Seasonal and Diurnal Characteristics and Drivers of Urban Heat Island Based on Optimal Parameters-Based Geo-Detector Model in Xinjiang, China

Abstract

In the context of sustainable urban development, elucidating urban heat island (UHI) dynamics in arid regions is crucial. By thoroughly examining the characteristics of UHI variations and potential driving factors, cities can implement effective strategies to reduce their impacts on the environment and public health. However, the driving factors of a UHI in arid regions remain unclear. This study analyzed seasonal and diurnal variations in a surface UHI (SUHI) and the potential driving factors using Pearson’s correlation analysis and an Optimal Parameters-Based Geographic Detector (OPGD) model in 22 cities in Xinjiang, northwest China. The findings reveal that the average annual surface urban heat island intensity (SUHII) values in Xinjiang’s cities were 1.37 ± 0.86 °C, with the SUHII being most pronounced in summer (2.44 °C), followed by winter (2.15 °C), spring (0.47 °C), and autumn (0.40 °C). Moreover, the annual mean SUHII was stronger at nighttime (1.90 °C) compared to during the daytime (0.84 °C), with variations observed across seasons. The seasonal disparity of SUHII in Xinjiang was more significant during the daytime (3.91 °C) compared to nighttime (0.39 °C), with daytime and nighttime SUHIIs decreasing from summer to winter. The study also highlights that the city size, elevation, vegetation cover, urban form, and socio-economic factors (GDP and population density) emerged as key drivers, with the GDP exerting the strongest influence on SUHIIs in cities across Xinjiang. To mitigate the UHI effects, measures like urban environment enhancement by improving surface conditions, blue–green space development, landscape optimization, and economic strategy adjustments are recommended.

1. Introduction

The expansion of urban scale and rapid changes in urban form are key features of contemporary urban development. These transformations alter the physical properties of the original subsurface, impacting heat exchange between the surface and atmosphere, and consequently modifying the local thermal environment [1]. An urban heat island (UHI) refers to a significant temperature increase in urban areas compared to the surrounding countryside and natural surroundings [2]. This phenomenon primarily stems from land use changes due to urbanization, anthropogenic heat sources like buildings and roads, and a lack of vegetation cover. A UHI has significant implications for the climate, energy consumption, ecology, and human health, leading to increased energy usage, worsened air quality, and higher urban temperatures [3,4]. Therefore, it is crucial to gain a deeper understanding of the evolving characteristics of the UHI and its driving factors to support initiatives aimed at reducing environmental pollution, as well as improving urban planning, management, and overall human well-being.

UHIs can be classified into atmospheric urban heat islands (AUHIs) and surface urban heat islands (SUHIs) based on the data sources used in heat island studies [5]. An AUHI relies on atmospheric temperature data from fixed or locally mobile weather stations, resulting in discrete point or linear data with limitations in scale conversion accuracy and analyses of large cities [6,7]. In contrast, an SUHI utilizes remote sensing data, enabling continuous monitoring over large areas and significantly enhancing the study of urban thermal environments [8]. With advancements in airborne and satellite remote sensing technology, researchers now have access to a wide range of remote sensing data sources with improved spatial and temporal resolution, making them essential tools for investigating urban thermal environments [9]. The primary data sources for urban surface thermal environment research include NOAA/AVHRR, ASTER, Landsat TM/ETM+, and MODIS [10,11]. Among them, MODIS stands out for its high temporal resolution, allowing it to accurately capture the temporal differences of UHIs [12]. Therefore, MODIS has been widely used in SUHI studies [13,14,15]. Li et al. have leveraged the MOD11A2 product to examine the seasonal and diurnal variations in the SUHII in Chinese prefecture-level cities [16]. Hamdi et al. have analyzed the seasonal and diurnal fluctuations in the UHII within the El-Mansoura conurbation, Egypt, using MODIS data [17]. Khan and Shahid have evaluated the UHII in five major cities in Pakistan utilizing MODIS data on daily surface temperatures [18]. Si et al. have explored the diurnal cycle, monthly variations, and interannual trends of SUHIIs in 34 major urban agglomerations in China by employing MOD11A1 data from 2003 to 2019 [19].

The strength of the SUHI is commonly assessed through the surface urban heat island intensity (SUHII) [20]. Research on the spatial and temporal variations in the SUHII has a well-established background, with numerous studies highlighting significant diurnal, seasonal, and spatial changes. At the spatial level, the UHI varies based on the climatic conditions of different regions [21]. On a temporal scale, previous research has highlighted the significant seasonal and diurnal variations in the UHI. Research conducted on over 9500 cities worldwide has shown that during summer, SUHII is typically stronger than in winter, with daytime effects being more pronounced than nighttime effects, displaying specific spatial distribution characteristics [22]. In China, it has been observed that the SUHII is generally more pronounced at nighttime than during the daytime. Further examination of the SUHI in Beijing identified a stronger effect during nighttime compared to daytime across various local climate zones [23]. Additionally, Cao et al. have concluded that the annual average nighttime SUHII in China is notably stronger than during the daytime [24]. Seasonally, the nighttime SUHII is most intense in winter; however, variations exist in both diurnal and spatial aspects of this effect, depending on the research methodology employed [25,26]. For instance, research on the diurnal and seasonal SUHIIs in the Zhengzhou metropolitan area have demonstrated that the daytime SUHII peaks in summer, while the nighttime intensity is strongest in winter [27]. In contrast, a prior study by Asfa Siddiqui on the UHI in Lucknow, India, reported that both daytime and nighttime SUHIIs were most pronounced in summer [28]. Xi et al. did not specifically address diurnal SUHII variations, but they indicated that the average UHII in Hefei is strongest in summer and weakest in winter [29].

A comprehensive review of the relevant literature on the factors influencing the UHI reveals that it is shaped by a combination of natural factors like the geographic location and climate, as well as socio-economic factors such as urbanization and industrialization [30,31,32,33]. Previous research indicates that natural factors like the geographic location, regional climatic background, elevation, topography, and vegetation cover significantly impact the UHI. Among these, vegetation cover is considered to have a particularly strong impact [34]. Human-modified urban thermal environments are primarily influenced by surface cover factors such as land use/land cover and urban landscape patterns. In addition, impervious surfaces, green space areas, the urban spatial form, and density all affect the distribution of the UHI [35,36]. Moreover, the city size also plays a crucial role in the UHI [37,38], with suggestions that the city size and urban form may be key in its formation and progression [39,40]. The characteristics of urban scale, such as buildings, roads, and population density, can exacerbate the UHI [41,42]. Urban form, referring to the spatial arrangement of buildings, roads, and green spaces within a city, can have varying effects on the heat island effect based on different thermal characteristics of different morphologies [43]. For instance, Yin et al. conducted a spatial regression analysis in Wuhan, demonstrating that optimizing the urban form can significantly impact the management of the UHI [44]. In recent years, a number of studies have demonstrated that socio-economic factors significantly influence the SUHII. Wu et al. examined the drivers of the SUHII in Guangdong Province and found that socio-economic indicators, such as population and gross domestic product, are key determinants of the nighttime SUHII [45]. Similarly, Sidiqui et al. investigated the relationships among SUHII spatial variations, land use patterns, and socio-economic factors in the city of Greater Geelong, Australia, revealing that socio-economic and demographic factors play crucial roles in the generation of the SUHII [46].

Nowadays, researchers employ various methods and models to investigate the relationship between the SUHII and its driving factors. Common approaches include Pearson’s correlation analysis, Ordinary Least Squares (OLS), Geographical Weighted Regression (GWR) model, gray correlation analysis, and geo-detector analysis [44,47,48,49,50]. The geo-detector model, introduced by Wang and Xu [51], has been widely utilized to study SUHI driving factors. For instance, Hu et al. utilized the geo-detector model to assess the explanatory power of impact factors on the urban thermal environment in Tianjin [52]. Similarly, Zhao et al. identified key impact factors of the urban thermal environment in Zhengzhou City using the geo-detector model, highlighting factors such as the Normalized Difference Building Index (NDBI), Normalized Difference Vegetation Index (NDVI), and anthropogenic factors [53]. In addition, Xiang et al. examined the dominant factors of the seasonal SUHII in the urban agglomeration of the middle reaches of the Yangtze River through a combination of Spearman’s correlation analysis and geo-detector methods [54]. Their analysis revealed differences in dominant factors between daytime and nighttime.

Globally, semi-arid, arid, and hyper-arid areas encompass approximately 41 percent of the world’s land area. Despite the harsh conditions prevalent in these environments, they support over two billion people, accounting for about one-third of the global population. This underscores their significance for human survival and habitation. Over the past 40 years, these regions have witnessed rapid population growth and urbanization, resulting in significant transformations of urban landscapes and the emergence of the UHI. And research has indicated that arid and semi-arid areas exhibit a greater UHII compared to humid and semi-humid regions [21]. While previous research has delved into the seasonal diurnal variations and driving factors of the UHI, there remains a gap in understanding the seasonal and diurnal characteristics and drivers of this phenomenon in the arid and semi-arid regions of northwest China, particularly in Xinjiang. As a crucial economic hub and frontier region in the urbanization process of Central Asia and western China, Xinjiang is grappling with significant UHI impacts and resulting ecological challenges due to its temperate continental arid climate and delicate ecological surroundings. Thus, this study aims to (1) analyze the seasonal and diurnal variation patterns of the SUHII across 22 cities in Xinjiang, a region characterized by arid and semi-arid climates in northwest China; (2) to explore the relationship between the urban size and SUHII; (3) to evaluate the impacts of natural factors, urban form, and socio-economic circumstances on the SUHII using OPGD analysis. The study also proposes strategies and recommendations to mitigate the UHI in Xinjiang. Furthermore, the findings of this research can offer insights and lessons for understanding the UHI in other arid zone cities across the global urbanization landscape.

2. Materials and Methods

2.1. Study Area

Xinjiang is situated in northwestern China and spans from 73°25′ E to 97°30′ E and 34°30′ N to 49°00′ N, covering 1,664,900 km², which is one-sixth of China’s total land area [55]. It is the largest provincial-level administrative region in China [56,57]. The topography of Xinjiang includes mountain ranges, basins, deserts, and grasslands, which significantly impact the region’s climate and environment [58]. Xinjiang exhibits various climate types, predominantly arid and semi-arid, such as temperate continental arid and alpine climates [59]. Xinjiang boasts abundant light and heat resources, with a total solar radiation ranging from 5440 to 6280 MJ/m² and an average annual sunshine hours of 2500–3400h, making it one of the regions in China with the highest sunshine hours [60]. The region maintains an average temperature of about 6–10 °C over many years, but experiences significant seasonal temperature variations. Summers are characterized by high temperatures exceeding 30 °C on average, while winters are extremely cold, with temperatures dropping below −10 °C. Xinjiang experiences limited annual precipitation, averaging between 100 and 200 mm, leading to widespread drought conditions across most of the region [61]. For this study, 22 cities in Xinjiang with urban built-up areas greater than 10 km² were selected, as depicted in Figure 1.

2.2. Data Sources and Preprocessing

The primary data utilized in this study include the MODIS land surface temperature (LST) dataset; land use/land cover (LULC) data; natural geographic feature data, including precipitation, topography, and NDVI; as well as socio-economic data such as the population density, GDP, and the urban green space rate (

Table 1. Datasets used in this study.

Data Type	Factors	Year	Spatial Resolution	Source
Land surface temperature (LST)		2020	1 km	National Tibetan Plateau/Third Pole Environment Data Center (https://data.tpdc.ac.cn/ (accessed on 24 January 2023))
Land use/land cover (LULC)		2020	30 m	Resources and Environmental Science Data Center (https://www.resdc.cn/ (accessed on 24 January 2023))
Natural geographic data	Precipitation	2020	1 km	National Tibetan Plateau/Third Pole Environment Data Center (https://data.tpdc.ac.cn/ (accessed on 16 May 2023))
	DEM	2020	30 m	NASA (https://search.earthdata.nasa.gov/ (accessed on 16 May 2023))
	NDVI	2020	1 km
Socio-economic data	Population density	2020	-	China Urban Construction Statistical Yearbook 2020
	GDP	2020	-	China City Statistical Yearbook 2021
	Urban green space rate	2020	-	China Urban Construction Statistical Yearbook 2020

). LST data were utilized to quantify the temperature in the study area, enabling the calculation of the SUHII. The LST dataset for daytime and nighttime on 15 January, 15 April, 15 July, and 15 October 2020 was chosen to represent the four seasons. In this study, the MODIS product converts land surface temperature units from thermodynamic Kelvin (K) to Celsius (°C) using a field calculator in ArcGIS. The LULC data for the study area in Xinjiang in 2020 were obtained by cropping vector maps of city boundaries using the Extract-by-Mask tool in ArcGIS. The urban and rural areas were then extracted based on the land use coverage classification system of the Institute of Geographic Sciences and Resources of the Chinese Academy of Sciences (IGSR). Urban built-up areas were defined as continuous impervious surfaces greater than 10 km². Landscape pattern indices were calculated to describe urban form. Natural geographic data provided environmental context, while socio-economic data reflected urbanization levels and human activities.

2.3. Methods

2.3.1. Surface Urban Heat Island Intensity

The surface urban heat island intensity (SUHII) is a measure of the UHI within a particular area or city [20]. It is commonly assessed by comparing the temperature in the urban area to that in the surrounding rural or natural areas [62,63]. Greater temperature variances signify a stronger UHI. Its formula is as follows:

$$SUHII=LS{T}_U-LS{T}_R$$

(1)

where LST_U is the LST of core urban areas, while LST_R is the LST of surrounding rural or natural areas.

2.3.2. Urban Form

In order to characterize the spatial pattern of urban form, the study selected the urban landscape pattern index and urban fractal dimension. The landscape pattern index, originating from landscape ecology, is a suitable method for quantifying the spatial pattern characteristics of urban form [64]. It effectively characterizes the spatial composition and structural features around elements and has been widely used in quantitative research on urban form and land use [65]. In some studies, researchers have integrated the landscape pattern index with a thermal environment analysis [66]. Seven types of landscape pattern indices were chosen based on previous studies to quantify the spatial pattern of urban form in this research, and the calculations were performed using Fragstats 4.2 software. The urban fractal dimension was calculated using the box counting method [67], a simple technique for estimating fractal dimension, implemented through Fractalyse 3.0 software. The specific urban form factors and their descriptions are detailed in

Table 2. Detailed description of urban form factors.

Category	Variable	Description	Equation
Urban forms	Largest Patch Index (LPI)	Percent area of the largest patch of urban land in the total landscape area.	$\begin{array}{c}LPI={\displaystyle \frac{\left[\mathit{max}(a_1,a_2,\dots ,a_j\right)]}{A}}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}(2)\end{array}$
	Landscape Shape Index (LSI)	LSI equals 1 when the landscape has the most regular shape; the value increases with the complexity of the landscape shape.	$\begin{array}{c}LSI={\displaystyle \frac{0.25E}{\sqrt{A}}}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}(3)\end{array}$
	Mean perimeter–area ratio (PARA_MN)	PARA_MN indicates the average perimeter-to-area ratio of urban patches, representing the regularity of urban patch shapes in the landscape.	$\begin{array}{c}PARA\_MN={\displaystyle \frac{P}{A}}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}(4)\end{array}$
	Percentage of Like Adjacencies (PLADJ)	PLADJ indicates the percentage of cell adjacencies involving urban patches that are similar.	$\begin{array}{c}PLADJ={\displaystyle \frac{N_{\mathit{Simlar}}}{N_{Total}}}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em} (5)\end{array}$
	Patch Cohesion Index (COHESION)	COHESION measures the physical connectedness of urban land; it increases as the urban patch becomes more clumped or aggregated in its distribution.	$\begin{array}{c}COHESION={\displaystyle \frac{\left[1-\frac{{{\displaystyle \sum }}_{j=1}^m{p}_{ij}}{{{\displaystyle \sum }}_{j=1}^m{p}_{ij}\sqrt{a_{ij}}}\right]}{\left[1-\frac{1}{\sqrt{A}}\right]}}(6)\end{array}$
	Aggregation Index (AI)	AI is used to determine the degree of compactness of the landscape.	$\begin{array}{c}AI=\left[{\displaystyle \frac{a_{ii}}{\mathit{max}\to a_{ii}}}\right]\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}(7)\end{array}$
	Urban fractal Dimension (DI)	DI is considered as a measure of compact-ness, i.e., compact cities have usually large values of DI.	$\begin{array}{c}D={\displaystyle \frac{\mathit{log}N}{\mathit{log}r}}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}(8)\end{array}$

.

2.3.3. Pearson’s Correlation Analysis

The Pearson correlation coefficient is frequently utilized to depict the linear correlation between variables x and y [68]. The Pearson correlation coefficient quantifies the strength of the correlation relationship, and its formula is as follows:

$$r={\displaystyle \frac{{{\displaystyle \sum }}_{i=1}^n\left(x_i-\overline{x}\right)\left(y_i-\overline{y}\right)}{\sqrt{{{\displaystyle \sum }}_{i=1}^n{\left(x_i-\overline{x}\right)}^2}\sqrt{{{\displaystyle \sum }}_{i=1}^n{\left(y_i-\overline{y}\right)}^2}}}$$

(9)

where r is the Pearson correlation coefficient, which is calculated by determining the covariance and standard deviation of two variables. The value of r falls within the range of [−1, 1]. A correlation of 0 indicates no linear relationship between x and y. The closer the r is to 1 or −1, the stronger the correlation. Conversely, the closer the r is to 0, the weaker the correlation.

2.3.4. Optimal Parameters-Based Geographic Detector (OPGD)

Geo-detector models can quantitatively measure the significance of independent variables in relation to dependent variables based on the theory of spatial differentiation [51]. Traditional studies of drivers of surface temperature and thermal environment differentiation typically rely on a correlation analysis using SPSS software (https://www.ibm.com/spss accessed on 24 January 2023) to test significance. However, this method may not be convenient for exploring the strength of factors’ influences. The introduction of the OPGD offers a new statistical measurement model that overcomes limitations of traditional geo-detectors, such as the lack of collinearity and no variable assumptions. OPGD also addresses issues with data discretization in traditional geo-detector models, eliminating the need for manual adjustments and reducing the impacts of human subjective factors [69]. OPGD has found widespread application in natural science, social science, and environmental pollution studies [70]. The factor detector is utilized to analyze driving factors of spatial divergence in the UHI in this study. The q value of each driver, calculated with different classification methods and numbers, represents the explanatory power of the driver on the SUHII. Its formula is as follows:

$$q=1-{\displaystyle \frac{{{\displaystyle \sum }}_{g=1}^L{N}_g{\sigma }_g^2}{N{\sigma }^2}}$$

(10)

where L represents the stratification of the dependent or independent variable; N_g and σ_g² denote the number of samples and the variance of stratification group g; N and σ² refer to the number of samples and the variance of the entire model, respectively.

2.4. Study Steps

In this study, we first calculated the SUHII for 22 cities in Xinjiang using MODIS LST data (ArcGIS 10.8). Next, we calculated and organized 14 natural factors, urban size and form factors, as well as socio-economic factors that may potentially influence the SUHII. We then employed Pearson’s correlation analysis to assess the relationship between the SUHII and its influencing factors across different seasons and during both daytime and nighttime. Finally, we utilized the OPGD model to investigate the primary factors affecting the SUHII in Xinjiang across various seasons and times of day.

3. Results

3.1. Seasonal and Diurnal Characteristics of the SUHI

Figure 2 illustrates the seasonal and diurnal SUHII distributions of 22 cities in Xinjiang in 2020, highlighting the high spatial heterogeneity of the SUHII in the region. The mean SUHII of these cities in 2020 was 1.37 °C (

Table 3. Descriptive statistics of the SUHII.

	Annual Average SUHII (°C)	Annual Daytime Average SUHII (°C)	Annual Nighttime Average SUHII (°C)
mean	1.37	0.84	1.90
maximum	2.86	3.30	3.90
minimum	−0.78	−2.11	0.30

), ranging from −3.95 °C (spring daytime in Wujiaqu) to 10.90 °C (winter daytime in Yining). Spatially, the annual mean SUHII distribution pattern in Xinjiang cities follows northern Xinjiang (1.59 °C) > southern Xinjiang (1.36 °C) > eastern Xinjiang (−0.10 °C). Specifically, Urumqi had the highest annual average SUHII value (2.86 °C), while Turpan had the lowest (−0.78 °C). Turpan and Wujiaqu were the only cities with annual average SUHII values below 0 °C, indicating the absence of a heat island throughout the year. Yining had the highest annual average daytime SUHII value (3.29 °C), while Wujiaqu had the lowest (−2.11 °C). Seven cities, including Altay, Hami, Kashgar, Kuitun, Tumushuke, Turpan, and Wujiaqu, had annual average daytime SUHII values below 0 °C, suggesting no daytime heat island throughout the year. Urumqi had the highest annual average nighttime SUHII value (3.89 °C), while Turpan had the lowest (0.30 °C). All 22 cities in Xinjiang experienced a nighttime heat island throughout the year, with annual average nighttime SUHII values above 0 °C.

Figure 3 illustrates the seasonal and diurnal variations in the SUHII in Xinjiang. In 2020, the SUHII values for spring, summer, autumn, and winter were 0.47 °C, 2.44 °C, 0.40 °C, and 2.15 °C, respectively, indicating a seasonal trend of summer > winter > spring > autumn. Analyzing the diurnal patterns, the annual average SUHIIs were higher at nighttime (1.90 °C) compared to during the daytime (0.84 °C) across the year. When looking at individual seasons, daytime SUHIIs were higher than nighttime SUHIIs in summer (2.81 °C, 2.08 °C) and winter (2.61 °C, 1.69 °C), while the reverse was observed in spring (−0.97 °C; 1.91 °C) and autumn (−1.10 °C; 1.91 °C). Notably, during spring and autumn daytime, some cities in Xinjiang experienced SUHIIs below 0 °C, suggesting the lack of a heat island in these periods. Overall, the SUHII in Xinjiang follows a pattern of summer daytime > winter daytime > summer nighttime > spring nighttime > autumn nighttime > winter nighttime > spring daytime > autumn daytime, with the heat island being more pronounced in summer than in winter, regardless of daytime or nighttime.

3.2. Relationship Between Urban Size and the SUHII

Figure 4 and Figure 5 illustrate the relationship between urban size and the SUHII. Urban size refers to the size of the urban built-up area, typically measured in km². In the study area, the urban built-up area of the 22 cities ranges from 13.71 km² (Aral) to 557.09 km² (Urumqi), with an average of 76.59 km². The cities are categorized into four groups based on size: 0–25 km², 25–50 km², 50–75 km², and over 75 km². Analyzing the SUHIIs of these cities reveals that larger cities exhibit a stronger SUHII during the daytime, while at nighttime, all cities except those in the 50–75 km² category show a similar trend. The annual mean SUHIIs during the daytime for cities of 0–25 km², 25–50 km², 50–75 km², and over 75 km² are −0.03 °C, 1.17 °C, 1.27 °C, and 1.29 °C, respectively. During the nighttime, the annual mean SUHIIs for these categories are 1.62 °C, 1.91 °C, 1.38 °C, and 2.56 °C. This indicates the presence of a UHI during the daytime in cities larger than 25 km² and during the nighttime in cities of all sizes.

3.3. Impact Factors of the SUHII

3.3.1. Descriptive Statistical Analysis of SUHII Impact Factors

In this study, 14 factors, including the LPI (Largest Patch Index), LSI (Landscape Shape Index), PARA_MN (mean perimeter-area ratio), PLADJ (Percentage of Like Adjacencies), COHESION (Patch Cohesion Index), AI (Aggregation Index), DI (Urban fractal Dimension), PRE (precipitation), AL (altitude), EL (elevation), NDVI (Normalized Difference Vegetation Index), PD (population density), GDP (gross domestic product), and UGSR (urban green space rate) were selected to examine their impacts on the SUHII in Xinjiang cities.

Table 4. Descriptive statistics of the driving factors.

Category	Variable	N	Minimum	Maximum	Mean	STD
Urban Forms	LPI	22	0.03	9.49	1.38	2.25
	LSI	22	2.38	8.55	4.00	1.57
	PARA_MN	22	14.48	841.12	153.34	240.19
	PLADJ	22	97.39	98.97	98.33	0.39
	COHESION	22	98.93	99.87	99.50	0.21
Natural Factors	AI	22	97.75	99.25	98.79	0.34
	DI	22	1.20	1.77	1.55	0.13
	PRE	22	25.72	301.00	144.11	79.79
	AL	22	421.98	1376.21	750.13	290.45
	EL	22	−129.98	402.41	27.07	109.42
	NDVI	22	0.07	0.29	0.20	0.06
Human Factors	PD	22	643.00	6355.00	3796.95	1247.49
	GDP	22	519,300.00	33,373,200.00	4,170,059.91	6,646,827.05
	UGSR	22	30.72	44.95	36.57	3.28

provides the descriptive statistics for these variables, including their minimum, maximum, mean, and standard deviation (STD) values. The results indicate that the LPI, PARA_MN, EL, and GDP exhibit larger STD values compared to their mean values, suggesting significant variation among the 22 cities in Xinjiang. Conversely, the urban form factors LSI, PLADJ, COHESION, AI, and DI show smaller STD values relative to the means, suggesting less variation in urban form among the cities. Particularly, DI shows a range between 1 and 2, with a maximum value of 1.77 and a minimum value of 1.20, suggesting fractal self-similarity in urban form and aligning with principles of fractal geometry [71]. Moreover, natural and socio-economic factors such as PRE, AL, NDVI, PD, and UGSR also exhibit STD values smaller than the means, implying relatively consistent values across the 22 cities in Xinjiang. A comparison with national averages reveals that the mean values of PRE, NDVI, and UGSR in Xinjiang cities are lower than the national average (706 mm, 0.38, and 38.24%, respectively), indicating lower precipitation levels, vegetation cover, and urban green space in these cities. Additionally, the average PD of the 22 cities in Xinjiang is higher than the national average of 2778 persons per square kilometer, indicating a higher concentration of population in urban areas. This trend may be linked to the presence of job opportunities and improved social services like transportation, healthcare, education, and cultural amenities in urban centers as opposed to rural villages in Xinjiang. Consequently, there is a steady migration of non-urban residents towards the cities [72].

3.3.2. Pearson’s Correlation Analysis of the SUHII Impact Factors

The Pearson correlation analysis was conducted to examine the relationship between the SUHII and impact factors across different seasons in 2020 within each city of the study area. The results, depicted in Figure 6, indicate seasonal and diurnal variations in the r value and significance level of the Pearson correlation coefficients between the SUHII and the impact factors in Xinjiang during 2020. The significance test at the 0.1 level revealed specific correlations. The spring daytime SUHII was significantly and positively correlated with AL (r = 0.802), while the summer daytime SUHII was negatively correlated with EL (r = −0.368). The autumn daytime SUHII was positively correlated with AL (r = 0.558), while the winter daytime SUHII was positively correlated with PRE (r = 0.758) and COHESION (r = 0.416), and negatively correlated with AL (r = −0.411) and EL (r = −0.469). Moreover, the spring nighttime SUHII exhibited a significant positive correlation with GDP (r = 0.368), while the summer nighttime SUHII was positively correlated with PLADJ (r = 0.455), GDP (r = 0.451), and COHESION (r = 0.376). The autumn nighttime SUHII was positively correlated with the GDP (r = 0.411), and the winter nighttime SUHII was positively correlated with the GDP (r = 0.483), PARA_MN (r = 0.405), and PLADJ (r = 0.382), while being negatively correlated with the UGSR (r = −0.368).

Overall, the daytime SUHII of Xinjiang cities exhibited a notable correlation with topographic factors, specifically AL and EL. The spring and autumn SUHIIs showed positive correlations with AL, while the summer and winter SUHIIs displayed negative correlations with both AL and EL. Furthermore, during winter, the daytime SUHII also demonstrated significant positive correlations with PRE and COHESION. Notably, there was a significant positive correlation between the nighttime SUHII and GDP across all cities in Xinjiang. Additionally, the summer nighttime SUHII exhibited significant positive correlations with PLADJ and COHESION, while the winter nighttime SUHII displayed positive correlations with PARA_MN and PLADJ, along with a negative correlation with the UGSR.

3.3.3. The Results of the OPGD Analysis

The OPGD was utilized to identify the impact factors influencing the SUHII across different seasons in each city of Xinjiang in 2020. Figure 7 illustrates that the selected key factors have varying degrees of influence on SUHII across different seasons in each city within the study area. The main drivers of the daytime SUHII in spring include AL, LSI, EL, and GDP, while EL, NDVI, LPI, GDP, PLADJ, and DI play key roles in summer. In autumn, significant factors are EL, GDP, AL, and PRE, and during winter, major drivers are PRE, COHESION, NDVI, AL, and PLADJ. For the spring nighttime SUHII, the influential factors are EL, PD, NDVI, GDP, PLADJ, AI, and LPI, while crucial factors for the summer nighttime SUHII are the GDP, PLADJ, NDVI, PD, AI, and PARA_MN. Important factors for the autumn nighttime SUHII include PD, GDP, PLADJ, COHESION, LPI, PRE, DI, and AI, and for the winter nighttime SUHII, significant factors are PRE, GDP, PLADJ, and AI. Notably, all these factors demonstrate an explanatory power exceeding 50%. This analysis reveals that the SUHII in each city is impacted by a combination of natural environmental, urban size and form, and socio-economic factors.

Among all driving factors, GDP has the greatest impact on SUHII, with an explanatory power of more than 50% for all study periods, except winter daytime. The other drivers exhibit significant seasonal and diurnal variations. For instance, LSI affects the SUHII only during spring daytime, PARA_MN impacts the SUHII solely during summer nighttime, AL influences the SUHII exclusively during daytime, while PD and AI affect the SUHII solely during nighttime. The study highlights the various urban form factors, natural factors, and socio-economic factors that influence the SUHII during both daytime and nighttime. Urban form factors such as LSI, LPI, PLADJ, DI, and COHESION play roles in the SUHII during the day, while natural factors like AL, EL, NDVI, and PRE impact the SUHII across different seasons. Socio-economic factors like the GDP are also found to affect the SUHII during the day. At night, factors such as LPI, PLADJ, DI, COHESION, AI, and PARA_MN influence the SUHII, while natural factors like EL, NDVI, and PRE play a role in the SUHII during specific seasons. Socio-economic factors like the GDP and PD are also identified as impacting the SUHII at night. Overall, the study highlights the seasonal variability in the drivers of SUHII.

4. Discussion

4.1. Seasonal and Diurnal Characteristics of the SUHI in an Arid Region

This study delves into the spatial distribution pattern of the SUHII in 22 cities in Xinjiang, a region located in the arid zone of Central Asia. It focuses on exploring the seasonal and diurnal differences in the SUHII in cities within the arid zone, aiming to uncover the drivers behind the SUHII in these urban areas. The results showed that the spatial distribution pattern of the annual average SUHII in Xinjiang cities follows a trend of North Xinjiang > South Xinjiang > East Xinjiang, possibly linked to regional economic disparities. Urumqi stands out with the highest SUHII at 2.86 °C, which is attributed to its status as the largest city in Xinjiang with substantial urban development, industrial activity, and construction intensity compared to other cities [73]. Overall, cities in Xinjiang exhibit significant heat islands, with summer and winter generally showing stronger effects compared to spring and autumn. The SUHII is more pronounced in winter and summer than in spring and autumn owing to seasonal climatic variations, such as low temperatures and atmospheric stability in winter, as well as high solar radiation and reduced adaptability of the evaporative system in summer. Additionally, anthropogenic influences, including heat emissions from heating and cooling activities, urban subsurface characteristics that promote high heat storage and the reduction of green spaces, and atmospheric pollution factors, such as the greenhouse effect and atmospheric stability, contribute to this phenomenon. Diurnal variations are more pronounced in spring and autumn, while less evident in summer and winter. Notably, the annual average SUHII is higher at nighttime than during the daytime, aligning with previous research indicating that the nighttime SUHII surpasses the daytime SUHII in arid regions [74]. This phenomenon may be attributed to materials like asphalt, bricks, and concrete absorbing and retaining solar radiation during the daytime, releasing it gradually at nighttime [75]. On the other hand, this difference may be attributed to the fact that the annual nighttime mean SUHII in Xinjiang is consistently above 0 °C, while the annual daytime mean SUHII is below 0 °C in seven cities. This results in a stronger annual nighttime mean SUHII than daytime SUHII in Xinjiang. The lower daytime mean SUHII in these cities could be due to their desert surroundings, where the built-up areas experience lower temperatures during the daytime compared to the surrounding deserts, possibly due to the cooling effect of vegetation [76,77]. The research also revealed that the summer and winter SUHIIs are stronger during the daytime, aligning with previous findings [78]. However, the study noted that spring and autumn SUHIIs are stronger at nighttime, possibly influenced by the seasonal variations in Xinjiang. The findings also suggest that the seasonal difference in the SUHII is greater during the daytime, indicating that the SUHII is more stable at nighttime and aligning with the findings of Wu et al. [74], Peng et al. [78] and Wang et al. [79]. In addition, our study observed a decreasing trend in the SUHII from summer to winter during the daytime, consistent with the findings of Zhou et al. [80]. However, unlike the study by Zhou et al. [80], our research also identified a decreasing trend in the SUHII at nighttime, possibly due to the unique diurnal temperature difference in arid regions. The decrease in the SUHII from summer to winter was attributed to surface cover and urban activities in summer, which led to higher temperatures compared to suburban areas. In contrast, both the city and suburbs were covered in snow and ice during winter, resulting in minimal temperature variations.

4.2. Driving Factors of the SUHII

4.2.1. Impacts of Natural Factors on the SUHII

The role of the natural environment as a key determinant of the UHI phenomenon is widely acknowledged. Our research identified EL, NDVI, PRE, and AL as significant natural factors influencing the SUHII in cities across Xinjiang. Among these factors, EL was found to have a more pronounced effect on the SUHII compared to AL, as confirmed by previous studies [74]. The OPGD analysis revealed that EL had a greater impact on the SUHII during the daytime in the hot season and at nighttime in the cold season. Pearson’s correlation analysis further demonstrated that EL exerts a significant inhibitory effect on the SUHII during daytime in both summer and winter. This suggests that cities situated at a lower EL exhibit a stronger SUHII compared to rural areas, consistent with previous research [81]. However, in other seasons, the relationship between EL and the SUHII was either insignificant or demonstrated a slight positive correlation, contrary to previous findings. This discrepancy was also observed in studies conducted in Jaipur City, India, and 201 prefecture-level cities in China. Researchers attributed this phenomenon to the complex impact of urban expansion on factors such as solar radiation, surface roughness, and vegetation cover, resulting in uncertainties regarding its influence on the SUHII [16,82]. Additionally, the unique characteristics of different cities were identified as factors contributing to this variability. The results from the OPGD indicate that the impact of the NDVI on the SUHII varies based on daytime and nighttime, with a greater effect observed during the daytime in summer and winter, and during nighttime in spring and autumn. The Pearson correlation analysis revealed that the relationship between the NDVI and SUHII in 22 cities in Xinjiang was not statistically significant. This contrasts with previous studies that suggest a significant negative correlation between the NDVI and SUHII, highlighting the unique climate conditions in the study area that influence this relationship across different seasons. Previous research by Mathew, Khandelwal, and Kaul demonstrated that the influence of the NDVI on the SUHII is more pronounced in winter and monsoon seasons compared to summer [82], supporting the notion that the association between the NDVI and SUHII is season-dependent. Additionally, Fujibe observed a lack of correlation between the SUHII and changes in urban surface cover, which aligns with Fumiaki Fujibe’s suggestion that urban warming may be more strongly linked to internal changes such as increased commercial activities and building heights rather than the spatial coverage of urban surfaces [83].

4.2.2. Impacts of Urban Size and Urban Form on the SUHII

Numerous studies have indicated that urban size and urban form are crucial factors in the formation and development of UHIs. In relation to urban size, our analysis of 22 cities in Xinjiang revealed a consistent trend wherein larger cities exhibited a stronger SUHII, which aligns with previous research [40,84,85,86]. They also found that the UHI is more pronounced in large cities compared to smaller ones, primarily due to the high density of buildings and roads in urban areas. These man-made structures absorb heat rapidly and have a low heat capacity, which results in swift urban warming. Furthermore, the dense population and frequent human activities in large cities contribute to significant heat emissions, further intensifying the UHI. However, a contrasting finding has been reported by Su et al., who observed a higher SUHII in medium and large cities compared to mega-cities in Chinese urban settings [37]. Su attributed this disparity to the advanced functional area planning, well-structured blue–green spaces, and efficient traffic management prevalent in mega-cities, which mitigate the influence of urban size on the SUHII. Consequently, medium and large cities exhibit higher SUHII levels than mega-cities. In our investigation, as the built-up area of all 22 Xinjiang cities is less than 600 km², our results align with previous studies focusing on small to medium-sized cities. In terms of urban form, our study identified PLADJ, AI, LPI, COHESION, DIMENSION, LSI, and PARA_MN as significant factors influencing the SUHII in Xinjiang cities, with PLADJ and AI emerging as the factors with the most impact. The OPGD revealed that the influence of PLADJ on the SUHII was more pronounced at nighttime compared to daytime across all seasons. Additionally, Pearson’s correlation analysis showed that PLADJ positively impacted the SUHII during daytime in summer and winter, as well as at nighttime in all seasons. Previous studies by Chen et al. [87] and Chunling et al. [88] in Beijing and Wuhan also reported similar findings, with correlation coefficients of 0.305 and 0.7555, respectively. The results of the AI and PLADJ analyses showed similarities, with AI having a greater impact on the SUHII at nighttime than during the daytime in all seasons, positively promoting the SUHII during nighttime. This aligns with the conclusions of Wang et al. [89] and Shen et al. [90] that higher AI values, indicating more agglomerated built-up areas, lead to a more severe SUHII. The effects of LPI, COHESION, DIMENSION, LSI, and PARA_MN on the SUHII varied seasonally, consistent with findings from Chen et al. [87]. It can be inferred that the combined effects of multiple factors contribute to the understanding of the SUHII across different seasons.

4.2.3. Impacts of Socio-Economic Factors on the SUHII

With the rapid urbanization and industrialization, the impacts of socio-economic factors on the SUHII are increasingly significant. Our research identified the GDP and PD as key socio-economic factors influencing the SUHII in cities in Xinjiang. Through Pearson’s correlation analysis and OPGD results, we determined that the GDP plays a crucial role in enhancing the SUHII across all seasons, except winter daytime, in 2020. Specifically, the GDP shows a positive correlation with the SUHII, particularly at nighttime, with a statistically significant relationship. This finding aligns with previous studies [91], where Chen also observed a minor impact of the GDP on surface temperature during winter daytime compared to other seasons. Other studies [23,92,93,94,95,96,97,98] have reported a strong positive correlation between the GDP and SUHII (0.39 ≤ r ≤ 0.92) at various scales, highlighting the impact of urban human activities and socio-economic development on UHIs. Additionally, the OPGD results indicated that PD has a greater impact on the SUHII at nighttime across all seasons, suggesting that PD primarily affects the SUHII during nighttime. The Pearson correlation analysis revealed that PD promotes the SUHII at nighttime in summer and suppresses it in spring and autumn. These results are consistent with previous research [16], where Li found a positive correlation (r = 0.09) between PD and the SUHII during summer nighttime in a study of 201 prefecture-level cities in east–central China, with the opposite trend observed at nighttime during other seasons.

4.3. Policy Suggestions

The SUHII in Xinjiang exhibits clear spatial variations, as well as distinct seasonal and diurnal characteristics. When developing policies to address the SUHII, it is crucial to prioritize the identification of regions and seasons with pronounced SUHIIs, strategically harnessing positive factors while minimizing the impacts of negative factors and devising tailored strategies for different regions [99]. Additionally, the complexity of reducing the SUHII necessitates adaptable policies for individual cities, with a comprehensive approach that considers the entire urban system rather than solely focusing on the UHI. Based on our findings, we propose the following insights for mitigating the SUHII in Xinjiang cities. First of all, it is crucial to manage the expansion of impervious surfaces and artificial buildings, adjust the urban subsurface structure, and enhance the development of urban blue and green spaces [100,101,102]. Utilizing environmentally friendly building materials like fiber-reinforced plastic and bamboo–glass fiber composites instead of traditional materials such as steel, glass, and concrete, along with implementing green roofs and reflective surfaces, can effectively control the SUHII [103,104]. Green roofs help lower surface temperatures, while reflective surfaces increase surface albedo, resulting in a more efficient cooling effect [105,106]. Secondly, it is essential to manage urban size and optimize the spatial layout of the city’s landscape. In order to combat the SUHII while also controlling the expansion of large built-up areas, a balance must be struck between SUHII mitigation and economic growth through the development of satellite cities and new towns. Finally, it is advisable to diversify surface cover types and decrease the proximity of similar surfaces by incorporating parks, urban ventilation corridors, and other measures to break the spatial continuity of the city [107,108,109]. Additionally, adjustments to the industrial structure and optimization of population distribution are essential. Large industrial enterprises can boost GDP growth by refining their industrial layout, enhancing production efficiency, and minimizing the environmental impacts of industrial activities in urban areas. Long-term strategies to combat the SUHII include reducing anthropogenic heat emissions by promoting energy conservation, enhancing energy efficiency, and encouraging the use of public transportation. It is noteworthy that Xinjiang serves as a significant renewable energy source for power generation. With its abundant wind and solar energy resources, the gradual replacement of fossil fuels with renewable energy sources may represent an effective approach for the future [104,110].

4.4. Limitations and Prospects

This study still has some limitations and uncertainties. In this study, we analyzed the seasonal and diurnal variations in the SUHII in Xinjiang, utilizing MODIS data as the primary source. MODIS data offer significant advantages for studying the SUHII due to their extensive global coverage and high temporal resolution, which facilitate continuous and dynamic monitoring of surface temperatures. This capability provides essential insights into the spatial and temporal distribution of the SUHII and its trends. However, MODIS data also have limitations; specifically, their spatial resolution is inadequate for analyzing fine urban structures, which may result in an inaccurate representation of small temperature differences within the city. Additionally, precipitation data for natural factors is only available on a monthly basis, making it impossible to analyze the impact factors of daytime and nighttime separately. Similarly, GDP and PD data for socio-economic factors are annual, preventing the assessment of seasonal variations in the impact strength, and the physical mechanism by which socio-economic factors affect the surface heat island is unclear. Secondly, the selection of socio-economic factors is incomplete, omitting transportation, pollution, and infrastructure, which limits a comprehensive understanding of human factors’ impacts on regional heat islands. Thirdly, the study only considers the two-dimensional urban form, neglecting the complexity of three-dimensional urban landscapes and their impacts on the SUHII. Finally, Pearson’s correlation analysis is a fundamental yet effective method in the study of the UHI. However, it is important to note that Pearson’s correlation analysis can only describe the strength and direction of the relationship between two variables; it does not provide insights into causality and is limited in its ability to quantify spatial heterogeneity. In future UHI studies, geographically weighted regression models can be utilized, as they can offer better explanations of causality and effectively quantify spatial heterogeneity. Additionally, the study focuses on the city as a whole using average data with lower spatial resolution, lacking detailed research at the grid scale and within the city. Further research should conduct a more detailed analysis of regional heat islands by examining the SUHII at the grid scale in arid zones, considering impact factors at seasonal and diurnal scales, accounting for the impacts of three-dimensional landscapes, and utilizing detailed results to mitigate the SUHII.

5. Conclusions

This study quantifies the seasonal and diurnal variations in the SUHII in Xinjiang, while investigating how natural factors, urban size and form, and socio-economic factors contribute to the regional heat island. The research utilizes land use data to calculate relative surface temperatures and characterize the SUHII. The study further computes the urban scale pattern index based on the urban built-up area range, and examines the seasonal differences in SUHII through Pearson’s correlation analysis and OPGD. Key findings include the following: (1) the SUHII in Xinjiang is typically more pronounced in summer and winter compared to spring and autumn. Moreover, the annual average SUHII is higher at nighttime than during the daytime, stronger during the daytime than at nighttime in summer and winter, and more pronounced at nighttime than during the daytime in spring and autumn. (2) The seasonal disparity of the SUHII is more significant during the daytime than at nighttime, with a decreasing trend observed from summer to winter for both daytime and nighttime SUHII. (3) Our study indicates that the drivers of the SUHII are intricate and seasonal rather than fixed. Natural factors like EL and NDVI, urban size and form factors like PLADJ and AI, and socio-economic factors like the GDP and PD play significant roles in influencing the SUHII in Xinjiang. Among these factors, socio-economic factors, especially the GDP, have the greatest impact on the SUHII across all seasons, except daytime in winter. Therefore, there is an urgent need to address the increasing SUHII and its negative effects on both the population and the environment. To combat this issue, it is crucial to enhance the urban environment, manage urban size, optimize urban landscapes, and reorganize economic activities and population distribution. Furthermore, promoting the adoption of new technologies and increasing public awareness about environmental and self-protection measures can also mitigate the impacts of the UHI. The methodology employed in this study can be replicated to examine the UHI and urban comfort in other arid zone cities, and future research could focus on analyzing the correlations between UHI changes and urban industries, transportation, and air pollution.

The main contribution of this study is its comprehensive analysis of seasonal and diurnal variations in the SUHI and its potential drivers across 22 cities in Xinjiang, China. This research addresses the existing gap in understanding the factors influencing SUHI dynamics in arid regions. Furthermore, the study provides a scientific foundation for comprehending the SUHI phenomenon in these areas and significantly contributes to the development of effective urban environmental improvement strategies. These strategies include optimizing surface conditions, enhancing blue–green spaces, and adjusting economic policies to mitigate the SUHI while promoting sustainable urban development.