity intensity; and population density characterizes the spatial distribution of human presence and its thermal impact. All factors were normalized to a common scale with a spatial resolution of 30 m. Initial weights were assigned based on expert judgment and literature synthesis to construct a spatial suitability surface. The CA-Markov model employed a 5 × 5 Moore neighborhood rule for spatial expansion and integrated the transition matrix with driving factors to simulate the dynamic evolution of LST types.

(2)

Scenario Design and Transition Rule Adjustment

To simulate the future evolution of Land Surface Temperature (LST), this study designed two typical development scenarios: the Natural Development Scenario (ND) and the Park City Scenario (PC), aiming to assess potential changes in the thermal environment of Tianfu New District under different pathways.

Natural Development Scenario (ND): This scenario assumes that the city will maintain its current pace of expansion, energy structure, and land use patterns, without any ecological or climate policy interventions. The transition probabilities between LST categories were statistically derived from five temporal LST classification maps (2000–2024), thereby capturing long-term empirical transition trends rather than being arbitrarily assumed. To validate model accuracy, we first simulated the 2024 LST distribution using transition probabilities from the preceding periods, achieving a Kappa coefficient of 0.83, which indicates high reliability. Furthermore, the CA–Markov framework was integrated with multiple spatial driving factors, with weights determined through the CRITIC method and literature synthesis. This combination ensured that the transition probabilities reflected both historical dynamics and realistic environmental constraints, providing a more robust basis for scenario simulation under the inertial development trajectory.

Park City Scenario (PC): This scenario simulates a potential pathway where the thermal environment is mitigated through park-city development strategies and green, low-carbon policy interventions. It is based on the LST transition patterns observed during 2019–2024, with adjusted transition probabilities to reflect policy effects: the probability of transitions from medium to high temperature zones is reduced by 10%, while transitions from medium to low temperature zones increase by 20%; transitions from high temperature zones to medium and low zones both increase by 10%; and transitions from low and sub-low temperature zones to higher temperature categories are reduced by 20% [43]. These modifications aim to capture the potential for thermal relief under ecological protection and energy-saving policies.

It should be noted that the Park City scenario represents an idealized pathway in which ecological and low-carbon policies are assumed to be fully implemented as planned. In reality, the actual effectiveness of these interventions may be constrained by financial resources, governance capacity, and public acceptance. Therefore, the scenario serves not as a guaranteed prediction but as an upper-bound reference case, highlighting the potential benefits of comprehensive ecological interventions under favorable implementation conditions.

Together, these two scenarios represent contrasting development trajectories—inertial expansion versus ecological intervention—and provide a comparative basis for evaluating the future evolution of the urban heat island effect.

3. Results

3.1. Spatiotemporal Evolution Characteristics of LST and RLST

As shown in Figure 3, both the Land Surface Temperature (LST) and Relative Land Surface Temperature (RLST) in Tianfu New District exhibited clear spatiotemporal evolutionary patterns from 2000 to 2025. In the LST zoning maps, medium-temperature zones dominated the region between 2000 (Figure 3b) and 2019 (Figure 3e), while sub-high and high-temperature zones gradually expanded, primarily concentrated in urban core areas, road-intensive zones, and built-up surroundings. By 2024 (Figure 3a), the high-temperature zones significantly expanded into continuous patches, particularly in the northern and central regions, demonstrating a pronounced intensification and concentration of the urban heat island effect. Meanwhile, low-temperature zones were mainly situated in southwestern mountainous areas and regions characterized by intersecting water bodies and forests, with their area decreasing annually, leading to a weakened cold island effect.

RLST maps (Figure 3f–j) further illustrate relative temperature changes across the region. In 2000, extreme heat areas were limited and confined to densely built urban locations. With ongoing urbanization, these extreme heat areas progressively expanded each year. By 2024 (Figure 3f), they had widely spread throughout the north-central city and along main traffic arteries, displaying a “core-radial” expansion pattern. Concurrently, cold and extremely cold zones, predominantly distributed within ecological patches and water-rich areas in the southwest, gradually shrank, exhibiting increasingly fragmented spatial patterns and intensifying spatial contrasts in the thermal environment.

In summary, between 2000 and 2024, Tianfu New District experienced an overall thermal evolution characterized by “medium-temperature dominance, hot-zone expansion, and cold-zone contraction.” Spatial-temporal heterogeneity in thermal distributions increased significantly, with the urban heat island core shifting progressively northward and eastward, and low-temperature zones retreating toward the southwestern corner, resulting in a distinct spatial configuration of an “urban thermal core surrounded by peripheral cold rings.” Such changes closely correlate with intensified land use activities and ecological space compression in the region.

More detailed statistical data (Figure 4) show that from 2000 to 2024, land surface temperature (LST) types in Tianfu New District exhibited a clear trend toward higher temperature categories, indicating an increasingly severe urban thermal environment. In 2000, the medium-temperature zone accounted for 42.12% of the area and was the dominant type. The sub-low (24.38%) and sub-high temperature zones (21.91%) had similar proportions, while low- and high-temperature zones occupied relatively small shares—5.32% and 6.27%, respectively. By 2024, although the medium-temperature zone still had the largest proportion (38.80%), it had slightly decreased compared to 2000. The sub-high temperature zone remained relatively stable (21.62%), while the high-temperature zone increased to 6.97%, indicating a significant intensification of the urban heat island effect. At the same time, the share of the low-temperature zone declined to 3.76%, further demonstrating the ongoing shrinkage of cold-source areas.

In terms of transition pathways, there was substantial conversion of medium-temperature zones into sub-high and high-temperature zones, especially between 2014 and 2019, during which the area of high-temperature zones increased rapidly, reaching 11.51% in 2019. Some low-temperature zones also transitioned directly into medium- or even high-temperature zones, suggesting that urban expansion and changes in land surface cover have significantly disrupted the thermal environment. Conversely, the proportion of reverse transitions (from high to lower temperatures) was relatively low, indicating a weak cooling trend and limited effectiveness in heat island mitigation. This period directly coincides with the official establishment of the Tianfu New District in 2014, which triggered large-scale construction of industrial parks, transportation infrastructure, and residential complexes in the northern and central regions. This rapid land use transformation from vegetated or agricultural land to impervious surfaces is the primary driver behind this significant thermal shift.

Three critical dynamics are evident. First, the dominant pathway of thermal degradation was the substantial conversion of medium-temperature zones into sub-high and high-temperature zones, a clear signature of urban expansion where vegetated surfaces are replaced by heat-retaining impervious materials. This process, which fundamentally alters the surface energy balance by favoring sensible heat over latent heat flux, accelerated markedly between 2014 and 2019, coinciding with the district’s period of intensified development. Second, while smaller in area, the direct transition of low-temperature zones to medium or high-temperature zones is particularly significant, as it represents the direct erosion of core cooling assets like forests and wetlands, often due to disruptive infrastructure projects. Finally, reverse transitions from higher to lower temperatures were minimal, highlighting the quasi-irreversible nature of thermal degradation in built environments. Even with recent ecological restoration efforts, the overall landscape trend demonstrates that the pace of development overwhelms the limited thermal recovery, underscoring the challenge of retrofitting thermal resilience into a rapidly urbanizing fabric.

In summary, the evolution of LST in Tianfu New District is not a simple warming trend but a structured process of thermal landscape restructuring. The analysis of transition pathways reveals a system dominated by irreversible conversions from cooler to hotter states, driven by land development dynamics. This demonstrates a cumulative and intensifying thermal stress, underscoring the urgent need to move beyond compensatory greening to a strategy that actively protects the integrity of remaining cold sources and the connectivity between them.

3.2. Cold-Source Extraction and Evolution of Cold-Corridor Networks

To mitigate ecological risks posed by rising surface temperatures and urban heat island effects, this study identified the Cold Island Core Source (CICS) patches in Tianfu New District from 2000 to 2024 based on Relative Land Surface Temperature (RLST) and ecological resistance surfaces. It further constructed a hierarchical network of cold corridors and revealed their spatial evolution characteristics (Figure 5).

In terms of cold island structure dynamics, the distribution and scale of cold-source patches have undergone significant adjustments with the acceleration of urbanization. In 2000 (

Table 4. Area of Cold Island Core Source (CICS) at different levels from 2000 to 2024.

CICS Level	2000 (km2)	2009 (km2)	2014 (km2)	2019 (km2)	2024 (km2)
Level 1	61.88325022	56.17535333	97.29454859	65.71397268	55.73893
Level 2	41.18752547	107.5291908	117.0509869	131.0495554	133.3324
Level 3	266.3971395	77.71425497	381.8785064	48.54054849	381.6604
Total	369.4679152	241.4187991	596.2240419	245.3040765	570.7318

), cold sources were primarily concentrated in the southwestern hilly areas and peripheral agroforestry zones, displaying a broad but highly fragmented spatial pattern. By 2009 and 2014, cold-source patches in the central and eastern regions had noticeably shrunk, and both first- and second-level cold sources had retreated toward the urban periphery. Particularly before 2014, although third-level cold sources experienced short-term expansion, core ecological spaces were severely disturbed, resulting in a progressively unbalanced ecological structure. By 2019, the cold island system had further degraded, with third-level cold-source areas plummeting to 48.54 km² and first-level cores contracting to 65.71 km², indicating substantial weakening in ecological support capacity. Fortunately, since 2019, with the advancement of the Park City initiative, ecological restoration efforts have begun to take effect. By 2024, the total cold-source area had rebounded to 570.73 km², with third-level cold sources recovering to 381.66 km² and second-level sources expanding significantly, reflecting the effectiveness of ecological buffer restoration around the urban periphery.

The ecological resistance surface, constructed based on land use, topography, and NDVI, reveals that from 2000 to 2024, Tianfu New District experienced a general shift toward higher resistance values (Figure 6). As urban construction progressed, ecological connectivity was progressively diminished, and the resistance to cold-air diffusion and species migration increased significantly. The central urban areas and northern industrial parks emerged as high-resistance zones, characterized by dense impervious surfaces and a low proportion of green space, thus acting as major obstacles to ecological corridor continuity. Particularly after 2014, high-resistance values rapidly spread across the northern and central regions, placing sustained pressure on ecological system integrity. By 2024, although resistance values in the urban core remained high, the surrounding hilly and green areas formed a degree of low-resistance buffer zones, providing spatial foundations for cold-air dispersion and ecological ventilation.

Building upon the above analyses of cold-source identification and resistance surfaces, a hierarchical cold-island ecological corridor network was constructed and its temporal evolution assessed (Figure 7). In 2000, the number of corridors was limited, consisting primarily of a few third-level corridors, with a “point-to-point” connection pattern. By 2009, second-level and third-level corridors had gradually increased, enhancing regional ecological connectivity. A number of significant corridors that span urban subdistricts were developed between 2014 and 2019, with third-level corridors showing functional diversity by growing across rural green spaces and secondary forest zones. Third-level corridors formed the dominant southwest–northeast ventilation spine, while first- and second-level corridors built a “bridge–buffer–transition” framework across northern and east–west urban areas, forming a clear and complementary cold-island network by 2024.

Overall, the spatial structure of Tianfu New District’s cold-source ecological network evolved from a “scattered point-type” pattern in 2000 to a “multi-level, radial, and networked” one by 2024. In particular, the main ventilation axis along the southwest–northeast direction, supported by peripheral second- and third-level corridors, has effectively enhanced regional airflow efficiency and thermal resilience. The ecological corridor network not only strengthened the connectivity and diversity of cold-air diffusion pathways but also provided visual and scientific support for future spatial optimization.

3.3. Simulation of Land-Surface-Temperature Patterns Under Different Development Scenarios

To further evaluate the changing trends of the thermal environment in Tianfu New District under different future development pathways, this study utilized land use simulation results from the PLUS model to construct land surface temperature (LST) zoning maps for two scenarios—Natural Development and Park City—for the years 2035 and 2050. The areas of five temperature categories (LTZ, SLTZ, MTZ, SHTZ, and HTZ) were statistically analyzed (Figure 8).

The results reveal two starkly different futures. Under the Natural Development Scenario, urban heat zones continue to expand while cooler zones significantly contract. In 2035 (Figure 8a), the Medium Temperature Zone (MTZ) dominates with an area of 807.52 km². The High Temperature Zone (HTZ) and Sub-high Temperature Zone (SHTZ) cover 102.81 km² and 257.48 km², respectively, forming a continuous belt of heat in the northern and central urban areas. By 2050 (Figure 8b), with continued urban expansion and ecological space compression, the MTZ shrinks to 694.78 km², while SHTZ and HTZ increase to 300.10 km² and 83.21 km², respectively, showing a clear trend of “temperature upgrading” from medium to high levels. Meanwhile, the total area of cool zones drops to 515.93 km², indicating further deterioration of the thermal environment and a more pronounced urban heat island effect. The Natural Development Scenario projects an unabated intensification of the urban heat island effect, where high-temperature zones expand and coalesce into a continuous heat belt in the urban core. By 2050, this leads to a landscape-wide “temperature upgrading” and a significant contraction of cool zones, indicating severe thermal deterioration.

In contrast, under the Park City Scenario, the expansion of heat zones is effectively curbed, and cooler zones show signs of recovery. In 2035 (Figure 8c), the MTZ area increases slightly to 834.82 km² compared to the natural scenario, while the HTZ area decreases to 103.20 km² and the SHTZ shrinks markedly to 226.28 km², suggesting initial success in controlling the heat island effect. By 2050 (Figure 8d), although the MTZ slightly declines to 745.50 km², the HTZ will be further reduced to 66.02 km²—significantly lower than the 83.21 km² in the natural scenario—and the total area of cool zones rebounds to 503.61 km², reflecting effective ecological regulation. The Park City Scenario demonstrates that this trajectory is not inevitable. Proactive planning interventions effectively curb thermal expansion and show clear signs of recovery in cooler zones. This comparison underscores the significant potential of development pathways that prioritize ecological land preservation and blue-green infrastructure to mitigate urban thermal risks, even amidst ongoing urbanization.

Overall, the urban heat island effect intensifies under the Natural Development Scenario, particularly in the northern areas and along central urban corridors. In the Park City Scenario, although urban construction continues, policies such as ecological land preservation and optimization of blue-green infrastructure help maintain a consistently lower HTZ area, demonstrating the significant potential of this development pathway to mitigate urban thermal risks.

4. Discussion

4.1. Drivers and Mechanisms of the Urban Thermal Pattern

Multi-temporal Landsat remote sensing data from 2000 to 2024 reveal that high-temperature zones in Tianfu New District have continuously expanded outward along the north–central development axis, while low-temperature patches have become increasingly fragmented and contracted. This divergent evolution is attributed to rapid land use change and the degradation of blue–green infrastructure functions. Impervious surfaces with high thermal inertia have progressively replaced permeable, low-heat-capacity vegetated surfaces, resulting in greater heat accumulation and delayed nocturnal cooling. At the same time, ventilation corridors such as riverside greenways and hilly forest belts have been disrupted, weakening convective heat dispersion and amplifying temperature differentials at the interface between “cold” and “hot” sources [44]. Moreover, high-temperature zones overlap with areas of high GDP and population density, confirming the amplifying effect of economic activity and population concentration on urban heat island (UHI) intensity.

By integrating multi-temporal Landsat data with ecological resistance surface modeling, this study reveals how the physical constraints on cold-air diffusion pathways evolve alongside urban expansion. A multi-level ventilation corridor system has gradually emerged along a southwest–northeast axis, highlighting the structural role of ecological planning in regulating urban thermal risk. Cold-island corridors function not only as ecological landscape units but also as critical infrastructure for urban thermal regulation. Their spatial connectivity reflects the city’s capacity to respond to climate variability and its overall resilience level [45].

While the MSPA method has been widely used to identify ecological sources and support biodiversity conservation [46], it has also recently been extended to analyze the spatial structures underlying UHIs [47]. Building on this foundation, the present study proposes an integrated analytical framework that combines cold-source identification from remote sensing with scenario-based LST forecasting using the PLUS model. In contrast to conventional static representations of land surface temperature, this approach emphasizes the dynamic, spatio-temporal, and predictive nature of cold-island systems. By simulating future LST patterns under multiple development scenarios, this research treats temperature not merely as a passive environmental outcome but as a spatial variable actively shaped by planning interventions. This epistemological shift reframes urban climate adaptation from a reactive posture toward a proactive strategy for shaping thermal environments.

Within this corridor-based predictive paradigm of thermal environments, rapidly growing districts such as Tianfu New District can be reinterpreted as experimental fields of spatial–thermal–policy coupling, where land use change, infrastructure configuration, and climate adaptation strategies form a tightly interwoven and dynamically evolving system.

4.2. Planning Implications and Policy Responses

Scenario simulation results indicate that the “Park City” development pathway offers significant ecological intervention potential in regulating the urban thermal environment. By 2050, under this scenario, high-temperature zones could be effectively limited to below 66 km², while the total cold-island area may increase from 245 km² in 2019 to over 571 km². These findings demonstrate that thermal stress in rapidly urbanizing regions is not inevitable but exhibits clear spatial structural dependencies. This validates the structural potential of embedded blue–green networks for regulating heat risks across multi-scalar spatial systems. To translate simulation advantages into practical governance outcomes, a compound governance framework is urgently needed—one that integrates spatial planning, technical interventions, and ecological network design.

First, performance-based constraint indicators should be adopted. For example, limiting high-temperature zones to no more than 5% of total construction land and mandating the provision of at least 55 km² of core cooling sources per 100,000 residents. The study by Marando et al. on 601 European Functional Urban Areas provides a clear, quantitative benchmark, finding that to achieve an average temperature reduction of 1 °C, a city requires at least 16% tree cover, increasing to 32% for a 2 °C reduction [48]. These stringent thresholds would provide measurable benchmarks for future evaluation and regulation.

Second, targeted strategies should be applied based on local conditions. In the northern heat-concentration belt, low thermal-inertia interventions such as high-reflectivity roofs, permeable paving, and tree-lined streets should be deployed to reduce sensible heat accumulation on surfaces [49]. In contrast, the southwest zone, rich in cold sources, should focus on limiting the expansion of impervious surfaces and protecting low-resistance ventilation corridors to preserve and stabilize ecological cooling flows. Furthermore, the planning focus should be on the targeted design of urban green infrastructure [50,51]. A compelling case in point is the framework applied to Fuzhou, China [38]. Instead of uniform greening, they constructed an “urban heat network” to precisely identify critical thermal structures. Based on this, they proposed highly differentiated green infrastructure strategies for different key nodes: for “heating nodes” (pinch points in the network, such as bare ground and construction sites), the priority is to protect small existing green spaces and increase vegetation cover; for “cooling nodes” (barriers to heat diffusion, such as forests and rivers), the core strategy is strict protection to prevent erosion and development. This network-based approach shifts green infrastructure planning from a generic patch-based perspective to precise, targeted interventions, maximizing cooling efficiency within limited urban space.

Third, cold-island corridors should serve as primary structural axes for constructing multi-scale blue–green infrastructure networks, encompassing suburban wetlands, urban green belts, and neighborhood-scale pocket parks. This facilitates continuous cold-air delivery from macro to micro scales [52]. This approach is supported by evidence that peri-urban forests are more effective in providing cooling services than scattered green spaces in the city center, offering both greater temperature reduction and a wider cooling extent [53]. The success of well-planned networks is evident in cities like Ljubljana, a former European Green Capital, which exhibits a very low percentage of area with negative cooling effects, indicating a rational and effective green infrastructure structure [54]. Key corridors must maintain at least a 10-m-wide uninterrupted green belt and incorporate wetland nodes to ensure thermal regulation capacity is embedded from the early development stages—avoiding the need for costly retrofits later [55].

While the “Park City” scenario demonstrates clear ecological benefits, its implementation is not merely a technical exercise but is contingent upon navigating complex institutional, financial, and social barriers. International experiences, where greening initiatives are often hampered by land ownership disputes and funding gaps, highlight that a purely technical approach is insufficient. Recognizing this, we propose a policy feasibility assessment framework to guide the transition of the cold-island network from a model to an on-the-ground reality. This framework addresses three critical dimensions: first, the legal and institutional, where China’s advantage in state-owned urban land is contrasted by complexities in converting collectively owned rural land for ecological use; second, the economic and financial, which demands diversification beyond precarious municipal budgets towards sustainable mechanisms like Public–Private Partnerships and green bonds to avoid creating unfunded “paper parks”; and third, the social and governance aspect, which requires supplementing efficient top-down planning with inclusive models for public participation and stakeholder coordination. Explicitly analyzing these feasibility dimensions is crucial for developing robust strategies that bridge the gap between idealized simulations and the complexities of practical implementation.

Furthermore, the role of large water bodies, particularly reservoirs, warrants special consideration as critical infrastructure for urban thermal regulation. Our findings, which identify these areas as persistent low-resistance zones and origins for primary cold-air corridors, underscore their foundational importance. In the context of future development, the concept of “hydro-technical development” must evolve beyond traditional engineering to become an integral part of climate adaptation strategy. Under the Natural Development (ND) scenario, these reservoirs might face risks from adjacent urban expansion, potentially degrading their buffer zones and compromising their cooling efficiency. In contrast, the Park City (PC) scenario offers a proactive vision where these reservoirs are leveraged as strategic assets. This involves not necessarily building new dams, but rather implementing ecological engineering measures around existing ones: for example, establishing extensive wetland buffer parks along reservoir shorelines, enhancing public access through blue-greenways that double as ventilation corridors, and integrating these systems with “sponge city” principles to manage stormwater runoff. Such hydro-technical planning, therefore, becomes a key mechanism for strengthening the resilience and connectivity of the cold island network modeled in this study, transforming static water bodies into dynamic, multifunctional climate infrastructure.

Ultimately, while the Park City scenario demonstrates substantial mitigation potential, its realization depends on policy continuity, stable financial investment, and societal participation. Future planning should integrate this feasibility framework to ensure that idealized policy targets are grounded in practical and adaptive management strategies.

4.3. Limitations of the Study

Although this study integrates scenario simulation and ecological network modeling with a high degree of systematization, several limitations remain. The 30-m resolution of remote sensing data cannot capture fine-scale thermal heterogeneity such as street-level shading, green roofs, or small water bodies, potentially underestimating localized cooling effects. The PLUS model extrapolates based on historical trends and does not incorporate future extreme climate scenarios; thus, thermal risk under worst-case conditions may still be underestimated. Additionally, while resistance values were objectively weighted using the CRITIC method, they did not dynamically reflect variations in urban greening or population density under different scenarios. Moreover, the transition probabilities in the PLUS model—although empirically derived from long-term observations and constrained by multi-factor drivers—still simplify potential non-linear feedbacks between climate change and urban development.

Moreover, the study period (2000–2024) covers a critical phase of rapid urban growth and policy transition, but remains limited in temporal depth. In the context of climate change, longer-term ecological monitoring and LST trend assessments are necessary to reveal the lag effects, path dependencies, and resilience evolution of cold-source networks. Future research should extend the observation window to 40 years or even half a century, integrating climate scenario modeling, historical image reconstruction, and urban development records to enable deeper analysis of long-term thermal responses. In addition, the “Park City” scenario assumes full realization of ecological and low-carbon policies, which may be overly optimistic under real-world constraints such as funding, governance capacity, and social acceptance.

Further research directions include: (1) Coupling downscaled climate models with the PLUS framework to enhance the robustness of climate adaptation assessments; (2) Incorporating high-resolution UAV or hyperspectral data to refine microclimate performance evaluations; (3) Integrating public health and thermal comfort indicators to expand cold-island planning from ecological functionality to socio-adaptive governance, enabling ecology–health co-benefit frameworks; (4) Constructing long-term urban thermal environment archives and combining them with past land use trajectories and future growth projections to promote cold-source research toward long-term observation, comprehensive evaluation, and adaptive planning. Relatedly, future modeling could further improve robustness by coupling PLUS with downscaled climate outputs and non-linear machine-learning approaches, and by representing policy implementation as graded or probabilistic rather than fully realized.

5. Conclusions

This study established an integrated analytical framework, focusing on the spatio-temporal patterns, drivers, and scenario outlook of cold-island network evolution under rapid urbanization, providing essential insights for urban thermal risk management. Key findings include:

(1) Between 2000 and 2024, rapid urbanization substantially reshaped the thermal environment, characterized by expanded and intensified high-temperature zones, alongside increasingly fragmented and shrinking cooling sources. Consequently, the region’s thermal landscape transitioned from moderate-temperature dominance to moderate-to-high-temperature dominance.

(2) Ecological restoration policies initiated since 2019 effectively reversed previous trends of fragmentation, increasing the total cold-source area to approximately 571 km² by 2024. Cold-island corridor networks evolved from isolated patches to an integrated and hierarchical structure, significantly enhancing ecological connectivity and resilience.

(3) Scenario analyses indicated that proactive ecological planning could effectively control the expansion of high-temperature areas, limiting them to around 66 km² by 2050—approximately 20% less than under natural development conditions. Concurrently, cooler zones would stabilize between 504 and 572 km², demonstrating substantial mitigation potential against future thermal risks.

In summary, these findings highlight the profound impact of urbanization on the thermal environment and validate the effectiveness of proactive ecological planning in mitigating these risks. The integrated analytical framework presented in this study offers practical methodological references and actionable policy insights that extend beyond the study area, guiding sustainable urban growth and climate adaptation in rapidly urbanizing regions.