The Preliminary Study of Environmental Variations Around the Du-Ku Highway Since 2000

Abstract

Highways and their surrounding areas in mountainous and plateau regions are particularly susceptible to environmental changes, which can significantly impact their safety. In the context of global warming, the magnitude of environmental changes around highways has been further amplified. These environmental disturbances pose substantial risks to highway infrastructure in mountainous regions. By using satellite data and remote sensing techniques, this study focused on the environmental variations of the Du-Ku Highway (DKH) in the Tianshan Mountains and the preliminary revealed shifts in surface water, land surface temperature (LST), normalized difference vegetation index (NDVI), and temperature vegetation dryness index (TVDI) since 2000. The quantitative results showed that the water bodies with area between 0.1 and 0.5 ha showing the most significant growth around the DKH. The LST values are primarily distributed between 280 and 285 K, while the NDVI values are mostly below 0.4, and the TVDI is mainly concentrated at the two extremes. In the context of global warming and its amplified impact on mountainous and plateau regions, these findings offer critical insights that can directly support mountainous highway construction and maintenance strategies by identifying environmental indicators, providing a scientific foundation for making data-driven decisions.

1. Introduction

As linear engineering projects connecting different locations, highways often traverse various geomorphological units such as mountains, plains, and basins. These different geomorphological units create varying environmental distributions around highways and bring about different safety hazards [1,2]. Moreover, the environmental characteristics around highways, such as the distribution of water bodies, vegetation, and permafrost, may change due to the construction and operation of highways. This is particularly evident for mountainous highways located in plateau areas [3,4,5,6].

For highways in mountainous regions with significant altitude variations, changes in the surrounding environment play a critical role in ensuring the safe operation of the highway. In recent years, global climate warming has led to an upward trend in both the frequency and intensity of extreme weather events, such as extreme precipitation, severe droughts, and glacial retreat [7,8,9,10,11,12]. Disasters such as debris flows and glacial lake outburst floods triggered by these climate events pose significant threats to the safe operation of highways in mountainous areas, and have caused environmental changes like desertification, reduced vegetation cover, and permafrost degradation [13,14,15,16]. The 2011 Zonag Lake outburst, for instance, led to an increase in surface water downstream along the Qinghai-Tibet Highway, resulting in the reconstruction of several bridges on the highway [17,18,19,20]. Additionally, debris flows triggered by intense rainfall in the Shenshuicao Gully, located in the Qilian Mountains, destroyed two highway bridges [21]. Besides the disasters caused by intense precipitation, drought-induced desertification can also lead to windblown sand-related issues, compromising the safe operation of highways [22].

To ensure the safe operation of highway projects in mountainous regions, various scholars have conducted research on the surrounding environments using different methods. Li et al. [1], for example, used the National Highway 318 as a case study and proposed a multi-scale engineering geological zonation approach, which classified hazardous sections of the highway based on a variety of data sources. Du et al. [23], employing InSAR and other geoinformation methods, conducted a detailed investigation of surface deformation around the Qinghai-Tibet Highway, and identified four areas with significant subsidence and one distinct uplift area. Huang et al. [24], focusing on sections of the Gonghe-Yushu Expressway that are prone to frequent damage, used ground-penetrating radar, drones and field surveys to analyze the types of highway distress, noting that damage was most frequent in waterlogged areas. Lin et al. [25], through field investigations and ground temperature monitoring, found that the permafrost environment surrounding the Qinghai-Tibet Highway has been affected, with desertification being observed in certain areas.

The aforementioned studies have demonstrated that changes in the surrounding environment of highways can significantly impact the safety of engineering projects, and highways located in plateau and mountainous regions are particularly vulnerable to such environmental disturbances. However, due to the harsh natural conditions, traditional field surveys can only obtain relatively limited data and cannot provide a comprehensive, large-scale assessment of the environment surrounding highways. With the accumulation of satellite data and advancements in computer technology, remote sensing technology is playing an increasingly important role in environmental surveys. Based on the climatic factors and topographic factors obtained by different remote sensing datasets, Mehmood et al. [26] found that a significant correlation has been found between elevation, LST and the occurrence of forest fires. By using Land sat and ERA5 data, the NDVI in Khyber Pakhtunkhwa province of Pakistan has been noted, and showed an upward trend, especially in areas with dense vegetation [27]. Using Landsat-9 data and ground-based data combined with several machine learning algorithms, Anees et al. [28] estimated the forest above-ground biomass (AGB) of the Himalayan region. Through the application of Sentinel-2 data and the integration of multiple machine learning models, Luo et al. [29] improved the estimation of the forest AGB. Previous studies have demonstrated that remote sensing technology performs exceptionally well for environmental monitoring. Against the backdrop of rapid advancements in remote sensing technology and continuous improvements in accuracy, investigations of the environment investigations derived by remote sensing around highways have become increasingly favored by researchers in recent years [23,30,31,32].

In this study, the 5 km area on either side of Du-Ku Highway (DKH) has been established as the research object. The DKH traverses the Tianshan Mountains, where the previous research on this region has primarily focused on the natural climate, hydrological conditions, and natural hazards, with limited attention given to human-engineered projects such as highways. In the context of climate change, environmental changes in the surrounding areas of mountainous highways may trigger natural disasters such as debris flows, floods, and landslides, posing significant risks to the highway safety. Therefore, monitoring the environment around these highways is crucial. This research focuses specifically on the DKH project, conducting a preliminary analysis of surface environmental changes with remote sensing technology since 2000. By selecting different environmental indexes, the study revealed changes in surface water bodies, vegetation and dryness index, and surface temperature around the DKH, a typical mountainous highway. These findings provide valuable insights and scientific support for the construction and maintenance of highways located in mountainous regions.

2. Study Area and Methods

2.1. Study Area

The DKH, also known as the Dushanzi-Kuqa Highway, is the southern section of China’s National Highway 217 (G217 National Road), connecting northern and southern Xinjiang. Due to its route through the Tianshan Mountains, it is also referred to as the Tianshan Highway. The Tianshan Mountains are located in central Asia and are characterized by a temperate continental climate, with distinct seasons and significant temperature variations. Precipitation in the Tianshan area is unevenly distributed due to the elevation differences, with more precipitation on the northern slopes than on the southern slopes. The runoff generation and collection processes in this region are complex and the main natural hazards in the Tianshan area include geological and meteorological hazards. Geological hazards consist of debris flows, collapses, and landslides, while meteorological hazards include drought, strong winds, snow damage, frost, and heavy rain-induced floods [33,34]. From an administrative division perspective, the DKH starts in the Dushanzi District of Karamay City in northern Xinjiang and extends southward to Kuqa City in Aksu, located on the northern edge of the Tarim Basin in southern Xinjiang, passing through counties such as Nileke, Xinyuan, and Hejing (Figure 1a). The DKH has a significant elevation range, with the highest point exceeding 3400 m at Tielimaiti Tunnel, and an average elevation above 2000 m. Due to climatic conditions such as winter snow and ice in the Tianshan Mountains, the highway is subject to seasonal traffic controls, typically opening from June to October each year. In this research, to achieve a more comprehensive coverage of the environmental conditions surrounding the road, the study area is defined as a 5 km zone on either side of the DKH (10 km wide), covering the entire length of the highway (Figure 1b).

2.2. Data and Methods

2.2.1. Research Data

The topographic and terrain data for the study area were obtained from the SRTM DEM dataset with a resolution of 30 m. For the extraction of surface water bodies, the study utilized the JRC Yearly Water Classification History, v1.4 dataset (YWCH), developed by the European Commission’s Joint Research Centre. The dataset covers annual classifications of global surface water from 1984 to 2021, with a resolution of 30 m. It provides long-term records of water body changes, including classifications such as permanent water and seasonal water, and is widely used in water resource management, environmental monitoring, disaster response, and climate change research. YWCH has an extensive temporal span and high resolution, allowing for systematic and continuous capture of water body changes, making it especially suitable for long-term monitoring of water resource dynamics. In this study, the permanent water band from this dataset was employed to analyze the changes in surface water bodies within the study area from 2000 to 2021.

The MOD11A2 and MOD13A2 datasets were used for the analysis of land surface temperature (LST), vegetation conditions, and drought conditions. Both datasets are global products generated by the MODIS sensor aboard NASA’s Terra satellite. MOD11A2 provides 8-day composite data on LST and emissivity with a resolution of 1 km, suitable for climate monitoring, agricultural management, and ecological analysis. MOD13A2 offers 16-day composite vegetation indices with a 1 km resolution, including NDVI and EVI, which are used to assess vegetation health and land cover, and are widely applied in ecological research and forest management. In this study, the two MODIS datasets were utilized to extract LST and NDVI for the study area from 2000 to 2024. The LST product of MOD11A2 is derived from the thermal infrared bands of MODIS data, using atmospheric correction and cloud-masking algorithms to minimize environmental interference, resulting in more accurate land surface temperature measurements. The NDVI product MOD13A2 specifically uses reflectance values from the visible (red) and near-infrared (NIR) spectral bands. The annual average of both the LST and NDVI data were calculated to minimize seasonal variability and transient atmospheric effects, which formed the basis for calculating the Temperature Vegetation Drought Index (TVDI) of the study area. By using these data, the environmental changes around the DKH are observed spatially and temporally.

2.2.2. Methods

1.

Calculation of some environmental indices:

The shape index (SI) of surface water bodies, also known as the shoreline development index, reflects the degree of shoreline development. A higher shape index indicates a more irregular shoreline. It is calculated by comparing the perimeter of the surface water body to the perimeter of an ideal circle with the same area, thereby measuring the deviation of the water body from an ideal circular shape. The calculation formula is:

$$SI={\displaystyle \frac{P}{2\sqrt{\mathrm{\pi }A}}}$$

(1)

where $SI$ is the shape index of the surface water body, $P$ is the perimeter of the surface water body, and $A$ is the area of the surface water body.

The fractal dimension (FD) of surface water bodies is an index that reflects the complexity of the water body’s shape. The value of the fractal dimension ranges from 1 to 2. The closer it is to 1, the simpler the shape of the surface water body; the closer it is to 2, the more irregular and geometrically complex the shape. The calculation formula is:

$$FD={\displaystyle \frac{2\ln \left(P/4\right)}{\ln A}}$$

(2)

where $FD$ is the fractal dimension of the surface water body, $P$ is the perimeter, and $A$ is the area of the surface water body.

Based on the LST and NDVI data obtained from MODIS data, the Temperature Vegetation Dryness Index (TVDI) for the study area can be calculated. By constructing an NDVI-LST feature space and utilizing the triangular relationship between NDVI and LST, the equations for the dry and wet boundaries are established to calculate the TVDI value for the study area. The method for calculating the TVDI is as follows:

$$\left\{\begin{array}{c}TVDI={\displaystyle \frac{LST-LS{T}_{MIN}}{LS{T}_{MAX}-LS{T}_{MIN}}}\\ LS{T}_{MIN}=a+b\times NDVI\\ LS{T}_{MAX}=c+d\times NDVI\end{array}\right.$$

(3)

where $LS{T}_{MIN}$ and $LS{T}_{MAX}$ are the equations of the wet and dry edges, respectively, with $a$ and b representing coefficients for the wet edge, and $c$ and $d$ representing coefficients for the dry edge. $NDVI$ and $LST$ are the pixel values involved in the calculation. In the NDVI-LST feature space, for pixels with the same NDVI value, the corresponding maximum and minimum LST values are determined. Linear regression equations are then obtained for these maximum and minimum values to establish the dry and wet edges of the TVDI. The TVDI value is 1 at the dry edge and 0 at the wet edge [35]. To ensure the robustness and reliability of two edges, this study first merged the annual NDVI and LST data and then performed random sampling on the merged dataset. After selecting a sufficient sample size, the equations for the dry and wet boundaries were determined. By obtaining the equations of dry and wet edges properly, the TVDI values of all pixels within the study area are computed.

2.

Research route:

The research route of this study is illustrated in Figure 2. After delineating the study area by creating buffer zones, LST, NDVI, and yearly surface water (YSW) distributions of the study area since 2000 were obtained using the Google Earth Engine (GEE) platform, and the topography and terrain of the study area were extracted. Subsequently, an LST-NDVI feature space was constructed by using the acquired LST and NDVI data, enabling the calculation of dry and wet edge equations. These equations were then used to calculate the TVDI for the study area. Following the classification and segmentation of the LST, NDVI, and TVDI, the sequences of LST (LSTs), NDVI (NDVIs), and TVDI (TVDIs) under different conditions were derived. As for the water bodies, after vectorizing the obtained water body data, the area, perimeter, SI, and FD of each water body in the study area were calculated, revealing changes in the water bodies around the DKH since 2000. Through a comprehensive spatiotemporal analysis of the YSW, LST, NDVI, and TVDI, this study elucidates the environmental changes in the surface conditions surrounding the DKH.

3. Results

3.1. The Topography and Terrain Along the DKH

Due to the significant elevation differences and the numerous peaks along the Duku Highway, there is considerable variability in the distribution of slopes and aspects along the route (Figure 3). From Dushanzi to Haxilegen Mountain, the elevation rises from below 1000 m to approximately 3500 m. In the early stages of this section, the slope is predominantly within 0° to 10°, with the aspect mainly concentrated between 90° and 180°. At around 28 km, the elevation reaches nearly 1900 m before beginning to descend, with slopes primarily ranging between 10° and 30° and aspects mainly concentrated between 180° and 270°. The elevation then continues to rise, with most slopes exceeding 20°, and aspects predominantly between 270° and 360°. Between Haxilegen Mountain and Yuximolegai Mountain, slopes are generally greater than 10°, with aspects mainly concentrated between 180° and 270°. Between Yuximolegai Mountain and Tielimaiti Mountain, the elevation first decreases and then increases, with slopes primarily ranging from 10° to 30° between the 170 km and 248 km marks, and aspects mainly between 180° and 270°. At approximately 220 km, the elevation is around 1700 m, reaching about 2700 m at 248 km. However, from 248 km to 376 km (Tielimaiti Mountain), the terrain is relatively flat, with slopes mostly concentrated between 0° and 10°, and aspects mainly between 90° and 180°. From Tielimaiti Mountain to Kuqa, the elevation shows a descending trend, and near the endpoint at Kuqa, the slopes are around 0° to 10°, with aspects ranging between 90° and 270°.

3.2. The Characteristics Surface Water Along the DKH

By analyzing the characteristics of surface water bodies surrounding the DKH, both the number and area of surface water bodies have shown an increasing trend since 2000. The total number of surface water bodies grew from 1701 in 2000 to 2059 in 2021, representing an increase of 21.04%. According to the linear fitting results, the average annual increase in the number of surface water bodies is 12.5 per year (Figure 4a). The minimum total area of surface water bodies in any given year occurred in 2004, at 69.98 km², while the maximum occurred in 2019, at 99.09 km², with an average annual increase of 1.107 km²/year. In contrast to the number and overall morphology of surface water bodies, the average fractal dimension and shape index of surface water bodies each year did not exhibit significant changes, with the average fractal dimension remaining around 1.05, and the shape index ranging primarily between 1.37 and 1.39 over the years.

Although the surface water in the study area shows an increasing trend in both area and number overall, water bodies in different areas exhibit varying trends. Based on the area distribution of each water bodies and considering the individual pixel area of the dataset (0.09 ha), the water bodies within the study area are categorized into three segments: less than 0.1 ha, between 0.1 to 0.5 ha, and greater than 0.5 ha. By classifying the surface water bodies into different segments, it becomes evident that water bodies smaller than 0.1 hectares constitute the largest proportion in terms of number, followed by those with areas between 0.1 and 0.5 hectares, while water bodies larger than 0.5 hectares represent the smallest proportion (Figure 5a). Although water bodies smaller than 0.1 hectares are the most numerous, their proportion has shown a declining trend over the years. In 2000, they accounted for 59.17%, reaching their lowest proportion in 2012 at 53.22%, before recovering to 56.5% in 2021. Conversely, the proportion of water bodies with areas between 0.1 and 0.5 hectares has shown an increasing trend, rising from 26.67% in 2000 to a peak of 30.87% in 2013, with a slight decrease thereafter, reaching 29.95% in 2021. Water bodies larger than 0.5 hectares have exhibited a slight downward trend in their proportion, decreasing from 14.16% in 2000 to 13.54% in 2021. In terms of the absolute number of water bodies across different size categories, all sizes have shown an increasing trend. Specifically, the number of water bodies smaller than 0.1 hectares has increased by 5.1 per year, those between 0.1 and 0.5 hectares by 6.2 per year, and those larger than 0.5 hectares by 1.3 per year (Figure 5b).

3.3. LST, NDVI and TVDI Around the DKH

The LST data around the DKH since 2000 has revealed the temperature distribution spatially and temporally. From Figure S1, the higher LST values are mainly distributed at the start and end points of the DKH, where the elevation is lower and the terrain is relatively flat. Conversely, in areas with high altitudes and relatively significant topographical variations, the LST values are lower. Classifying the LST in the study area based on different temperature ranges reveals the area change trends corresponding to different LST values (Figure 6). The area with temperatures below 275 K is the smallest, with a minimum of 258.26 km², representing 5.47% of the total study area (in 2000), and a maximum of 621.1 km², accounting for 13.16% of the total area (in 2003). The area with a LST between 275 and 280 K ranges from 14.81% (in 2008) to 27.22% (in 2010) of the total area. The largest proportion is found in the 280 to 285 K range, with areas ranging from 787.89 km² (in 2003) to 1716.53 km² (in 2020), accounting for 16.7% and 36.38% of the total area, respectively. Beyond 285 K, the proportion of each temperature range decreases as the temperature increases. The minimum proportions for the 285 to 290 K, 290 to 295 K, and above 295 K ranges are 8.24%, 7.26%, and 7.31%, respectively, while the maximum proportions are 25.39%, 24.72%, and 23.95%, respectively. From the area distribution of different LST ranges in specific years, it is evident that the LST in the study area is primarily concentrated between 275 and 290 K.

The overall NDVI values in the study area are relatively low. At the starting and ending points of the DKH, the NDVI value is relatively low. However, at the mountainous segments where the elevation becomes higher, the NVDI value shows an increasing trend (Figure S2). Considering that both the start and end points of the study area are located in urban regions, this phenomenon could be attributed to human activities, such as urban development and grazing, which reduce the NDVI in these areas. In higher-altitude mountainous regions with less human interference, the NDVI remains relatively high. By classifying this into five intervals ranging from less than 0.1 to greater than 0.4, it is observed that the area with NDVI values less than 0.1 shows a decreasing trend, while the areas within the 0.1 to 0.2, 0.2 to 0.3, and 0.3 to 0.4 intervals exhibit slight increasing trends. The interval with NDVI values greater than 0.4, however, shows a slight decreasing trend (Figure 7). The NDVI distribution across different years also indicates that the NDVI values in the study area are primarily concentrated in regions with values less than 0.3. Specifically, the area with NDVI values less than 0.1 ranges from 1006 to 1432 km², accounting for 22.28% to 31.7% of the total area. The area with NDVI values between 0.1 and 0.2 ranges from 884 to 1595 km², with a proportion ranging from 25.43% to 38.32%. The area with NDVI values between 0.3 and 0.4 accounts for 10.4% to 17.92%, while the area with NDVI values greater than 0.4 accounts for less than 5.2%.

After obtaining the LST and NDVI data for the study area, the dry and wet boundaries of the TVDI can be constructed within the NDVI-LST feature space (Figure 8). Using the dry and wet boundary equations, the TVDI distribution for the study area can then be calculated. Analysis of the NDVI-LST feature space over the years reveals that most data points in the study area fall within the NDVI range of 0.2 to 0.4 and the LST range of 280 to 290 K. Regarding the slopes of the dry and wet boundary equations, the slope of the dry boundary equation is generally concentrated around −18, with a minimum of −23.6 in 2021 and a maximum of −9.1 in 2006. The median slope of the wet boundary equation is 20.6, with a minimum of 17.08 in 2013 and a maximum of 24.4 in 2003.

From the trends in the TVDI across different intervals in the study area, it is observed that areas with TVDI values less than 0.2 and greater than 0.8 exhibit a declining trend, while areas within the 0.2 to 0.4 and 0.4 to 0.6 intervals show an increasing trend (Figure 9). The 0.6 to 0.8 interval remains relatively stable. In terms of the proportion of the TVDI across different years, the area distribution of the TVDI exhibits a pattern of high proportions at both extremes and lower proportions in the middle. Specifically, the area with TVDI values greater than 0.8 ranges from 1385 to 2172 km², accounting for 29.37% to 46.03% of the total area, making it the most significant range. This is followed by the area with TVDI values less than 0.2, which accounts for 19.88% to 25.14%, with an area between 937 and 1186 km². The areas within the 0.2 to 0.4 and 0.4 to 0.6 intervals range from 558 to 1127 km² and 399 to 1195 km², respectively. The smallest area proportion is observed in the 0.6 to 0.8 interval, with areas ranging from 199 to 933 km², accounting for 4.23% to 19.78% of the total area.

The distribution patterns of the three surface environmental indices, LST, NDVI and TVDI show distinct differences when considering their data distributions and the temporal variations in their median values (Figure 10). The median LST value ranges from a maximum of 286.38 K to a minimum of 280.76 K. Although there is a noticeable downward trend in the median values during 2010 and 2011, the overall variation is relatively small. The LST data is concentrated around the median. In contrast, the NDVI data distribution exhibits two high-density (peak) regions, with the median falling primarily between these two peaks. The median NDVI value ranges from a maximum of 0.23 to a minimum of 0.17. The median TVDI value ranges between 0.35 and 0.55, with a generally stable trend, though a notable decline is observed in 2010 and 2011. The TVDI data is more concentrated at the extremes, with less concentration around the median.

4. Discussion

4.1. Climate Conditions Influence the Surface Environment

Climate conditions play a pivotal role in shaping the surface environment, influencing ecosystems and engineering projects in profound ways. With the development of remote sensing technology, obtaining and analyzing surface environment with climate conditions using satellite data has become increasingly mainstream. Based on the MODIS data and machine learning models, Shoaib et al. [36] evaluate the impact of climate factors (rainfall, temperature) and human activities (urbanization) on vegetation cover. With ERA 5 data, MODIS data and DEM data, Kaleem et al. [37] quantified the impact of elevation and climate variables on the net primary productivity (NPP) in northern Pakistan, which is a typical mountainous region.

Since the surface environment is influenced by multiple factors, the integrated use of meteorological and remote sensing data can reveal multi-layered patterns and trends, especially around the construction. With meteorological data surrounding the highway, Song et al. [38] indicated that the resilience of vegetation is related to the terrain of the highway segment, vegetation coverage, climate conditions, and the intensity of engineering activities. Based on the daily mean values of the meteorological data and UAV data, Luo et al. [39] demonstrated the ground surface temperature around the Qinghai-Tibet Highway.

Previous studies have demonstrated that climate conditions, especially temperature and precipitation, could significantly impact the surface environment around highways. By using ERA5 data, the mean annual air temperature (MAAT) and precipitation around the DKH region were obtained (Figure 11). The data reveals a clear upward trend in the MAAT, with an increase rate of 0.06 °C per year. The highest temperature was recorded in 2022, with a temperature of 1.9 °C, while the lowest temperature occurred in 2014, at −0.78 °C. Conversely, the annual precipitation exhibits a decreasing trend, with a reduction rate of 3.84 mm per year. The highest recorded precipitation was 695 mm in 2000, while the lowest, 595 mm, occurred in 2014. Specifically, in 2010 and 2011, a significant increase in precipitation accompanied by a marked decrease in temperature appeared. This change led to a reduction in LST across the study area, which inhibited vegetation growth. Consequently, despite increased precipitation, the NDVI did not show a substantial rise. The decrease in LST and the relative stability of the NDVI resulted in a downward trend in the TVDI during 2010 to 2012. However, with subsequent increases in the MAAT and reductions in precipitation, the TVDI began to rise again (Figure 10).

The increase in air temperature and decrease in precipitation often contribute to droughts, which can lead to a reduction in surface water bodies. However, in the DKH region, both the number and area of surface water bodies have shown an increasing trend, contrary to the decreasing precipitation trend. Given the significant glacier distribution in the Tianshan Mountains, through which the road passes, it is hypothesized that the increase in surface water is primarily fed by the melting glaciers.

4.2. The Variance and Trend of Different Indexes

Monitoring variance and trends in LST and the NDVI and TVDI is essential for assessing environmental variations, especially in areas vulnerable to climate change and human interventions. Recent studies have highlighted the importance of these indicators for understanding changes in soil moisture, vegetation health, and drought conditions. For instance, Li et al. [40] analyzed the spatiotemporal variations in drought using LST, NDVI, and TVDI data, providing insights into how these indexes reveal environmental stress patterns across different climate zones. Similarly, Ding et al. [41] found that these indicators offer a valuable framework for analyzing vegetation responses to climate-induced drought in the Yellow River Basin. Additionally, Ali et al. (2024) demonstrated the efficacy of combined LST and NDVI trend analyses in evaluating vegetation health under varying drought and thermal stress, underscoring the indicators’ importance in climate-sensitive studies [42].

As a typical vulnerable region to climate change, the analysis of the variance and trends of the LST, NDVI and TVDI around mountainous highways is significant. By calculating the standard deviation of pixel values across different years and performing linear regression, the variability and trends of LST, NDVI, and TVDI for each point in the study area can be determined (Figure 12). The analysis of LST variability and trends reveals that most of the DKH region falls within the low variability, positive trend category, indicating a slow upward trend in LST across most of the area. However, in the region at latitude 43° and longitude 84°, LST exhibits high variability and a negative trend, suggesting that LST in this area is decreasing. The NDVI generally shows a low variability, positive trend, indicating a slow increase in vegetation around the DKH. The TVDI shows low variability with no significant trend, indicating that the drought conditions in most of the DKH region remain relatively stable. However, in the area at latitude 43° and longitude 84°, there is a high variability, negative trend, suggesting a decrease in drought conditions in that region.

The analysis of LST, NDVI, and TVDI trends across the DKH region highlights the complex interplay of climatic and environmental factors influencing this area. The low variability and positive trends observed for LST and NDVI suggest gradual warming and a concurrent increase in vegetation cover, likely due to regional changes in climate and adaptation processes in local vegetation. These trends, however, may also indicate subtle shifts in ecosystem responses to long-term temperature and precipitation patterns, possibly driven by changes in seasonal precipitation, solar radiation, or soil moisture. Conversely, regions of high variability and distinct negative trends, such as those observed at latitude 43° and longitude 84°, warrant closer examination. These areas may be experiencing unique local conditions that disrupt the general upward trends, potentially due to a combination of topographical influences, human activity, or microclimatic anomalies. For instance, the decrease in LST and TVDI variability in these areas could imply stabilization of surface temperatures alongside reduced drought stress, possibly enhancing vegetation growth or altering species composition over time.

5. Conclusions

Amid the broader context of climate change, the environment surrounding many man-made structures has undergone significant transformations. As linear infrastructures spanning diverse terrains, highways are subject to particularly complex impacts. This study focuses on the DKH, an area characterized by significant topographical variation. By analyzing surface water bodies, LST, NDVI and TVDI data from 2000 onwards, the study draws the following conclusions:

• Both the area and number of surface water bodies around the DKH have shown an increasing trend, while the FD and SI of the water bodies have remained relatively stable. In terms of distribution, water bodies smaller than 0.1 hectares account for the largest proportion, while those larger than 0.5 hectares account for the smallest. The greatest increase was observed in water bodies ranging from 0.1 to 0.5 hectares. The highway projects in mountainous regions should account for potential changes in local water bodies to mitigate effects on hydrological influence in the future.
• By categorizing different intervals, the study revealed that the LST in the area along the DKH is predominantly between 280–285 K, accounting for 16.7% to 36.38% of the region, with a median LST varying from 280.76 K to 286.38 K. The NDVI values are generally low, mostly below 0.4, and are characterized by two distinct peaks with the median NDVI values peaking at 0.23 and bottoming out at 0.17. The TVDI is predominantly found at the extremes, with a significant area showing values above 0.8 (29.37% to 46.03%) and below 0.2 (19.88% to 25.14%), while the median TVDI values fluctuate between 0.35 and 0.55. Due to the low value of NDVI, a vegetation buffer around the mountainous highway might reduce disasters of the engineering.
• Analysis of the variability and trends of LST, NDVI, and TVDI at individual points shows that LST and NDVI generally fall within low variability, positive trend intervals across most regions, while TVDI remains largely in low variability, no significant trend intervals. However, in the region at latitude 43° and longitude 84°, both the LST and TVDI exhibit high variability and negative trends. These high variability zones should receive more attention to avoid the occurrence of disasters.

This study comprehensively utilizes various remote sensing datasets to reveal the spatiotemporal changes in the surface environment around the DKH from multiple perspectives. However, some limitations remain, such as the difficulty of capturing deeper characteristics using only annual data due to significant seasonal variations along the DKH, and the lack of effective exclusion of rivers from the surface water body data. Future research should address these shortcomings through more detailed exploration and investigation.