Next Article in Journal
Benchmark Operational Condition Multimodal Dataset Construction for the Municipal Solid Waste Incineration Process
Previous Article in Journal
Bridging Divides for Sustainable Urban Development: How Public-Space Design Fosters Social Cohesion in a Multiethnic Informal Settlement—The Case of Hesar, Hamedan (Iran)
Previous Article in Special Issue
Investigating Warm-Season Heatwaves Along the Lithuanian Baltic Sea Coast Applying Copernicus Datasets
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 18
Issue 5
10.3390/su18052283
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Influence Mechanism and Spatial Heterogeneity of Urban Spatial Structure on the Thermal Environment: A Case Study of the Central Urban Area of Jinan

by
Junning Wang
,
Xiaoqing Zhang
*,
Qing Li
and
Yuhan Chen
College of Geography and Environment, Shandong Normal University, Jinan 250358, China
*
Author to whom correspondence should be addressed.
Sustainability 2026, 18(5), 2283; https://doi.org/10.3390/su18052283
Submission received: 9 December 2025 / Revised: 22 February 2026 / Accepted: 24 February 2026 / Published: 27 February 2026
(This article belongs to the Special Issue Advancing Sustainability Through Remote Sensing: Addressing Climate Change Challenges)
Browse Figures
Versions Notes

Abstract

Urban expansion and spatial restructuring significantly influence the urban thermal environment. This study investigates the central urban area of Jinan, developing a multi-dimensional spatial structure index system that integrates terrain, 2D/3D morphology, and layout based on multi-source data. Land surface temperature (LST) was derived from remote sensing imagery. Using road networks and triangulated irregular networks (TINs) generated from a digital elevation model (DEM), hybrid analysis units were created. Pearson correlation and bivariate global/local spatial autocorrelation analyses were applied to examine the mechanisms and spatial heterogeneity of how urban spatial structure affects LST. The results showed that (1) LST was strongly associated with urban spatial structure. Among the 12 significantly correlated indicators, building density showed the strongest positive correlation with LST (r = 0.5883), while DEM mean had the strongest negative correlation (r = −0.7444), indicating that compact built-up areas intensified heating, whereas terrain most strongly moderated surface temperature. (2) LST and indicator correlations varied with elevation. LST showed a negative correlation with the standard deviation of DEM, suggesting that greater terrain variability enhances cooling effects. This spatial variation in the dominant drivers of the thermal environment reflects a clear divergence of influencing factors across different elevational zones. The thermal environment exhibits a pronounced north–south split: cooling effects prevail in the south due to terrain, while warming effects dominate in the north due to building forms. (3) Bivariate spatial autocorrelation revealed clear spatial heterogeneity. High–high clustering of LST and spatial structure indicators in the northern plain denoted heat-aggregated zones. Low–low clustering in the topographically complex, sparsely built south formed cold-source zones, and transitional areas showed mixed high–low and low–high clustering. (4) Based on these findings, a zonal governance framework was advocated, prioritizing terrain assessment followed by spatial structure optimization. This promoted a shift from uniform to precise, zone-based thermal environment management, laying a scientific foundation for sustainable spatial planning.

1. Introduction

With the acceleration of urbanization, the supply of urban land resources is becoming increasingly scarce [1]. The city has not only expanded rapidly on the two-dimensional plane but has also expanded into three-dimensional space, improving land-use efficiency [2]. The continuous extension of the city on the two-dimensional and three-dimensional levels has changed the nature of the underlying surface of the city. Impervious surfaces composed of cement, asphalt, concrete, and other chemical materials have replaced the original natural surface [3], resulting in a significant increase in the sensible surface heat flux. In addition, the phenomenon in which the temperature inside the city is higher than suburban temperatures is called the urban heat island (UHI) effect; this is due to factors such as artificial heating generation by urban production and urban living, along with the high heat storage capacity of buildings [4]. Apart from its direct impact, the UHI effect indirectly has a wider impact on human activity by changing, amongst others, the regional atmospheric structure, surface energy balance, and urban hydrology [5], which could harm economic and social development, the suitability of human habitation, and the health of residents in the long-term [6,7]. Nowadays, how to alleviate the urban heat island effect by optimizing the spatial layout of the city and reducing the negative impact of the UHI effect has gradually become a hot topic of research [8,9].
Depending on the data sources, UHIs are divided into atmospheric UHIs and surface UHIs [10,11]. The data sources of atmospheric UHIs are atmospheric temperature data measured by thermometers or fixed weather stations [12], but because the measured data are usually discontinuous point data or discrete linear data, they only have temporal continuity. They do not have long-term stability or continuity over space. Thus, temperature data are usually only used for historical temperature comparisons between years and months [13,14,15]. Urban surface heat islands are characterized by their land surface temperatures (LSTs) [16,17]. In 1972, Rao first proposed the use of thermal infrared remote sensing data to obtain LSTs for studying UHIs [18,19]. Since then, remote sensing data has been used to study UHIs. With continuous improvements in remote sensing technology and surface temperature retrieval algorithms [20], remotely sensed LST data have many advantages, such as strong spatial continuity, high spatial resolution, large areal coverage, and the ability to form a relatively complete meteorological network [21,22], and it has been widely used in the study of the UHI effect.
The continuous extension of the city into the three-dimensional space breaks its original open space pattern, reduces the efficiency of urban heat diffusion, and further intensifies the UHI effect [23,24,25]. Therefore, urban spatial structures have gradually become one of the most important factors affecting urban thermal fields [26,27,28]. Nowadays, most studies use factors such as building height, building density, building volume, and sky visibility [29,30,31] as the main indicators for exploring the correlation between spatial structures and the urban thermal field. However, current research on the parameters of urban spatial structures that affect the UHI focuses mainly on the two-dimensional level. Few studies have been conducted on the three-dimensional spatial structures [32,33], and their research methods mainly involved ordinary linear regression analysis and correlation analysis. These studies ignored the spatial heterogeneities in urban spatial structure parameters and the urban thermal field [34,35,36]. Researching urban spatial structures and urban thermal fields, the choice of spatial research unit plays a vital role, as the relationship between the two may vary with variations in the scale of the spatial research unit [37,38]. In current research on the relationship between the two at the urban scale, the most commonly used research units include grids and blocks [39].
In this study, Jinan’s central urban area was selected as the research area, and the urban thermal field was represented by the urban surface temperature. It aims to investigate the mechanisms through which urban spatial structure influences the thermal environment under complex topographic conditions. Existing research has predominantly focused on flat terrains or the impact of individual factors, while an in-depth spatial analysis of the synergistic and divergent mechanisms between these factors within complex topographic cities remains insufficient. To address this, the study integrated road network-generated blocks with triangulated irregular networks (TINs) derived from a digital elevation model (DEM) to construct hybrid spatial analysis units. A multi-dimensional urban spatial structure indicator system was also established based on multi-source data, incorporating terrain, 2D/3D morphological features, and spatial layout. By employing Pearson correlation analysis and bivariate local spatial autocorrelation analysis, this research not only quantified the correlations between various spatial structure indicators and LST but also highlighted their spatially heterogeneous impact patterns on the thermal environment. This approach elucidated the differentiation in thermal patterns, which were dominated by topography in the south and driven by building morphology in the north. The findings are intended to provide scientific support for optimizing urban spatial layout and implementing precise, zone-based regulation strategies to mitigate the urban heat island effect.
The article is structured as follows: Section 2 introduces the overview of the study area and data sources. Section 3 elaborates on the research methodology, including the approach for delineating spatial research units, the method for retrieving land surface temperature, and the framework for constructing urban spatial structure indicators. Section 4 presents the results of the correlation analysis between urban spatial structure and land surface temperature, as well as the spatial heterogeneity analysis. Section 5 discusses the practical implications of the findings in Section 4 and proposes corresponding spatial planning strategies. It also addresses the limitations of the current study and suggests ways to mitigate similar issues in future research. Finally, Section 6 summarizes the main conclusions and provides a scientific basis for future urban planning.

2. Study Area and Data Sources

2.1. Study Area

This study was conducted in the central urban area of Jinan, Shandong Province, China. The area, approximately 2094 km2 in size, comprises the five districts of Huaiyin, Licheng, Lixia, Shizhong, and Tianqiao. In Jinan’s central urban area, the buildings are densely distributed, with the layout of the buildings and the urban spatial structure having distinct and significant characteristics. The characteristics of the spatial distribution of green spaces and water bodies were relatively clear. Moreover, as the area with the highest UHI intensity, it serves as an ideal sample for investigating how urban spatial structures influence thermal fields.

2.2. Data Sources

In this study, the research area was delineated using administrative boundary vector data of Jinan’s central urban area, which was acquired from Map World of Shandong (http://www.sdmap.gov.cn/, accessed on 30 August 2025). Topographic data were extracted from a Digital Elevation Model (DEM) obtained via Google Earth, with a spatial resolution of 9.55 m. The building vector data used in this study were obtained from the Building height of Asia in the 3D-GloBFP dataset. This dataset is hosted on the Zenodo open-access platform (https://zenodo.org/records/11397015, accessed on 3 September 2025). According to the original study [40], the dataset achieves a building footprint extraction accuracy exceeding 80% and a height estimation accuracy of over 85%, making it suitable for meso- to macro-scale urban morphological analysis. Based on this dataset, the administrative boundaries of Jinan’s central urban area were overlaid to extract building vector data within the corresponding extent, which served as the foundation for subsequent spatial pattern analysis. The resulting dataset includes building height and area attributes, comprising more than 480,000 individual building records. The DEM data were processed using Shui Jing Zhu software (version X3.1). Landsat 8 remote sensing images were obtained from USGS (USGS; https://earthexplorer.usgs.gov, accessed on 4 September 2025). Road network data for Jinan were sourced from OpenStreetMap (https://trac.openstreetmap.org, accessed on 15 August 2025). The overview map of the study area is shown in Figure 1.

3. Methodology

3.1. Construction of Hybrid Space Research Units

To account for the dominant influence of topography over building layout on the thermal environment [41], and to reflect the distinct spatial structures between the southern mountainous area and the northern plain in Jinan, this study constructed hybrid spatial units by integrating road network blocks with Triangulated Irregular Network (TIN) facets derived from Digital Elevation Model (DEM) data.
A zonal strategy was implemented based on the dominant factors influencing land surface temperature (LST). In the southern mountainous area, where topography is the primary control, a TIN was generated from 9.55 m resolution DEM data, with each triangle serving as a fundamental topographic unit. Boundaries of continuous natural features, such as large water bodies, were manually merged to preserve them as independent units. In the northern built-up area, characterized by flat terrain and high building density, basic block units were defined using OpenStreetMap road data, specifically as polygons enclosed by expressways, national and provincial highways, county roads, and major urban roads.
The specific procedures were as follows. In ArcMap 10.2, the five categories of road data were merged, and duplicate segments were removed. Following the method proposed by Xu Haiyang [42], the morphological principles of “dilation and erosion” were applied to extract continuous single-line road centerlines from dual-line roads. This process involved creating an 80 m buffer, merging the buffers and converting them into raster format, applying binarization, and extracting road centerlines through vectorization using the ArcScan toolbox. Subsequently, the “Feature to Polygon” tool was used to generate preliminary block polygons from the processed road network data. To ensure topological continuity across the study area, TIN boundaries were prioritized in areas where TIN units and road-based block polygons intersected. The initially delineated units then underwent area screening and manual adjustments. Units with an area smaller than 90,000 m2 were merged with the adjacent unit sharing the longest boundary. For blocks in the northern built-up area with an area exceeding 3,000,000 m2, manual subdivision was conducted by integrating high-resolution building footprint data, secondary road networks, and Google Earth imagery, based on land cover and urban function consistency. Through the above procedures, this study established a system comprising 2685 spatial units for subsequent analyses.

3.2. Land Surface Temperature Inversion

Urban thermal fields are usually reflected by LST and atmospheric temperature [43]. LST has the characteristics of strong spatial continuity, wide surface coverage, and interaction with air temperature [11]. The LST is generally obtained by inverting the thermal infrared bands in remote sensing images. The most commonly used methods for LST inversion include the radiative transfer equation, separation window, and single-channel algorithm [44]. Among these, the radiative transfer equation method is the most basic surface temperature inversion method. This method can be applied to thermal infrared remote sensing in various bands [45]. In this study, the radiative transfer equation was used to invert the surface temperature. The LST value was reflected by the surface thermal radiation, which was obtained by subtracting the thermal radiation absorbed by the atmosphere and the upward and downward thermal radiation of the atmosphere itself from the total intensity of the thermal radiation received by the satellite sensor. The LST could be calculated by the Planck function, as shown in Equations (1) and (2):
G T S = C λ − C ↑ − τ 1 − ε C ↓ / τ ε
T S = B 2 / ln B 1 / G ( T S ) + 1 − 273
In Equations (1) and (2), the surface-specific emissivity is represented by ε, while C ↓ and C↑ correspond to the downward and upward atmospheric radiance, respectively. G(TS) is the thermal radiance of the black body, Cλ denotes the thermal infrared radiance value received at the satellite sensor, τ stands for the atmospheric transmittance within the thermal infrared band, and TS is the land surface temperature (°C). The B1 and B2 values are obtained from the image source file [46], B1 = 774.89 W/(m2·μm·sr), B2 = 1321.08 K. The atmospheric upward and downward radiance values and transmittance could be obtained by downloading the same atmospheric auxiliary data used in generating the remote sensing data. These data could be downloaded from the USGS EarthExplorer (https://earthexplorer.usgs.gov, accessed on 4 September 2025). This study employed the radiative transfer equation method to retrieve land surface temperature. The retrieval accuracy was influenced by the estimation of land surface emissivity (LSE) and atmospheric correction parameters. Land surface emissivity was estimated based on the NDVI threshold method, while atmospheric parameters were obtained from the NASA online calculator. This scheme represented a relatively accurate method that balanced precision and operational feasibility.

3.3. The System and Quantitative Expression of Urban Spatial Structure Indexes

The urban spatial structure is mainly determined by the distribution of urban topography, building spaces, green spaces, and water bodies. Among them, buildings are the most important constituent elements of a city and the most important factors affecting UHIs. Therefore, according to Chen [41], this study selected buildings as the main body, and an urban spatial structure index system was constructed based on three dimensions: One-dimensional height indicators primarily characterize spatial structure in terms of vertical elevation. Two-dimensional plane indicators mainly describe the horizontal expansion of urban spatial structure. Three-dimensional spatial indicators focus on expressing the volumetric extension of urban buildings in three-dimensional space. The specific indexes are listed in Table 1.
The values of the above indexes were obtained using the spatial analysis function of ArcGIS 10.2. For the raster data of DEM type, we used ArcGIS 10.2, the raster analysis tool of “zonal statistics as table”, and counted the values of the H1 and H2 indexes into the corresponding research units. The P3 vector data were obtained using the raster analysis tool of “zonal statistics as table” in ArcGIS 10.2. For other statistical indicators of vector data (H3~S3), we first assigned the ID of the block grid data to the corresponding building using the identification function in ArcGIS 10.2, and then according to the building, calculated their height, area, volume, and other related indexes according to the newly assigned ID.

3.4. Pearson Correlation Coefficient

The Pearson correlation coefficient is the most commonly used method for studying the degree of linear correlation between elements. Its value interval was [−1, 1]. The closer the value is to 1 (or −1), the stronger the linear positive correlation (or negative correlation) between the elements. The closer the value is to 0, the weaker the correlation between the elements. This is shown in Equation (3).
R = ∑ i = 1 n ( x i − x ¯ ) ( y i − y ¯ ) ∑ i = 1 n x i − x ¯ ∑ i = 1 n y i − y ¯
In Equation (3), R is the correlation coefficient value between variables, n is the sample size of the variable, x ¯ and y ¯ are the average values of x and y, respectively.

3.5. Bivariate Local Spatial Autocorrelation

The bivariate local spatial autocorrelation method can reveal the spatial agglomeration and spatial differentiation of one attribute value in each spatial attribute unit and another attribute value in the adjacent unit. We can also obtain the spatial aggregation characteristics of attributes and a specific spatial research unit that has an abnormality through the LISA (local indicators of spatial associations) map obtained after the analysis. In this study, we adopted the adjacency relationship as a queen connection, which means that two spatial research units are adjacent if they have a common edge or vertex. The specific calculation is shown in Equation (4) [41].
I p q = W p ∑ q = 1 n Z p q W q
In Equation (4), Ipq is the bivariate spatial autocorrelation index value between the spatial units p and q, and Wp and Wq are the normalized values of the attribute value variances in the spatial research units p and q, respectively. The spatial weight matrix, denoted as Zpq, is defined based on the queen contiguity rule between spatial units p and q. The significance of the local indicator Ipq is typically assessed using its Z-score and the corresponding p-value. the critical Z-score thresholds of |1.96| and |2.58| correspond to p-values of 0.05 and 0.01, respectively. In practice, an absolute Z-score greater than or equal to 1.96 (p < 0.05) indicates a statistically significant spatial correlation between the two variables. This study employed GeoDa software (version 1.22) to calculate Local Moran’s I and generate LISA (Local Indicators of Spatial Association) cluster maps for identifying statistically significant spatial clusters and outliers. To account for multiple comparisons and ensure robust significance inference, a conditional random permutation approach was applied [47,48]. For each spatial unit, 999 random permutations of neighboring attribute values were conducted while keeping the value of the focal unit fixed, generating an empirical reference distribution. Statistical significance was determined by a pseudo p-value, derived by comparing the observed Local Moran’s I to this distribution. Results with a pseudo p-value < 0.05 were considered statistically significant and visualized in the LISA cluster map.
The process flow diagram is shown in Figure 2.

4. Results

4.1. Spatial Pattern of LST in the Central City of Jinan

Several studies [49] have shown that the correlation analysis results between urban spatial structures and LSTs in different seasons are similar, but the correlation between the two is more obvious in summer. Based on this, to obtain a more representative summer land surface temperature dataset that would reflect the stable spatial thermal pattern distribution of the city, and to minimize interference from single day extreme weather events, this study averaged the retrieval results from three qualified Landsat 8 images (cloud cover < 5%) acquired on 29 August 2023, 12 June 2024, and 20 June 2024. The results are shown in Figure 2.
It can be seen from Figure 3 that there was a large difference in the LST in Jinan’s central urban area in summer. The LST in summer was between 18.1 °C and 54.3 °C, and the largest difference was approximately 36.2 °C. The average LST in the area covered by buildings was 18.6 °C~50.1 °C. The low-temperature areas were mainly distributed in Jinan’s central urban area, and some were scattered in the Daming Lake, Yellow River, Queshan Reservoir, Wohushan Reservoir, and other water bodies. This phenomenon occurred because the topographic features of Jinan City are high in the south, low in the north, high in the east, and low in the west. The southern part of the central urban area is mountainous with high vegetation coverage and dense water distribution, and the LST in this area was low. The high-temperature areas were mainly distributed in the northern part of Jinan’s central urban area, including the southern part of Tianqiao, the eastern part of Huaiyin, the northern part of Shizhong, Lixia, and the northern part of Licheng; the LST in this area reached its maximum value because of the lower elevation and denser distribution of buildings in the northern part of the central urban area. Overall, the LST in Jinan’s central urban area was higher in the north and lower in the south; this trend was the opposite to that of the terrain of the central urban area, which was higher in the south and lower in the north. The high-temperature areas were mainly found in the areas with higher building densities, indicating that the distribution pattern of the buildings had a certain degree of influence on the LST.

4.2. Influence of Urban Spatial Structure Indicators on Urban Thermal Field

In this study, space units covered by buildings were selected as the research objects, and firstly, the Pearson correlation coefficient was used to study the correlation between urban spatial structure indexes (H1~S3) and LST in the whole area. Secondly, the influence of terrain height on LST was more significant when analyzing the distribution of surface temperature in the central city of Jinan. To further explore the influence of other urban spatial structure indicators on surface temperature in different elevation regions, the study classified H1 using the natural discontinuity method and re-analyzed the correlations between the remaining indexes (H1~S3) in the classified study units and the LST. The natural discontinuity classification method ensures that the internal difference within a set of data is the smallest, and the difference between classes is large. The classification results for H1 are listed in Table 2.
The correlation coefficient R value of the urban surface temperature and urban spatial structure index was calculated for the whole region and in different elevation ranges using MATLAB R2019a, and the significance test value was calculated. The results are presented in Table 3.
Table 3 shows that the correlations between all indexes, except for P3, and the LST reached a confidence level of 0.001 in the whole area, which indicates that the correlations between all other indexes and LST were extremely significant except for the adjacent distance between buildings. H1, H2, H5, H6 and LST were significantly negatively correlated, and the above indicators were mainly related to topography and topographic fluctuation. The higher the topography, the greater the topographic undulation, the lower the urban surface temperature. H3, H4, P1, P2, P4, S1, S2, and S3 all showed extremely significant positive correlations with the LST. Among them, the correlation between H3, P4 and S1 and the LST was more significant, which would indicate that the greater the height, the density and the spatial proportion of buildings in the research unit, the more obvious the effect of increased surface temperature in the area. H3 and H4 showed very significant positive correlations with LST, H5 and H6 had opposite correlations with LST. The reason might be that high-rise buildings are densely distributed in areas with flat terrain, whereas low-rise buildings are mainly distributed in areas with large terrain fluctuations. This phenomenon showed that the influence of terrain on the surface temperature was much higher than that of the buildings.
Table 4 shows that there were obvious differences in the correlations between urban spatial structure indexes and LST in the research units for the five different elevation ranges. At the elevation ranges of level 1, H3, H4, H5, H6, P1, P3, S1, S2, and S3 showed significant positive correlations with LST. Building height and its changes had a significant effect on LST in areas with a flat terrain. When the height of the buildings was higher and the height difference was significant, the LST value in the research area was higher. Among the above indexes, H3 and S1 had the most obvious effects on the improvement of land surface temperature in the area. The higher the building height and volume, the higher the surface temperature in the study area. Furthermore, the negative correlation between P3 and LST suggested that in densely built-up areas with gentle terrain, increasing the distance between buildings could effectively mitigate the urban heat island effect by lowering LST.
At the elevation range of level 2, H2, H5 and H6 were negatively correlated with LST, P2, P4 and S2 showed significant positive correlations with land surface temperature, while the rest of the indexes were not significantly correlated with LST. Thus, in this elevation range, terrain variation, the standard deviation of base area and volume of buildings were still the main factors affecting LST.
In the research areas within the third, fourth, and fifth elevation zones, topographic relief remained the most dominant factor influencing LST. The impact of topographic relief on LST intensified as its variability increased. Within the level 3, H2, H5 and H6 exhibited a significant negative correlation with LST, while the correlations of all other metrics with LST were not statistically significant. At the elevation range of level 4, H2 and H6 showed a significant negative correlation with LST, whereas P4 demonstrated a significant positive correlation. This indicated that within this specific elevation range, less variation in building height and a denser spatial distribution of buildings are associated with a more pronounced increase in regional LST. In the level 5 elevation range, H5 was found to be significantly negatively correlated with LST. A comparative analysis of H5 across all zones revealed a notable shift: it showed a significant positive correlation with LST in the first elevation zone, but a significant negative correlation in the second, third, fourth, and fifth zones. This pattern suggested that in areas of gentle terrain, higher buildings were linked to elevated LST, whereas in higher elevated regions, an increase in building height was associated with a decrease in LST. This finding further underscored that the influence of topography on temperature substantially outweighed that of building characteristics.

4.3. Spatial Heterogeneity of Urban Spatial Structure Indicators on LST

In addition to the influence of urban spatial structures in the same study area, the spatial structures formed in the surrounding area also significantly affected the change in LST. Therefore, this study adopted the bivariate local spatial autocorrelation analysis method to explore the overall influence of surface temperature on the spatial structure indexes in the surrounding area and the spatial heterogeneity of its influence. In this study, LST was used as the first variable, and the spatial structure indexes in the surrounding area were used as the second variables. Using the bivariate local Moran’s I tool in the GeoDa software, the Moran’s I value and the test value Z between the two variables were calculated. Table 4 presents the results.
According to Table 5, the absolute Z values between all indexes and LST were all greater than 2.58, reaching a confidence level of less than 0.01. This could indicate that the correlations between LST and all indexes of urban structure in the surrounding area, were extremely significant. The Moran’s I values between the LST and the H1, H2, H5, H6, and P3 indexes in the surrounding area were all less than 0. We realized that the LST decreased with an increase in the above values of the indexes in the surrounding area. The Moran’s I values between LST and H3, H4, P1, P2, P4, S1, S2, and S3 in the adjacent area were all greater than 0, showing very significant positive correlations. Among the above indexes, the average DEM, the average absolute height of buildings, the standard deviation of DEM and the standard deviation of absolute height of buildings in the adjacent area were more obvious in reducing the LST of the area. The indexes that could effectively improve the LST in the area were the average height of buildings, the building density and the average volume of buildings in the adjacent area.
To further analyze the spatial differences in the influence of urban spatial structure indexes on urban surface temperature in the surrounding regions, the bivariate local spatial autocorrelation analysis could obtain the LISA maps between LST and urban spatial structure indexes, which could explore the spatial heterogeneity of urban spatial structure indexes and urban thermal fields. For convenience of comparison, similar research results were analyzed together, and the results are shown in Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8 and Figure 9.
According to Figure 4, in the bivariate agglomeration analysis between the LST and the average DEM and the average absolute height of buildings, the proportion with no significant correlation was the most widely distributed in space. Secondly, the high–low outlier areas were relatively widespread in spatial distribution, primarily located in the northwestern part of the building zone. Comparison with the topographic map of Jinan’s central urban area revealed that these areas featured low altitude, minimal terrain fluctuation, and relatively dense building distribution. Consequently, the LST demonstrated significantly high–low outlier characteristics with both the average DEM and the average absolute height of buildings in adjacent areas. Areas exhibiting a low–low clustering spatial structure were found in the northern part of the building zone, where building distribution was relatively dispersed and the terrain was gentle. Within these areas, the LST, along with the two indicators—the average DEM and the absolute building height in adjacent regions—demonstrated a significant low–low clustering spatial pattern. In contrast, the high–high cluster areas were less frequently observed in the spatial distribution and were primarily concentrated in the central–southern part of Jinan’s central urban area. Within this region, the terrain began to rise in elevation and building distribution became relatively dense. In the northern fringe of the building zone, the LST exhibited a low–low clustering characteristic with these two indicators in adjacent areas. This pattern occurred because this area featured sparse building distribution and gentle terrain, resulting in lower LST. In the southern part of the central building zone, the LST showed significant low–high outlier characteristics with both the average DEM and the average absolute height of buildings in adjacent regions. These areas experienced substantial terrain fluctuation and higher altitude, causing the LST to be significantly influenced by topography and thus remaining relatively low.
Figure 5 shows that, in the bivariate agglomeration analysis between LST and the average building height and standard deviation of building height, the highest proportion of spatial correlations was insignificant. Secondly, in the southern, northwestern, and southwestern parts of the study area, LST and the two indexes showed significant low–low cluster and high–low outlier characteristics. The area was located at the edge of the built-up area, with a sparser distribution of buildings; therefore, the average building height and its variation in vertical space were small. In the central part of the built-up area, LST demonstrated noticeable low–high and high–high clustering spatial characteristics with the average building height and the standard deviation of building height in the adjacent areas. This was attributed to the dense distribution of buildings in this region, where the average building height was greater, and the variation in building heights was more obvious. Therefore, the surface temperature varies because of the different terrains.
Figure 6 illustrates that the results of the bivariate analysis between the LST and the standard deviation of the DEM; the standard deviation of the absolute height of the buildings was significantly affected by the terrain and the layout of buildings in the vertical space. The degree of terrain undulation increased and the vertical spatial difference in buildings was larger in the southern part of the built-up area, so the LST and these two indexes were high–high clusters and low–high outliers. In contrast, the terrain was flatter and the buildings had less vertical spatial differences in the north of the built-up area, and the LST and the two indexes showed relatively significant characteristics of the low–low cluster and the high–low outlier. In terms of spatial distribution, it was necessary to analyze the spatial correlations between LST and these two indexes in different regions according to indicators such as the height of the building and its degree of change.
It can be seen from Figure 7 that in the bivariate analysis between the LST and the average of the building footprint mean and the standard deviation of building footprint mean, the proportion of the distribution with no significant spatial correlation was the highest. In the southern part of the built-up area, the LST and the two indexes in the adjacent area were significant low–low clusters. The reason for this phenomenon was that because these areas were located on the edges of mountains or built-up areas, the LST was less affected by topography and building density, and at the edges of the built-up area, the volume of buildings and the size of buildings and their differences were small. The high–high clusters were mainly distributed in the periphery of the central built-up area. The reason for this phenomenon might be that this area contains more green space and water bodies, as well as more urban parks, factories, and other buildings, resulting in a large difference in the base area of buildings.
In Figure 8, the bivariate analysis between the LST and the proximity distance of buildings in the adjacent area shows obvious spatial distributions of the low–low cluster low outlier and the low–high outlier in parts of the southern mountainous region. Building distribution was relatively sparse in the southern mountains, where temperature was primarily influenced by topographic factors. In the town areas of the southern mountainous region, where building distribution was relatively dense, a notably significant low–high outlier was also observed. Scattered high–low and high–high clustering characteristics were present at the edges of the built-up areas.
From Figure 9, we can see that in the bivariate agglomeration analysis between LST, building density, and plot ratio, there were significant spatial characteristics of the high–high cluster between the surface temperature and these two indexes in the adjacent area. In the surrounding edge areas of buildings, the spatial distribution characteristics of the low and high–low outliers were mainly evident. The main reason for the difference in distribution was the central building-intensive area, where buildings were the most densely distributed and land use intensity was the greatest. The surrounding edges of the built-up area were generally close to the mountain, or there were many water bodies, green spaces, cultivated land, workshops, as well as unused land. Therefore, the distribution of buildings was sparse and the intensity of land use was low. The differences in surface temperature were mainly due to the differences in building density.
Figure 10 shows the results of the bivariate agglomeration analysis between LST and the average building volume and the standard deviation of the building volume. The spatial correlation between surface temperature and the two indexes in the adjacent area was the widest but was not significant. At the southern of the built-up area, the land surface temperature and the two indexes were the low–low clusters and the high–low outliers. There were few buildings in this area, mostly old-style buildings and bungalows, so the building volume and its differences were small. Topography was the main factor causing the variations in land surface temperature. However, in the central built-up area, the spatial distribution between LST and the two indexes were the high–high clusters and the low–high outliers. This was mainly due to the differences in building density.

5. Discussion

Although the land surface temperature retrieved by the above method contains some degree of uncertainty, this study aims to reveal the relative spatial pattern of summer LST rather than absolute values. Moreover, the associated errors are spatially consistent, thus having limited impact on the core conclusions regarding spatial heterogeneity. Due to the lack of synchronous in situ measurements, direct validation was not conducted in this study. To assess the reasonableness of the retrieval results, the adopted method was compared with previously validated approaches used in similar urban thermal environment studies [50,51]. The results show that our method is consistent with other studies, and the derived relative spatial pattern of summer LST in Jinan’s central urban area aligns with existing findings, indirectly supporting the reliability of our results. On this basis, this study further confirms the general relationship between urban structure and the urban heat island effect. Based on three Landsat 8 images from the summers of 2023 and 2024, this study characterized the overall spatial pattern of the summer thermal environment in Jinan’s central urban area by retrieving and averaging land surface temperature. Although the two years data impose certain limitations in reflecting long-term trends and the impacts of extreme weather events, the primary focus of this research is to reveal the stable influence of urban spatial structure on the thermal pattern, rather than to analyze the dynamic process of interannual variation. By averaging multiple summer images with minimal cloud cover, the summer temperature variations within Jinan’s central urban area were integrated, thereby highlighting a more stable spatial pattern. Previous studies [50,51], which predominantly used single-date or two-year image analyses and compared them with earlier LST patterns, have similarly shown that the spatial distribution of LST in Jinan remains consistent over multiple years. This indicates that the thermal spatial pattern dominated by the urban underlying surface is stable. Therefore, the use of multi-temporal averaging helps to better extract this stable feature, aligning with the objective of this study. Future research could incorporate longer time series of remote sensing and meteorological data to further distinguish between the thermal field shaped by urban morphology and the anomalous thermal field induced by weather fluctuations.
This study employs spatial heterogeneity analysis to quantify the distinct contributions of topography and building morphology to the urban thermal environment across Jinan. This analysis is framed by the city’s unique context, where higher terrain in the south contrasts with a denser built environment in the north. In the southern mountainous regions, LST is predominantly governed by topography. The cooling mechanisms in these areas stem from the vertical temperature lapse rate, enhanced ventilation and heat dissipation facilitated by the terrain, and greater vegetation coverage. Statistical models confirm that within this region, topographic factors demonstrate significantly higher explanatory power for variations in the thermal environment compared to building morphological factors, thereby establishing topography as the dominant controlling element. Conversely, in the built-up areas of the northern plains, the urban thermal pattern is dominated by intensive spatial configurations characterized by high density and high floor area ratios. Existing research has utilized geographically weighted regression models to uncover the non-stationary effects of spatial morphology on surface temperature [51,52]. Building on this, the present study employs a combination of correlation analysis and bivariate spatial autocorrelation analysis to refine this observed heterogeneity into a clear spatial differentiation pattern: cooling dominated by topography in the south contrasts with warming driven by building morphology in the north. It clarifies the operational boundaries and transition conditions of these two dominant forces within a city of complex terrain, representing a spatially refined extension of classical theory within a specific urban context.
The spatial heterogeneity patterns identified in this study, particularly the spatial transition in the influence weights of terrain and buildings on the thermal environment, and the identification of key areas like high–high heat clusters and low–low cold sources—primarily rely on a combined analysis of Pearson correlation and bivariate local spatial autocorrelation. To ensure the validity and reliability of the conclusions, this study addresses inherent key issues in spatial analysis. First, regarding edge effects, our analysis focused on core clustering areas within the study region with complete neighborhoods, such as the large northern high–high heat cluster where local Moran’s I values are stable, reliably revealing the coupling relationship between building density and high temperature. Sporadic significant units at the edges were not emphasized, considering potential statistical artifacts from boundary effects. Second, conclusions based on fixed spatial units are inherently related to their scale and zoning method. Addressing the modifiable areal unit problem (MAUP) encountered, this study primarily identifies the macro-scale spatial pattern of Jinan’s central urban area, the southern terrain leads to dominant cooling effects, while the northern building morphology results in predominant warming. This pattern is rooted in the fundamental geographical reality of higher southern terrain and flatter northern terrain, coupled with the basic urban development reality of higher intensity in the north and lower in the south. Its core driving mechanisms, based on topographic climatology effects and urban canopy effects, would not undergo fundamental changes due to minor adjustments in the size or configuration of the analytical grid. The hybrid zoning method adopted in this study better captures the macro-scale spatial pattern of interaction between natural terrain and the artificial built-up area. Finally, the study ensured the authenticity of identified spatial patterns through considerations for multiple comparisons. When conducting simultaneous significance tests across numerous spatial units, the study employed the criteria of spatial continuity and geographical consistency to mitigate against random fluctuations and outliers. The paper focuses on and discusses significant units that form continuous spatial clusters rather than isolated points. For instance, the high–high cluster in the central urban area forms a continuous spatial entity, which is unlikely to be a result of random variation. Furthermore, the spatial distribution of these continuous clusters shows reasonable consistency with the known geographical context of Jinan’s central urban area, such as the high-density built-up core and the orientation of major mountain ranges. The observed overlap between heat clusters and Jinan’s old urban core, and the correspondence between cold areas and the southern mountains, further geographically validates the authenticity of the statistical patterns. The spatial heterogeneity patterns analyzed and discussed are not statistical artifacts but genuinely reflect the spatial coupling relationship between urban spatial structure and the thermal environment in Jinan. This lays a solid foundation for proposing targeted urban planning directions and zoned governance strategies in the next step.
Spatial heterogeneity analysis not only helps to validate general patterns but, more importantly, enables the precise identification of areas that deviate from these norms or present anomalies. Thus, the analysis serves as a direct basis for informing subsequent planning interventions. As indicated, high–high heat clusters are predominantly located in the core areas of Jinan’s old urban districts, such as southern Tianqiao District and northern Lixia District. After years of development, these areas have become warning zones for the urban thermal environment. Without intervention, new urban development may replicate its conditions. For such areas, urban renewal should focus on enhancing ventilation capacity and urban green space. Under the premise of maintaining an overall balance in total development intensity, it is recommended to appropriately reduce development intensity in specific plots and reallocate the freed-up space. Existing research also suggests implementing rooftop greening, vertical greening, and creating pocket parks on vacant or underutilized land in these areas to rapidly increase green coverage, raise NDVI, and offset the thermal effects of dense construction. Low–low cold source areas are mainly located in the southern mountainous regions [52,53]. For these zones, ecological protection red lines should be delineated to restrict development intensity and building height. Studies also indicate that the relatively low temperature advantage of the southern mountains should be utilized to plan cold air corridors, facilitating the penetration of cool air into the urban area to alleviate pollution and stifling heat during stable weather conditions. For the transitional zones exhibiting high–low aggregation at the mountain front, their core ecological function is maintaining ventilation corridor patency. Therefore, building controls are necessary: first, regulating building morphology, suggesting staggered or point-group layouts; second, controlling building density and continuity to avoid large, contiguous building blocks that could form wind barriers and obstruct ventilation. Anomalous areas like low–high or high–low clusters reveal the local regulatory effects of specific factors such as water bodies, large reservoirs, green spaces, and special land uses on the thermal environment. Subsequent planning should conduct detailed studies of such areas, summarize effective experiences, and gradually apply them in planning for similar contexts.
Using Jinan as a topographic case study, the findings regarding specific spatial patterns are inherently local in nature. However, the core mechanism through which urban spatial structure influences the thermal environment offers valuable insights for understanding thermal differentiation in other cities. This mechanism was revealed using a methodology that employs hybrid research units, establishes multidimensional indicators, and combines correlation analysis with bivariate spatial autocorrelation. For cities with pronounced topographic features, the methodology developed in this study can be directly applied to identify terrain-dominated ventilation corridors and heat-prone basins. For flat cities with internally heterogeneous structures, the analytical focus can shift primarily to how differentiation in building morphology influences urban heat island patterns. A key implication of this research is to advance thermal environment management from undifferentiated approaches toward spatially differentiated, factor-specific zonal governance. Other cities can adopt a similar approach to identify internal high-temperature risk areas and ecological cool sources, thereby formulating targeted, zone-based intervention strategies. Future research may extend this framework to additional cases, adapting it to locally dominant meteorological processes and underlying surface characteristics to systematically examine and enhance its broader applicability.
It must also be acknowledged that this study has several limitations. First, the land surface temperature data were derived from the mean values of only two summer images. While effective in depicting the spatial distribution of the urban thermal field, the limited temporal span constrains the long-term representativeness of the findings. Future research should construct long-term temperature datasets spanning multiple consecutive years and seasons. This would enable a deeper exploration of the interannual variability and long-term evolution of the urban heat island effect, as well as its dynamic linkages to climate change and urban development stages. Second, the hybrid zoning method, which integrates natural and administrative boundaries, better reflects the actual structural characteristics of Jinan’s central urban area but inevitably encounters the Modifiable Areal Unit Problem (MAUP). Existing research indicates that, based on analytical units at different scales, the characteristics of the impact of spatial structure on land surface temperature exhibit significant differences [52,54]. Future studies could conduct comparative and cross-validation analyses across multiple spatial scales to systematically assess the robustness of this study’s conclusions, thereby mitigating potential errors inherent in single-scale analysis. Third, while the models in this study identified correlations between urban structural factors and UHI intensity, they could not strictly separate the quantitative influences of other variables— such as vegetation shading and transpiration effects, as well as anthropogenic heat emissions from industrial and transportation sources. Future work could attempt, based on more refined land cover data and energy consumption statistics, to apply attribution analysis models to quantify the contribution rates of various influencing factors. This would offer more scientifically grounded support for precise policy formulation and differentiated planning interventions.

6. Conclusions

Based on the above research, we could draw the following conclusions:
(1)
LST is significantly influenced by urban spatial structure. Correlation analysis indicates that five indicators, the mean and standard deviation of the DEM, the mean and standard deviation of absolute building height, and building adjacency distance, show a clear negative correlation with LST. In contrast, eight indicators, the mean and standard deviation of building height, the mean and standard deviation of building footprint area, building density, mean building volume, and floor area ratio, exhibit a significant positive correlation with LST. Among these, the mean DEM and the mean absolute building height play a particularly prominent role in reducing LST both locally and in adjacent areas, while the mean building height, building density, and mean building volume significantly intensify local thermal effects.
(2)
The correlation between LST and urban spatial structure indicators varies systematically with elevation. In low elevation, flat areas (e.g., at elevation levels 1 and 2), LST is primarily and positively influenced by building-related indicators such as building height, density, and footprint area. In contrast, in high-elevation areas with significant topographic relief (e.g., elevation levels 2 to 5), LST shows a significant negative correlation with the standard deviation of DEM, and the cooling effect strengthens with greater terrain variation. Through correlation analysis and bivariate spatial autocorrelation analysis, the study further clarifies the heterogeneity of the thermal environment as a distinct north–south divergence: cooling dominated by topography in the south contrasts with warming driven by building morphology in the north. In the context of a city with complex terrain, the research delineates the operational scopes, spatial boundaries, and transition conditions of these two dominant forces within the urban area. This not only reveals the localized differences in the formation mechanisms of the urban thermal environment but also represents a spatially refined extension of classical theory within a specific urban context.
(3)
The impact of urban spatial structure on LST exhibits significant spatial heterogeneity. Bivariate clustering analysis shows that in the densely built-up, flat northern areas, LST forms high–high clusters with most two-dimensional and three-dimensional morphological indicators. In the topographically complex, sparsely built southern mountainous areas, low–low cold-source clusters are dominant. In transitional zones, anomalous clustering patterns such as high–low or low–high are observed. These results demonstrate that integrating global correlation with bivariate local spatial autocorrelation effectively identifies heat clusters, cold sources, and transitional zones, shifting from a holistic understanding to precise spatial identification. This provides methodological support for systematically discerning spatial heterogeneity in the urban thermal environment.
(4)
Based on these findings, this study proposes a governance principle of prioritizing terrain assessment followed by precise optimization of urban spatial structure. Specifically, in densely built-up areas with gentle terrain, urban renewal should promote intensity transfer and morphological optimization, focusing on controlling building density and floor area ratio. In areas with significant topographic relief, it is essential to leverage topographic advantages to ensure urban ventilation potential and maintain unimpeded natural cooling pathways. For different types of transitional zones, differentiated graded management strategies should be implemented. These strategies reflect a shift from uniform management to zoned, precise regulation, offering a scientific basis for thermal environment governance and spatial planning in mountainous cities.

Author Contributions

Formal analysis, Q.L.; Writing—original draft, J.W.; Writing—review & editing, X.Z.; Visualization, Y.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Zhang, S.; Cao, X. Different pathways, same goal: Configuring different types of policy tools to improve the urban land green use efficiency. Humanit. Soc. Sci. Commun. 2025, 12, 1571. [Google Scholar] [CrossRef]
  2. Liu, X.; Wu, X.; Li, X.; Xu, X.; Liao, W.; Jiao, L.; Zeng, Z.; Chen, G.; Li, X. Global Mapping of Three-Dimensional Urban Structures Reveals Escalating Utilization in the Vertical Dimension and Pronounced Building Space Inequality. Engineering 2025, 47, 86–99. [Google Scholar] [CrossRef]
  3. Bai, Y.; Wang, M.; Yan, Y.; Wang, H. Exploring the impact of 2D/3D urban morphology on land surface temperature within the diurnal cycle in Tianjin. Sci. Rep. 2025, 15, 39740. [Google Scholar] [CrossRef]
  4. Kim, Y.; Yeo, H.; Kim, Y. Estimating urban spatial temperatures considering anthropogenic heat release factors focusing on the mobility characteristics. Sustain. Cities Soc. 2022, 85, 104073. [Google Scholar] [CrossRef]
  5. Xu, Y.; Jia, R.; Liu, J.; Han, D.; He, T.; Xu, X.; Liu, L.; Sun, Z.; Qiao, Z. The population exposure risk of urban heat island effect: From the perspective of urban spatial expansion in China. Build. Environ. 2024, 258, 111565. [Google Scholar] [CrossRef]
  6. Zhao, Q.; Tao, L.; Song, H.; Lin, Y.; Ji, Y.; Geng, J.; Yu, K.; Liu, J. Mitigation pathways of urban heat islands and simulation of their effectiveness from a perspective of connectivity. Sustain. Cities Soc. 2025, 124, 106300. [Google Scholar] [CrossRef]
  7. Amani, M.J.; Tanzadeh, R.; Moghadas Nejad, F.; Kabiri Nasrabad, M.M.; Chalabii, J.; Movahedi Rad, M. Urban Sustainability Through Pavement Technologies: Reducing Urban Heat Islands with Cool Pavements. Buildings 2025, 15, 504. [Google Scholar] [CrossRef]
  8. Yang, M.; Ye, P.; He, J. Green and Blue Infrastructure for Urban Cooling: Multi-Scale Mechanisms, Spatial Optimization, and Methodological Integration. Sustain. Cities Soc. 2025, 129, 106501. [Google Scholar] [CrossRef]
  9. Gao, D.; Wang, Z.; Gao, X.; Chen, S.; Chen, R.; Gao, Y. Constructing an ecological network based on heat environment risk assessment: An optimisation strategy for thermal comfort coupling society and ecology. Sustainability 2024, 16, 4109. [Google Scholar] [CrossRef]
  10. Almeida, C.R.d.; Teodoro, A.C.; Gonçalves, A. Study of the urban heat island (UHI) using remote sensing data/techniques: A systematic review. Environments 2021, 8, 105. [Google Scholar] [CrossRef]
  11. Voogt, J.A.; Oke, T.R. Thermal remote sensing of urban climates. Remote Sens. Environ. 2003, 86, 370–384. [Google Scholar] [CrossRef]
  12. Zhou, D.; Sun, S.; Li, Y.; Zhang, L.; Huang, L. A multi-perspective study of atmospheric urban heat island effect in China based on national meteorological observations: Facts and uncertainties. Sci. Total Environ. 2023, 854, 158638. [Google Scholar] [CrossRef]
  13. Gawarkiewicz, G.; Chen, K.; Forsyth, J.; Bahr, F.; Mercer, A.M.; Ellertson, A.; Fratantoni, P.; Seim, H.; Haines, S.; Han, L. Characteristics of an advective marine heatwave in the Middle Atlantic Bight in early 2017. Front. Mar. Sci. 2019, 6, 712. [Google Scholar] [CrossRef]
  14. Ren, Z. A Literature Review on the Use of Weather Data for Building Thermal Simulations. Energies 2025, 18, 3653. [Google Scholar] [CrossRef]
  15. Lai, Y.; Dzombak, D.A. Use of historical data to assess regional climate change. J. Clim. 2019, 32, 4299–4320. [Google Scholar] [CrossRef]
  16. Radoux, J.; Dominique, M.; Hartley, A.; Lamarche, C.; Bos, A.; Defourny, P. Land Cover Types Drive the Surface Temperature for Upscaling Surface Urban Heat Islands with Daylight Images. Remote Sens. 2025, 17, 2815. [Google Scholar] [CrossRef]
  17. Tanoori, G.; Soltani, A.; Modiri, A. Machine learning for urban heat island (UHI) analysis: Predicting land surface temperature (LST) in urban environments. Urban Clim. 2024, 55, 101962. [Google Scholar] [CrossRef]
  18. Zhang, J.; Wang, P.; Jin, A. Multidimensional characteristics of urban green space and its impact in mitigating urban heat Island effects: A case study of Guangzhou. Sci. Rep. 2025, 15, 39959. [Google Scholar] [CrossRef]
  19. De Razza, S.; Zanetti, C.; De Marchi, M.; Pappalardo, S.E. Mapping urban heatwaves and islands: The reverse effect of Salento’s “white cities”. Front. Earth Sci. 2024, 12, 1375827. [Google Scholar] [CrossRef]
  20. Sibo, D.; Chen, R.; Zhaoliang, L.; Mengmeng, W.; Hanqiu, X.; Hua, L.; Penghai, W.; Wenfeng, Z.; Ji, Z.; Wei, Z. Reviews of methods for land surface temperature retrieval from landsat thermal infrared data. Natl. Remote Sens. Bull. 2021, 25, 1591–1617. [Google Scholar] [CrossRef]
  21. Tang, W.; Zhou, J.; Ma, J.; Wang, Z.; Ding, L.; Zhang, X.; Zhang, X. TRIMS LST: A daily 1 km all-weather land surface temperature dataset for China’s landmass and surrounding areas (2000–2022). Earth Syst. Sci. Data 2024, 16, 387–419. [Google Scholar] [CrossRef]
  22. Ma, J.; Zhou, J.; Liu, S.; Göttsche, F.-M.; Zhang, X.; Wang, S.; Li, M. Continuous evaluation of the spatial representativeness of land surface temperature validation sites. Remote Sens. Environ. 2021, 265, 112669. [Google Scholar] [CrossRef]
  23. Lin, Z.; Xu, H.; Yao, X.; Yang, C.; Yang, L. Exploring the relationship between thermal environmental factors and land surface temperature of a “furnace city” based on local climate zones. Build. Environ. 2023, 243, 110732. [Google Scholar] [CrossRef]
  24. Yu, H.; Yu, L.; Zhang, C.; Qi, Y. How does urbanization process affect urban heat island effect? Interpretation of 31 cities in China based on local climate zones. Environ. Res. Lett. 2025, 20, 114013. [Google Scholar] [CrossRef]
  25. Hadi, Z.A.; Dom, N.C.; Ishak, A.; Salleh, S.; Abdullah, S.; Precha, N. Geospatial Dynamics Of Dengue Fever Density In Kuantan, Malaysia: GIS-Based Approach. Geogr. Environ. Sustain. 2025, 18, 54–64. [Google Scholar] [CrossRef]
  26. Zhou, W.; Huang, G.; Cadenasso, M.L. Does spatial configuration matter? Understanding the effects of land cover pattern on land surface temperature in urban landscapes. Landsc. Urban Plan. 2011, 102, 54–63. [Google Scholar] [CrossRef]
  27. Yang, J.; Yang, Y.; Sun, D.; Jin, C.; Xiao, X. Influence of urban morphological characteristics on thermal environment. Sustain. Cities Soc. 2021, 72, 103045. [Google Scholar] [CrossRef]
  28. Azhdari, A.; Soltani, A.; Alidadi, M. Urban morphology and landscape structure effect on land surface temperature: Evidence from Shiraz, a semi-arid city. Sustain. Cities Soc. 2018, 41, 853–864. [Google Scholar] [CrossRef]
  29. Wang, L.; Li, R.; Jia, J.; Zhai, Y.; Tian, Y.; Xu, D.; Chen, Y.; Zhang, X.; Ren, Z.; Ye, Z.; et al. Integrating morphology and vitality to quantify seasonal contributions of urban functional zones to thermal environment. Sustain. Cities Soc. 2025, 120, 106136. [Google Scholar] [CrossRef]
  30. Huang, C.; Liu, K.; Ma, T.; Xue, H.; Wang, P.; Li, L. Analysis of the impact mechanisms and driving factors of urban spatial morphology on urban heat islands. Sci. Rep. 2025, 15, 18589. [Google Scholar] [CrossRef]
  31. Han, W. Analyzing the scale dependent effect of urban building morphology on land surface temperature using random forest algorithm. Sci. Rep. 2023, 13, 19312. [Google Scholar] [CrossRef] [PubMed]
  32. Liu, S.; Zhao, J.; Chen, Y.; Zhang, S. Urban Morphology Classification and Organizational Patterns: A Multidimensional Numerical Analysis of Heping District, Shenyang City. Buildings 2024, 14, 3157. [Google Scholar] [CrossRef]
  33. Xiao, Z.; Peng, Z.; Yu, Z.; Liu, X. Generating natural cities using 3D road network to explore living structure: A case study in Hong Kong. Smart Cities 2023, 6, 1485–1506. [Google Scholar] [CrossRef]
  34. Jian, L.; Xia, X.; Wang, Y.; Liu, X.; Zhang, Y.; Yang, Q. Spatiotemporal dynamic relationships and simulation of urban spatial form changes and land surface temperature: A case study in Chengdu, China. Front. Public Health 2024, 12, 1357624. [Google Scholar] [CrossRef]
  35. Kang, X.; Pan, J. Evaluating The Spatial-Seasonal Variation, Heterogeneity and Distribution of Urban Thermal Environment: Case Study of Nanjing, China. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2019, 42, 95–102. [Google Scholar] [CrossRef]
  36. He, Q.; Yan, M.; Zheng, L.; Wang, B. Spatial stratified heterogeneity and driving mechanism of urban development level in China under different urban growth patterns with optimal parameter-based geographic detector model mining. Comput. Environ. Urban Syst. 2023, 105, 102023. [Google Scholar] [CrossRef]
  37. Guo, A.; Yang, J.; Xiao, X.; Xia, J.; Jin, C.; Li, X. Influences of urban spatial form on urban heat island effects at the community level in China. Sustain. Cities Soc. 2020, 53, 101972. [Google Scholar] [CrossRef]
  38. Chen, Y.; Yang, J.; Yu, W.; Ren, J.; Xiao, X.; Xia, J.C. Relationship between urban spatial form and seasonal land surface temperature under different grid scales. Sustain. Cities Soc. 2023, 89, 104374. [Google Scholar] [CrossRef]
  39. Fleischmann, M.; Feliciotti, A.; Romice, O.; Porta, S. Methodological foundation of a numerical taxonomy of urban form. Environ. Plan. B Urban Anal. City Sci. 2022, 49, 1283–1299. [Google Scholar] [CrossRef]
  40. Che, Y.; Li, X.; Liu, X.; Wang, Y.; Liao, W.; Zheng, X.; Zhang, X.; Xu, X.; Shi, Q.; Zhu, J.; et al. 3D-GloBFP: The first global three-dimensional building footprint dataset. Earth Syst. Sci. Data 2024, 16, 5357–5374. [Google Scholar] [CrossRef]
  41. Chen, Y.; Shan, B.; Yu, X. Study on the spatial heterogeneity of urban heat islands and influencing factors. Build. Environ. 2022, 208, 108604. [Google Scholar] [CrossRef]
  42. Xu, H.; Yu, B.; Chen, G.; Liu, Y.; Pu, Y. Accuracy Evaluation and Extraction of Urban Block Based on OpenStreetMap Data. Geospat. Inf. 2019, 17, 71–74+10. (In Chinese) [Google Scholar] [CrossRef]
  43. Cetin, M.; Ozenen Kavlak, M.; Senyel Kurkcuoglu, M.A.; Bilge Ozturk, G.; Cabuk, S.N.; Cabuk, A. Determination of land surface temperature and urban heat island effects with remote sensing capabilities: The case of Kayseri, Türkiye. Nat. Hazards 2024, 120, 5509–5536. [Google Scholar] [CrossRef]
  44. Della Rocca, F.; Pasquariello, P.; Masiello, G.; Serio, C.; De Feis, I. Machine Learning to Retrieve Gap-Free Land Surface Temperature from Infrared Atmospheric Sounding Interferometer Observations. Remote Sens. 2025, 17, 694. [Google Scholar] [CrossRef]
  45. Ouyang, Q.; Zheng, B.; Liu, J.; Luo, X.; Wu, S.; Sun, Z. Construction of a Cold Island Spatial Pattern from the Perspective of Landscape Connectivity to Alleviate the Urban Heat Island Effect. ISPRS Int. J. Geo-Inf. 2025, 14, 209. [Google Scholar] [CrossRef]
  46. Wei, X.; Ma, W.; Kong, L. Relationship between change of land surface temperature and land surface characteristics in medium and small cities. Remote Sens. Inf. 2019, 34, 115–119. [Google Scholar] [CrossRef]
  47. Anselin, L. Local indicators of spatial association—LISA. Geogr. Anal. 1995, 27, 93–115. [Google Scholar] [CrossRef]
  48. Ord, J.K.; Getis, A. Local spatial autocorrelation statistics: Distributional issues and an application. Geogr. Anal. 1995, 27, 286–306. [Google Scholar] [CrossRef]
  49. Zhou, W.; Tian, Y. Effects of urban three-dimensional morphology on thermal environment: A review. Acta Ecol. Sin. 2020, 40, 416–427. (In Chinese) [Google Scholar] [CrossRef]
  50. Yu, M.; Lu, X.; Xing, H.; Xu, Y.; Li, J.; Chen, X. Anaiysis of temporal and spatiai changes of surface thermal enviroment in Jinan city center. Sci. Surv. Mapp. 2021, 46, 100–107+157. (In Chinese) [Google Scholar] [CrossRef]
  51. Shen, Z.; Zhao, Y. Effect of spatial morphology on land surface temperature and their interaction: A case study of main urban area of Jinan City. J. Arid Land Resour. Environ. 2024, 38, 65–74. (In Chinese) [Google Scholar] [CrossRef]
  52. Shen, Z.; Wang, J.; Li, S. Effects of blue-green landscape patterns on land surfacetemperature in different geographic units of Jinan City. Geogr. Sci. 2025, 45, 1143–1156. (In Chinese) [Google Scholar] [CrossRef]
  53. Qiang, F.; Mengxue, X.; Bing, Z.; Lifang, W. Based on the Analysis of the Driving Factors of Thermal Environment in the Main Urban Area of Ji’nan City under the “Seasonal-Source-Sink” Situation. Ecol. Environ. Sci. 2026, 35, 75. (In Chinese) [Google Scholar] [CrossRef]
  54. Wang, J.; Meng, F.; Lu, H.; Lv, Y.; Jing, T. Individual and combined effects of 3D buildings and green spaces on the urban thermal environment: A case study in Jinan, China. Atmosphere 2023, 14, 908. [Google Scholar] [CrossRef]
Sustainability 18 02283 g001
Figure 1. The overview map of Jinan’s central urban area.
Figure 1. The overview map of Jinan’s central urban area.
Sustainability 18 02283 g001
Sustainability 18 02283 g002
Figure 2. The flowchart of the research.
Figure 2. The flowchart of the research.
Sustainability 18 02283 g002
Sustainability 18 02283 g003
Figure 3. Distribution map of LST in Jinan’s central urban area in summer.
Figure 3. Distribution map of LST in Jinan’s central urban area in summer.
Sustainability 18 02283 g003
Sustainability 18 02283 g004
Figure 4. Bivariate agglomeration analysis between the LST and H1 and H5. (a) LST and H1; (b) LST and H5.
Figure 4. Bivariate agglomeration analysis between the LST and H1 and H5. (a) LST and H1; (b) LST and H5.
Sustainability 18 02283 g004
Sustainability 18 02283 g005
Figure 5. Bivariate agglomeration analysis between the LST and H3 and H4. (a) LST and H3; (b) LST and H4.
Figure 5. Bivariate agglomeration analysis between the LST and H3 and H4. (a) LST and H3; (b) LST and H4.
Sustainability 18 02283 g005
Sustainability 18 02283 g006
Figure 6. Bivariate agglomeration analysis between the LST and H2 and H6. (a) LST and H2; (b) LST and H6.
Figure 6. Bivariate agglomeration analysis between the LST and H2 and H6. (a) LST and H2; (b) LST and H6.
Sustainability 18 02283 g006
Sustainability 18 02283 g007
Figure 7. Bivariate agglomeration analysis between the LST and P1 and P2. (a) LST and P1; (b) LST and P2.
Figure 7. Bivariate agglomeration analysis between the LST and P1 and P2. (a) LST and P1; (b) LST and P2.
Sustainability 18 02283 g007
Sustainability 18 02283 g008
Figure 8. Bivariate agglomeration analysis between the LST and P3.
Figure 8. Bivariate agglomeration analysis between the LST and P3.
Sustainability 18 02283 g008
Sustainability 18 02283 g009
Figure 9. Bivariate agglomeration analysis between the LST and P4 and S3. (a) LST and P4; (b) LST and S3.
Figure 9. Bivariate agglomeration analysis between the LST and P4 and S3. (a) LST and P4; (b) LST and S3.
Sustainability 18 02283 g009
Sustainability 18 02283 g010
Figure 10. Bivariate agglomeration analysis between the LST and S1 and S2. (a) LST and S1; (b) LST and S2.
Figure 10. Bivariate agglomeration analysis between the LST and S1 and S2. (a) LST and S1; (b) LST and S2.
Sustainability 18 02283 g010
Table 1. The system of urban spatial structure indexes and their meanings.
Table 1. The system of urban spatial structure indexes and their meanings.
DimensionalityIndicatorsMeaning
One-
dimensional height		Average height of DEM (H1)The average height of the terrain in the study area
Standard deviation of DEM (H2)Degree of topographic relief in the study area
Average building height(H3)The average height of buildings in the study area
Standard deviation of building height (H4)Fluctuation of building heights in the study area
Average absolute building height (H5)The average absolute building height in the study area
Standard deviation of absolute building height (H6)Fluctuation of absolute building heights in the study area
Two-
Dimensional plane		Building footprint mean (P1)The average building footprint area within the study area
Building footprint standard deviation (P2)The variability in building footprint area across the study area
Adjacent distance between buildings(P3)The average building proximity within the unit
Building density (P4)The numerical building density
Three-
dimensional space		Building volume mean (S1)The average volumetric density of the study area
Building volume standard deviation (S2)The dispersion of building volumes in the study area
Volume ratio (S3)Development intensity of the study area
Table 2. DEM elevation classification results.
Table 2. DEM elevation classification results.
Elevation CategoryRange of Elevations (Unit: m)
Level 118.5~62.5
Level 262.5~142.9
Level 3142.9~255.0
Level 4255.0~396.5
Level 5396.5~862.6
Table 3. Correlation analysis results of LST and urban spatial structure indexes (R).
Table 3. Correlation analysis results of LST and urban spatial structure indexes (R).
IndicatorsR
H1−0.7444 ***
H2−0.6889 ***
H30.4443 ***
H40.3398 ***
H5−0.7300 ***
H6−0.5241 ***
P10.3928 ***
P20.3241 ***
P3−0.0119
P40.5883 ***
S10.4330 ***
S20.3799 ***
S30.2914 ***
(Note: *** indicates a significance level of 0.001).
Table 4. Correlation analysis results of LST and urban spatial structure indexes (R) in different elevation ranges.
Table 4. Correlation analysis results of LST and urban spatial structure indexes (R) in different elevation ranges.
IndicatorsR (Level 1)R (Level 2)R (Level 3)R (Level 4)R (Level 5)
H2−0.0303 −0.1755 ***−0.1674 **−0.1990 ***−0.2039 ***
H30.2667 *** 0.0012−0.0418−0.0568−0.0577
H40.1556 *** −0.032−0.10060.01810.0009
H50.2275 *** −0.3375 ***−0.2201 ***−0.0709−0.4233 ***
H60.1457 ***−0.1586 **−0.1393 **−0.1896 ***−0.0684
P10.2081 ***0.1627 **0.08810.07950.0683
P20.1790 0.2089 ***0.12650.0590.0734
P3−0.0009 ***0.10070.09110.00110.0186
P40.5145 0.203 ***0.07820.1824 ***0.1417 **
S10.2570 *** 0.08250.05370.04990.038
S20.2324 *** 0.1710 ***0.10340.04480.0376
S30.1871 ***0.03450.0710.06790.0424
(Note: ** indicates a significance level of 0.01, and *** indicates a significance level of 0.001).
Table 5. Analysis of bivariate global spatial autocorrelation results between LST and urban spatial structure indexes.
Table 5. Analysis of bivariate global spatial autocorrelation results between LST and urban spatial structure indexes.
IndicatorsMoran’s I Z
H1−0.7129−61.5133
H2−0.6500−57.3546
H30.440444.9261
H40.347336.7669
H5−0.6995−60.7993
H6−0.5018−47.7239
P10.341534.4066
P20.281628.5795
P3−0.0284−2.9485
P40.508747.0945
S10.402240.8793
S20.342433.9729
S30.213822.6202
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Wang, J.; Zhang, X.; Li, Q.; Chen, Y. The Influence Mechanism and Spatial Heterogeneity of Urban Spatial Structure on the Thermal Environment: A Case Study of the Central Urban Area of Jinan. Sustainability 2026, 18, 2283. https://doi.org/10.3390/su18052283

AMA Style

Wang J, Zhang X, Li Q, Chen Y. The Influence Mechanism and Spatial Heterogeneity of Urban Spatial Structure on the Thermal Environment: A Case Study of the Central Urban Area of Jinan. Sustainability. 2026; 18(5):2283. https://doi.org/10.3390/su18052283

Chicago/Turabian Style

Wang, Junning, Xiaoqing Zhang, Qing Li, and Yuhan Chen. 2026. "The Influence Mechanism and Spatial Heterogeneity of Urban Spatial Structure on the Thermal Environment: A Case Study of the Central Urban Area of Jinan" Sustainability 18, no. 5: 2283. https://doi.org/10.3390/su18052283

APA Style

Wang, J., Zhang, X., Li, Q., & Chen, Y. (2026). The Influence Mechanism and Spatial Heterogeneity of Urban Spatial Structure on the Thermal Environment: A Case Study of the Central Urban Area of Jinan. Sustainability, 18(5), 2283. https://doi.org/10.3390/su18052283

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop