The turbidity of glacier meltwater runoff serves as a key indicator of glacier material output， and accurately obtaining turbidity data is essential for understanding the impact of glacier melting on downstream environments. Traditional methods for measuring the turbidity of glacier meltwater runoff primarily rely on water-contact observations， which are limited by high costs， low efficiency， and difficulty in continuous monitoring. Therefore， this study conducted experiments on Kuoqionggangri No.1 Glacier in the central-southern Qinghai-Xizang Plateau， and proposed a visual monitoring approach for glacier meltwater runoff turbidity based on low-cost field cameras and deep learning methods. Using images and turbidity data of the meltwater runoff from the No.1 Glacier， a turbidity prediction model was established using MobileNetV1. The results showed a significant correlation between the color and turbidity of the meltwater runoff from Kuoqionggangri No.1 Glacier， with relatively substantial diurnal variations in turbidity. The proposed prediction model for glacier meltwater runoff turbidity could achieve turbidity prediction. Within the range of 0~5 000 nephelometric turbidity unit （NTU）， the model achieved a mean absolute error （MAE） of 183.93 NTU and a coefficient of determination （R²） of 0.45. In the range of 0~200 NTU， the MAE of the turbidity predictions was 9.14 NTU， showing reductions of 19.16 NTU， 15.97 NTU， 3.21 NTU， 5.14 NTU， and 2.81 NTU compared to ShuffleNet， GhostNet， DenseNet121， InceptionV3， and ResNet50， respectively. The R² value was 0.93， representing increases of 0.3， 0.26， 0.04， 0.05， and 0.09， respectively. This study provides a novel approach for low-cost， high-frequency， and continuous monitoring of turbidity variations in glacier meltwater runoff in plateau regions.

The Qinghai-Xizang Plateau has a complex natural environment， characterized by the most extensive and highest-elevation permafrost in low-latitudes regions. Meanwhile， the plateau’s crust is strongly uplifted， and disasters such as earthquakes and landslides occur frequently. Unlike the instability of slopes in non-frozen soil areas， the melting of permafrost with high temperature and ice content is the primary cause of instability in permafrost slopes， and landslides can develop even on low and gentle slopes. If a strong earthquake occurs during the warm season—a period characterized by frequent freeze-thaw instability events—the slope is affected by the freezing and stagnant water effect， and the moisture content in the upper soil of the permafrost roof increases， forming a locally saturated area. Combined with seismic action， it is likely to evolve into a series of catastrophic landslides. Currently， an insufficient understanding of the dynamic response characteristics and amplification effects of soil slope underlying permafrost seriously restricts the development of geological hazard prediction theories and prevention technologies in high-intensity permafrost areas. This study took a typical low and gentle permafrost slope of Fenghuo Mountain in a high-intensity area of the Qinghai-Xizang Plateau as the research object. Based on field investigation and laboratory test results， a fluid structure coupling dynamic simulation model was constructed， and numerical simulation experiments were conducted to investigate the dynamic response characteristics of the slope under different seismic waves and earthquake intensities. The results showed that the low and gentle permafrost slope followed the acceleration elevation amplification effect， but the acceleration response would decay in the saturated soil layer above the permafrost roof. The attenuation effect gradually became more significant with the increase in earthquake intensity， with an average attenuation of 52.1% under a seismic load of 1.00 g. The attenuation effect gradually increased from the lower part of the slope to the slope shoulder. Analysis of the physical mechanism of acceleration attenuation indicated that the significant differences in physical and mechanical properties between permafrost and saturated soil layers led to a decrease in seismic wave velocity， increased dispersion， and waveform changes. Additionally， it was affected by the repeated reflection and refraction of seismic waves at the permafrost roof， coupled with the deformational flow of soil under dynamic action， resulting in significant attenuation of wave energy and amplitude at this location. The dynamic action can also induce the excess pore water pressure in the active layer of permafrost slopes， with the induced intensity increasing as the earthquake intensity increased. The excess pore water pressure in the saturated layer above the permafrost roof was stronger than that in the natural soil layer. The generation and accumulation processes of excess pore water pressure are influenced by the acceleration time history. Finally， a mechanism for the excitation of excess pore water pressure in saturated soil layers based on intraregional vibration shear action was proposed through instantaneous analysis. The excitation mechanism at the slope shoulder was mainly affected by the non-coordinated movement and shear action of the frozen soil layer and active layer. As the elevation decreased， the motion of the saturated layer in the lower part of the active layer increased. Additionally， this mechanism at the mid and the toe of the slope was affected by the non-coordinated motion and shear effect of the saturated layer relative to the upper and lower layers. This study addresses the sustainable development of the ecological environment and the safety requirements of major engineering construction and operation on the Qinghai-Xizang Plateau. The research findings have important scientific significance for improving the prevention and mitigation capabilities of earthquake and landslide disasters on the Qinghai-Xizang Plateau.

The Xinjiang-Xizang Highway （G219）， as the main transportation route in and out of Xizang， plays a crucial role in the regional development. However， frequent adverse geological phenomena along this Highway have intensified road damage， seriously compromising highway safety and stability. This study employed a deep learning model to identify and extract rock debris along the Xinjiang-Xizang Highway. Subsequently， a frost cracking model was utilized to explore the development patterns and influencing factors of rock debris. Finally， fractal dimension theory was coupled with rock strength degradation model to reveal the formation and evolution processes of rock debris. The results showed that the deep learning model could effectively identify rock debris with vertical zonation and horizontal orientation. Analysis using the frost cracking model revealed that temperature was the key factor controlling the growth dynamics of rock debris. Weathered debris from frost cracking served as a major material source during the formation alongside other adverse geological phenomena. Based on fractal dimension theory combined with an improved rock strength degradation model， the average rate of dimensional reduction evolution along the Xinjiang-Xizang Highway was estimated to be approximately （3.40~5.67）×10^-8⋅a^-1. This study highlights the influence of frost cracking on adverse geomorphic processes and elucidates the mechanism of growth and development of rock debris. The findings offer novel insights for predicting rock debris， thereby providing scientific references for geological hazard prevention and hazard mitigation decision-making.

Driven by global climate change， extreme climate events on the Qinghai-Xizang Plateau have evolved rapidly， significantly reshaping regional ecosystems and surface characteristics. Based on daily meteorological data from 45 meteorological stations on the Qinghai-Xizang Plateau from 1961 to 2020， this study calculated and analyzed the spatiotemporal variation trends of extreme climate indices to investigate the spatiotemporal characteristics of extreme climate events. The results showed that： temporally， the Qinghai-Xizang Plateau exhibited an overall warming and humidification trend over the past 60 years. The number of cold nights decreased most rapidly at a rate of -5.3 d⋅（10a）^-1， while the number of warm nights increased most rapidly at 5.6 d⋅（10a）^-1. Annual precipitation increased at a rate of 9.43 mm⋅（10a）^-1， with asymmetric warming between day and night—the minimum value of the lowest temperatures （3.1 ℃） showed the greatest increase ［0.5 ℃⋅（10a）^-1］. The fitting differences between extreme precipitation indices and temperature indices revealed that the regional warming characteristics of the Qinghai-Xizang Plateau were more pronounced than humidification. Spatially， most stations on the plateau exhibited asymmetric changes in cold and warm extremes and diurnal temperature variations， and the variations in extreme precipitation events showed high spatial heterogeneity. Additionally， the spatial distribution of the variation trends in ice days and consecutive dry days showed north-south asymmetry， and consecutive wet days exhibited both increasing and decreasing trends. Correlation analysis showed that significant relationships （P<0.05） were observed among different temperature extreme indices， between the number of frost days/ice days and temperature extremes， and between temperature extremes （including the number of frost/ice days） and altitude/latitude， with correlation coefficients greater than 0.5. The correlations among the other temperature indices， as well as between them and latitude/longitude， were not significant （P>0.05）， with correlation coefficients less than 0.5. In contrast， the extreme precipitation indices were significantly correlated with each other and with latitude （P<0.05）， with correlation coefficients greater than 0.5， while their correlations with altitude or longitude were relatively low， with correlation coefficients less than 0.5. This study provides a scientific basis for regional climate prediction and disaster early warning on the Qinghai-Xizang Plateau， holding significant importance.

Snow data are crucial for research on climate change， hydrological processes， and disaster assessment. However， obtaining accurate snow cover data in terms of spatial distribution remains a challenge at present. This is primarily due to the dynamic characteristics of snow cover， making acquiring related data more difficult. Traditional field measured data offer high accuracy but are limited in providing large-scale spatiotemporal distribution information of snow cover. Satellite remote sensing can obtain snow cover data with large spatial coverage and long observation periods but is greatly affected by weather and terrain. Additionally， snow cover simulations using models are constrained by differences in model structures and driving data. Reanalysis data， which integrate remote sensing， model simulations， and observational data， can provide accurate snow cover data and offer the potential to capture spatiotemporal variations of snow cover precisely. However， their applicability in specific regions still requires validation. To address this issue， this study conducted a comparative analysis of snow cover percentage （SCP） between MODIS data and reanalysis data at daily， interannual， and intra-annual scales during the hydrological years from 2000 to 2020. Based on defined snow parameters， this study calculated snow cover days （SCD）， snow onset date （SOD）， snow end date （SED）， and snow duration days （SDDs）. The Theil-Sen trend analysis and Mann-Kendall significance test were used to comprehensively compare the spatial variation patterns and significance of snow-related parameters derived from remote sensing and reanalysis data. Using meteorological station data from ground observations， this study compared and analyzed the correlation coefficient （r）， mean bias error （MBE）， and root mean square error （RMSE） between the snow depth from reanalysis data at the corresponding station coordinates and the observed snow depth at the stations. The applicability of ERA5-Land reanalysis snow cover and snow depth data in Xinjiang was assessed. The results showed that： （1） ERA5-Land and MODIS exhibited consistent trends in daily snow cover variations， reflecting seasonal snow cover variations in Xinjiang. In months with low snow cover percentages， ERA5-Land showed good agreement with MODIS. However， notable differences occurred during months with high snow cover （around January）. The two datasets exhibited strong positive correlation and small systematic bias， although ERA5-Land tended to slightly underestimate snow cover. （2） Variations in snow parameters from ERA5-Land were highly consistent with MODIS. ERA5-Land showed a larger area of unstable snow cover and tended to underestimate stable snow cover. In some low-altitude areas， ERA5-Land had delayed SOD and SED compared to MODIS， and SDDs were shorter than those from MODIS. In high-altitude areas， ERA5-Land simulated SOD， SED， and SDDs with higher accuracy： SOD was not significantly delayed， SED was not significantly advanced， and SDDs were not significantly shortened. （3） Comparison between ERA5-Land reanalysis snow depth data and station data showed that the correlation was the weakest at the Tianshan Daxigou station （r=-0.16）. Although the MBE （2.85 cm） and RMSE （8.91 cm） were relatively small， there was an overall overestimation. At other stations， ERA5-Land demonstrated good correlation with observed snow depth， with correlation coefficients ranging from 0.42 to 0.94. Although ERA5-Land captured the variation trend of snow depth （mean r=0.72）， there were significant differences in correlation and errors among different stations. Overall， ERA5-Land slightly overestimated snow depth （mean MBE=6.46 cm）， indicating the need for further optimization to improve simulation accuracy. Snow cover in Xinjiang is not only a crucial source of freshwater but also a key factor in regulating regional climate and maintaining ecological balance. Therefore， accurate monitoring and assessment of snow dynamics are crucial for the sustainable development of Xinjiang. As the latest global reanalysis dataset， ERA5-Land provides long-term and large-scale data support for snow cover-related research. However， although widely used globally， its applicability in Xinjiang has not been thoroughly assessed. Xinjiang’s unique topography and climatic conditions may significantly influence the accuracy of reanalysis data assessment. Therefore， a systematic assessment of ERA5-Land’s applicability in Xinjiang holds important scientific and practical significance. In the context of global climate change， this study provides a scientific basis for the selection and application of ERA5-Land data in snow cover hydrological modeling in Xinjiang， aiming to provide references for related research on snow cover hydrology using reanalysis data in Xinjiang.

Using conventional meteorological observation data and NCEP 6-hourly reanalysis data， a diagnostic analysis was conducted on the causes of the severe cold wave with snowfall in the central and western regions of Inner Mongolia from January 19 to 22， 2024. The characteristics of reflectivity factor， radial velocity， temperature， humidity， wind， and other elements observed by vertical detection instruments—including Doppler weather radar， millimeter-wave cloud radar， wind profiler radar， microwave radiometer， GNSS/MET water vapor—were summarized. The results showed that： （1） This process was characterized by prolonged snowfall， intense temperature drop， and persistent multi-day low temperatures， with minimum temperatures at multiple stations reaching or exceeding extreme thresholds on the following day. The 500 hPa westerly trough， surface inverted trough， and cold high-pressure system were the main influencing systems. （2） The snowfall in Xilingol League exhibited characteristics of cold-pad snowfall， but the cold pad was shallow and weak. Low-level easterly winds observed in Doppler weather radar radial velocity maps， along with southerly airflow， provided sufficient moisture for the widespread snowfall in regions represented by Hohhot and formed strong moisture convergence. （3） The cold centers reached -44 ℃ and -24 ℃ at 500 hPa and 850 hPa， respectively， with the cold high-pressure center at 1 080 hPa in the key area. The 24-hour pressure change was ≥13 hPa and the 3-hour pressure change was ≥3 hPa. The thickness variation between 850 hPa and 500 hPa could serve as a reference for forecasting severe cold waves. （4） Characteristics from the microwave radiometer and GNSS/MET water vapor showed that water vapor increase/decrease preceded the onset/cessation of snowfall by 2~3 hours. The drop in 2-meter air temperature lagged behind the temperature drop in the 2~4 km layer measured by the microwave radiometer by approximately 3 hours. The reflectivity factor intensity of the millimeter-wave cloud radar was ≤25 dBz， with strong reflectivity corresponding to periods of intense snowfall. Cloud development and lowered cloud base heights were observed when the low-level southwesterly winds strengthened on wind profiler radar. The weakening of southwesterly winds in the middle and upper layers corresponded to the lowering of the cloud top height. Snowfall began when the echo reached the ground， and the snow tended to end when the low-level winds shifted to northwesterly. The combination of wind field vertical variation and reflectivity factor characteristics can determine the onset and cessation of snowfall. This study deepens forecasters’ multi-faceted understanding of severe cold waves with snowfall events and provides a scientific basis for their refined forecasting.

The engineering construction in seasonal permafrost regions has been persistently influenced and restricted by soil frost heaving， development of effective methods to restrain frost heaving has emerged as an important issue in these areas. This paper proposes the use of wheat straw as a kind of soil reinforcement material to address this problem. Firstly， based on experimental requirements， a high-precision temperature controlled frozen soil tensile strength device is independently developed. Using the device the tensile strength tests are conducted on straw-reinforced frozen silt samples under various conditions， and the influence of straw content， straw length， test temperature and soil water content on the tensile strength are analyzed. Secondly， based on the above experimental results and measured freezing temperature of straw-reinforced silt previously， tensile strength tests and one-dimensional frost heaving tests are conducted on specimens with fixed straw content and variable straw length after freezing at the temperature close to freezing temperature. The relationships between tensile strength near freezing temperature and frost heaving amount， frost heaving ratio are studied， and the mechanisms of straw reinforcement to suppress frost heaving are analyzed from the perspective of the influence of tensile strength on frost heaving. The research results indicate that from a macro perspective， soil frost heave is mainly caused by the growth of ice lenses， which require the generation of tensile forces greater than the soil's tensile strength. Straw reinforcement can effectively improve the tensile strength of soil， thereby restraining the development of frost heaving. The relationships between tensile strength， frost heaving amount， and frost heaving ratio are observed to be approximately linear. From a micro perspective， the mechanism of straw reinforcement to restrain frost heaving is mainly attributed to the following three factors： ice cementation， friction between soil particles， ice， and straw， and the effect of the straw network. Among which， the cementing effect of ice on soil particles can enhance the cohesion of frozen soil， the friction between soil particles， ice， and straw partially inhibits the sliding of soil particles. The network effect of straw strengthens the overall integrity of the soil， constraining its deformation. These three factors collectively contribute to an increase in the tensile strength of straw reinforced frozen silt. In addition， in terms of tensile strength of straw-reinforced frozen silt， it is related to the straw length， the content of the straw， the test temperature， and the water content of soil， among which the water content and the test temperature are the most influential factors； an optimum straw content or straw length could be found to obtain the maximum tensile strength.

The northwestern region of China has extensive areas of saline soil and frozen soil， where concrete infrastructures are consistently subjected to the coupled deterioration effects of salt erosion and freeze-thaw cycles. The resulting degradation of mechanical properties and reduction in durability have become one of the key factors affecting structural service life and safety. Under such coupled conditions， external salts penetrate the concrete through capillary pores， interacting with phase-change stresses induced by freeze-thaw cycles. This interaction leads to pore structure damage， micro-crack propagation， and degradation of hydration products， collectively accelerating the deterioration process of concrete materials. To mitigate performance degradation in this complex environment and enhance concrete’s resistance to freezing and erosion， researchers worldwide have extensively investigated durability improvement through the incorporation of mineral admixtures. Biochar， as an environmentally friendly material with excellent physical adsorption and pore-filling capabilities， demonstrates significant potential for concrete modification. It can effectively restrict the migration of harmful substances such as chloride ions， optimize the internal pore structure of concrete， and consequently improve its durability under harsh environmental conditions. To quantitatively evaluate the effectiveness of biochar-modified concrete against salt erosion and freeze-thaw cycles， this study systematically investigated the macroscopic physical and mechanical properties and microscopic structural properties of concrete with different biochar contents （0%， 1%， 2%， and 3%）. Specimens were immersed in a 3% NaCl solution by mass and subsequently subjected to different numbers of freeze-thaw cycles （0， 10， 20， 30， 40， and 50 cycles）. At the macroscopic level， changes in physical and mechanical properties of biochar-modified concrete—including water absorption， compressive strength， chloride ion distribution characteristics， and mass loss rate—were measured to reveal the influence mechanisms of biochar on pore structure evolution and damage accumulation processes during salt erosion and freeze-thaw cycles. At the microscopic level， scanning electron microscopy （SEM） was employed to conduct qualitative and quantitative analyses of the microscopic morphology of concrete with biochar contents of 0% and 1%， both before and after freeze-thaw cycles. Particular attention was paid to observing the evolution of pore size and distribution， the propagation patterns of micro-cracks， and changes in the content of key hydration products， including calcium silicate hydrate （C-S-H） gel and calcium hydroxide （C-H）. The results showed that the incorporation of biochar significantly improved the overall performance of concrete under the combined environment of salt erosion and freeze-thaw cycles. Macroscopically， biochar addition effectively enhanced the compressive strength after exposure to salt erosion and freeze-thaw cycles while reducing porosity， water absorption coefficient， and mass loss. Microscopically， biochar particles efficiently filled the micro-pores within the concrete matrix， reducing the number of harmful large pores and inhibiting the initiation and propagation of micro-cracks induced by freeze-thaw cycles. Simultaneously， biochar promoted the cement hydration process， increased the formation of gel products such as C-S-H， decreased average pore diameter， and optimized the overall pore structure. These modifications collectively enhanced the compactness and frost resistance of the concrete. Notably， there was an optimal range for biochar content regarding the improvement of concrete properties. As the biochar content increased， the compressive strength initially increased and then decreased， while durability indicators including porosity， strength loss rate， water absorption， and mass loss rate first decreased and then increased. Comprehensive macro-micro analysis revealed that concrete with 1% biochar content exhibited optimal physical and mechanical performance and damage resistance in the environment of salt erosion and freeze-thaw cycles. Its compactness， frost resistance， and resistance to chloride ion penetration were significantly superior to those of ordinary concrete. After 50 salt erosion and freeze-thaw cycles， compared to ordinary concrete， the 1% biochar-modified concrete demonstrated a 25.71% increase in compressive strength， a 47% reduction in water absorption， and a 49.7% decrease in mass loss rate. Furthermore， the distribution and migration depth of chloride ions within the concrete were significantly reduced. This study provides crucial experimental evidence and theoretical support for promoting and applying biochar as a novel sustainable modification material in concrete structures located in cold and arid regions.

With the development of engineering construction， frozen soil is widely used as building foundations in cold regions. In these practical projects， frozen soil is often under complex stress path conditions. Therefore， investigating the mechanical characteristics and damage behavior of frozen soil under different stress paths is of great significance for the construction and maintenance of engineering in cold regions. Although the mechanical properties of frozen soil under different stress paths have been systematically studied， significant differences are observed in the setting of loading rates during experiments. This discrepancy makes it difficult to effectively compare the experimental data and research conclusions under unified conditions and systems， thereby limiting the universality and general applicability of stress path experimental data and conclusions. To provide a scientific basis for determining the loading rate under different stress paths， this study used the conventional triaxial compression tests performed at a strain rate of 1%⋅min^-1 under strain-controlled conditions as the benchmark. The stress loading rate along the stress path direction was progressively increased， and three constant stress loading rates of 0.15 MPa⋅min^-1， 0.2 MPa⋅min^-1， and 0.25 MPa⋅min^-1 were set under the stress control mode to conduct conventional triaxial compression experiments on frozen soil samples. Through comparison with the conventional triaxial loading curve conducted according to the specifications， the stress loading rate along the stress path direction in this experiment was determined to be 0.2 MPa⋅min^-1. Under this loading rate， cyclic loading and unloading experiments under multiple stress paths were achieved through stress control. Six types of linear stress paths （k=\sqrt[]{\mathrm{3}}/\mathrm{3}， k=1， k=2， k=3， k=∞， k=-3） and their cyclic loading and unloading experiments were conducted. Based on experimental results， the influence of stress path on the deformation， strength， elastic modulus， and damage characteristics of frozen silty clay was analyzed. The experimental results showed that the curves of monotonic loading and cyclic loading-unloading experiments basically coincided during the loading stage， indicating that cyclic loading-unloading experiments under different stress path conditions in the stress control mode did not affect the deformation behavior of frozen silty clay. When 0<k<3， a larger k corresponded to a greater axial strain at the transition point from volumetric shrinkage to volumetric expansion in the volumetric strain curve. When k>3 or k<0， the volumetric deformation characteristics of frozen silty clay were mainly characterized by volumetric expansion， and the volumetric shrinkage was not significant. The strength of frozen silty clay showed a trend of first increasing and then decreasing with the change of stress path slope k. As the mean stress increased， the peak volumetric strain showed a trend of first increasing and then decreasing， and the influence of stress path on the strength of frozen silty clay became smaller. When 0<k<3， a smaller k led to lower stiffness of frozen silty clay. When k>3 or k<0， the stiffness under the k=∞ stress path was greater compared to the k=-3 stress path. Compared with the other five stress paths， the rate and amplitude of elastic modulus decrease were the highest under the k=-3 stress path. The damage variable was defined based on the degradation of elastic modulus， and the damage of frozen silty clay gradually increased with the development of axial strain and eventually stabilized. When 0<k<3， a smaller k resulted in greater damage to frozen silty clay. When k>3 or k<0， the damage under the stress path was less than k=-3. Based on the analysis of the damage evolution patterns of frozen silty clay， a damage evolution function considering the influence of stress path was established， which can well describe the damage behavior of frozen silty clay under different stress paths. The research methods and results can provide scientific basis for the standardization of frozen soil mechanics experiments and the verification of constitutive models.

The freezing of saturated saline soil is a dynamic， coupled process that involves water， heat， salt， and mechanics. This study aims to establish a numerical model based on porous medium mechanics to describe the coupled behavior of water， heat， salt， and mechanics components of saturated carbonate soil and validate the model through unidirectional freezing tests. The model incorporates the law of conservation of energy， Darcy’s law， and the effective stress principle， while introducing an ice-water pressure balance equation that accounts for interfacial energy， replacing the Clausius-Clapeyron equation to describe ice-water phase equilibrium. The basic assumptions are that the soil is an isotropic， saturated， poroelastic medium； the soil particles are incompressible； and sodium bicarbonate crystallization is neglected to ensure the model’s validity and accuracy. To obtain reliable validation data and visually reveal the freezing characteristics of carbonate saline soil， a unidirectional freezing test was conducted. Silty clay from Nong’an County， Changchun City， Jilin Province， was used as the test soil sample. The test was conducted in a TMS 9018-800 environmental freeze-thaw cycler at Kashi University. By controlling the top plate temperature （-20 ℃）， the bottom plate temperature， and the ambient temperature （+2 ℃）， unidirectional freezing conditions were simulated over 96 hours. To monitor temperature， volumetric moisture content， and electrical conductivity at different locations in real time， five 5TE three-in-one sensors were deployed along the height of the soil column. Additionally， displacement sensors were installed on the top plate to continuously record the overall frost heave deformation of the soil column. After freezing， samples were taken from the soil column （at six depths： 0， 5， 10， 15， 20， and 25 cm from the bottom） to measure the final moisture and salinity of each layer. In addition， to determine the initial freezing temperature of the model， the temperature changes at its center were monitored using the same soil sample in a low-temperature environment. It was finally determined that the initial freezing temperature of the soil sample under the experimental conditions （w=21%， c=0.5%） was 271.7 K. Based on the mathematical model established above and the experimental data obtained， the control equations were numerically solved and simulated using multi-physics field coupling calculation software. The simulation results were compared and analyzed with the experimental measurement data. The results showed that： （1） Displacement field： the simulation curve was highly consistent with the experimental data in terms of trend. The simulated value of the final frost heave after 96 hours of freezing （1.61 cm） was very close to the experimental value （1.59 cm）， with a relative error of only 1.2%， which fully proved the reliability of the model in predicting soil deformation. （2） Temperature field： the simulation results reflected the characteristics of the three stages of temperature change with time at different heights （rapid cooling， slow cooling， and steady stage）. The temperature-time curves of each measuring point closely agreed with the measured values. Slight deviations （up to about 2.3 ℃） occurred in local time periods （such as the first 25 hours at a height of 0.20 m）， likely attributable to the simplified treatment of temperature-dependent soil thermophysical parameters such as thermal conductivity and specific heat capacity in the model. However， the overall accuracy met the engineering requirements. （3） Moisture field： the simulated and experimental values for the total soil moisture content along the height of the soil column at t = 96 h showed the same trend. Both clearly demonstrated the typical phenomenon of migration and accumulation of unfrozen water near the freezing front driven by the temperature gradient， resulting in a significantly higher moisture content in this area than the initial value. （4） Salt field： the distribution of salt content along the height of the soil column after 96 hours of freezing was similar to that of the moisture field. The salt content in the unfrozen area decreased， while the salt concentration near the freezing front reached peak values， which was consistent with the experimental results.

With the continuous development of the economy， an increasing number of rock engineering projects in cold regions have been put into construction， such as mine excavation， roads， and slope projects. Extreme temperature differences have caused frost heave cracking， freeze-thaw sliding， and slope instability， posing serious threats to the long-term stability and safety of rock engineering projects while bringing significant challenges to engineering construction and operation. Previous studies on the energy evolution of freeze-thaw rocks have mostly focused on instantaneous failure aspects， while studies considering the freeze-thaw cycle on the evolution patterns of rock creep energy and creep damage characteristics remain relatively limited. This study aims to explore the creep characteristics， energy evolution， and creep damage characteristics of red sandstone under freeze-thaw cycles. Firstly， microscopic inspection tests were conducted on red sandstone subjected to different numbers of freeze-thaw cycles （0， 10， 20， and 30） to analyze the influence of freeze-thaw cycles on the microscopic damage of rocks. Subsequently， uniaxial compression graded loading creep tests were conducted to obtain the creep curves of freeze-thawed sandstone and analyze the evolution patterns of creep strain and corresponding creep strain rates at different stress levels. Finally， by calculating the energy of freeze-thawed sandstone under graded loading creep， the influence of freeze-thaw cycles on elastic energy and dissipated energy was explored， and the evolution patterns of creep damage of freeze-thawed sandstone were analyzed based on the energy dissipation principle. The results showed： The repeated freeze-thaw action continuously developed the pores and fractures of red sandstone， gradually expanded the cross-grain failure area， and intensified the surface damage. As the number of freeze-thaw cycles increased， the first spectral peak of red sandstone showed an increasing trend， while the second and third spectral peaks did not show obvious changes， indicating that the freeze-thaw cycle action promoted the development of small-sized pores in red sandstone. With an increasing number of freeze-thaw cycles， the steady-state strain rate of red sandstone increased， and the steady-state strain rate showed an exponential function growth trend with the increase of stress level. However， the stress level， creep time， and steady-state creep time of red sandstone decreased. Through analysis of the energy evolution patterns of the entire freeze-thaw creep process， it was found that under different stress levels， the repeated freeze-thaw action led to a significant increase in internal pore defects of red sandstone， thereby weakening the resistance of the specimens to deformation and further diminishing the elastic energy storage capacity. In addition， the cumulative elastic energy of freeze-thawed sandstone increased linearly with the stress level. The dissipated energy curve of freeze-thawed rocks experienced “rapid dissipation-slow dissipation-accelerated dissipation” stages. The greater the number of freeze-thaw cycles， the smaller the cumulative dissipated energy. Moreover， the cumulative dissipated energy of sandstone under different numbers of freeze-thaw cycles showed an exponential growth. During the graded loading creep process， the dissipated energy of red sandstone exhibited a certain cumulative behavior. The freeze-thaw cycle action reduced the energy damage of rocks during graded loading creep， and the damage in the creep stage was more sensitive to the freeze-thaw cycle than the loading. The energy damage gradually intensified， especially in the creep stage close to failure， and the energy damage accumulated rapidly to reach a peak. Therefore， when studying the evolution patterns of rock creep damage under graded loading， the cumulative effect of historical loading and creep stages should be considered. The experimental results enhance the understanding of creep damage and energy evolution patterns of freeze-thawed rocks under freeze-thaw cycles， and can provide a reference for evaluating the long-term stability and safety of rock engineering projects in cold regions.

Frost heave is a significant challenge to the stability of engineering foundations in cold regions， and identifying its dominant influencing factors is crucial for engineering prevention and control. This study focused on soil from the Qinghai-Xizang Plateau and systematically investigated the effects of soil type， cold-end temperature， and water supply pressure on frost heave through orthogonal experimental design combined with statistical analysis. Four types of soil samples were selected for the experiment， including two types of sandy soil， one type of silty soil， and one type of clayey soil. Based on an L₁₂ （4×3²） orthogonal array， 12 frost heave test groups were designed. Temperature boundary conditions were controlled using a high-precision cold bath system. Frost heave displacement was monitored by an electro-hydraulic servo loading device. Varying water supply pressures were simulated by a constant-pressure water supply device. Parameters including frost heave amount， freezing depth， and frost heave rate were measured. The contributions of each factor were quantified using multiple linear regression and the geodetector method. The results showed that： （1） Soil type had the most significant effect on frost heave response. Silty soil， with high fine-particle content （51.55%）， showed an extreme frost heave rate （17.10%）， significantly higher than that of sandy soil and clayey soil. （2） Lower cold-end temperatures and higher water supply pressures both intensified frost heave. However， soil type played a more dominant regulatory role. For instance， clayey soil， due to its low permeability， was less affected by water supply pressure. （3） Geodetector analysis demonstrated that soil type most strongly explained the spatial variation of frost heave rate （q = 0.765）， significantly exceeding water supply pressure （q = 0.134） and cold-end temperature （q = 0.052）. （4） Multiple regression further confirmed the negative effect of soil type， indicating that frost heave suppression was enhanced when fine-particle content exceeded a certain threshold. This study reveals the dominant role of soil properties in frost heave and the synergistic effects of cold-end temperature and water supply pressure. It is recommended that engineering in cold regions prioritize soil selection optimization and integrate temperature and moisture conditions， to reduce frost heave risks. These findings provide a theoretical basis for research on frost heave mechanisms and engineering prevention and control.

The frost heave effect of frozen soil is a major challenge to the safety of engineering projects in the Qinghai-Xizang Plateau， and the fissure structures within frozen soil are the direct manifestation of this effect. In this study， unidirectional freezing tests were conducted on silty clay， and the development patterns of freezing-induced fissures in the specimens during unidirectional freezing were analyzed based on experimental data and images. The results showed that under different operating conditions， the height of cryogenic structures in the soil remained stable within the range of 6~11 cm. Under slow freezing conditions （with a cooling rate of 1 ℃⋅h^-1）， the height of the cryogenic structures associated with freezing fissures was proportional to the temperature gradient. The cold-end temperature and the cooling rate jointly influenced the formation of fissures in the frozen soil. At low cooling rates， a lower final freezing temperature resulted in a larger proportion of fissure area. In contrast， under rapid freezing conditions， extremely low temperatures may restrict fissure development or even induce fissure closure. During the freezing process， the width of the freezing fissures fluctuated between 1~2 mm， while fissure length was unstable with significant fluctuations. In saline specimens， an ice crust formed at the cold end during freezing， leading to increased salt concentration on both sides of the ice lenses. This reduced the thermal conductivity， impeded the advancement of the freezing front， and prevented the formation of distinct thin-layered cryogenic structures at the cold end and thick-layered structures at the warmest end， with only micro-thin-layered cryogenic structures observed. Fissures in saline soils develop at an inclined angle rather than strictly vertically， and most exhibit azimuths between 120° and 180°， trending closer to the horizontal direction and exhibiting pronounced directional characteristics. The findings provide experimentally derived insights into the geometrical characteristics of freezing-induced fissure structures， offering valuable references for engineering applications and safety assurance in the permafrost regions of the Qinghai-Xizang Plateau.

With the strategic focus of transportation infrastructure construction shifting towards cold regions in Northeast China， Northwest China， and the Qinghai-Xizang Plateau， the long-term safety and stability of rock engineering in these areas face severe challenges from freeze-thaw cycles. Sandstone， as a common rock type in geotechnical engineering， is particularly susceptible to damage induced by the phase change of pore water during freeze-thaw processes， leading to degradation of its mechanical properties. While existing research has provided considerable insights into the damage characteristics of rocks under low-cycle freeze-thaw conditions （typically less than 100 cycles）， studies on damage evolution under high-cycle freeze-thaw conditions （100 to 200 cycles）， which are critical for assessing the long-term performance of structures in cold regions， remain relatively limited. Furthermore， the combined effects of different sub-zero temperatures and high cycle numbers on the macro-meso damage correlation are not fully understood. This study aims to systematically investigate the damage characteristics and mechanical property evolution of sandstone subjected to high-cycle （up to 200 cycles） and high-frequency freeze-thaw actions under different sub-zero temperatures， thereby bridging the knowledge gap and providing experimental support for the construction and maintenance of rock engineering in cold regions. Sandstone specimens， sourced from a mountain tunnel in Southwest China， were processed into standard cylindrical samples （50 mm diameter × 100 mm height）. A total of 36 specimens were prepared and divided into 12 groups for freeze-thaw cycle testing， considering three freezing temperatures （-5 ℃， -10 ℃， -20 ℃） and four cycle numbers （50， 100， 150， 200）. Prior to testing， the freezing and thawing durations for a 24-hour cycle were calibrated for each temperature. The freeze-thaw cycles were conducted by submerging the saturated specimens in distilled water within open containers， freezing them at the target temperature for 12 hours， and then thawing at room temperature （20 ℃） for 12 hours. After completing the prescribed number of cycles， the mass loss rate of each specimen was calculated. Nuclear magnetic resonance （NMR） tests were performed on selected specimens （including an untreated reference specimen） to analyze the evolution of pore structure， including T₂ spectrum， pore size distribution， porosity， and NMR imaging. Subsequently， triaxial compression tests were conducted on all freeze-thaw cycled specimens and three untreated reference specimens under confining pressures of 5 MPa， 10 MPa， and 15 MPa using a GCTS triaxial compression testing system. The stress-strain curves， peak strength， and elastic modulus were obtained to evaluate the degradation of mechanical properties. The key results and findings were as follows： （1） Pore structure evolution： NMR analysis revealed that the damage and expansion of the sandstone’s pore structure under freeze-thaw actions were primarily dominated by the development of mesopores （1~100 μm）， with negligible development of micropores and small pores. A significant and abrupt expansion in mesopore volume was observed when the number of freeze-thaw cycles reached 150， showing a 60% increase compared to the volume after 100 cycles. NMR imaging visually confirmed that as the number of cycles increased and the temperature decreased， the number of pores increased and connectivity was enhanced. （2） Macroscopic damage indicators： the porosity of sandstone increased from an initial 12.6% to 14.2% after 150 freeze-thaw cycles， accompanied by a mass loss rate of 1.2%， indicating pronounced freeze-thaw damage. The mass loss rate exhibited a sharp increase， particularly when the temperature dropped below -10 ℃ and the cycle number exceeded 100. Under the same temperature condition， the mass loss rate after 200 cycles was 20 times greater than that after 50 cycles， highlighting a significant cumulative damage effect under long-term cycling and lower temperatures. （3） Mechanical property degradation： the stress-strain curves of sandstone after freeze-thaw cycles exhibited marked nonlinear characteristics. Both the peak strength and elastic modulus exhibited nonlinear decay with decreasing freezing temperature and increasing number of freeze-thaw cycles， with a sudden decrease around 100 cycles， after which the degradation rate tended to stabilize. The mechanical properties also showed a strong dependence on confining pressure， with higher confining pressures mitigating the strength and stiffness loss to some extent. （4）. Macro-meso correlation： strong correlations were established between microscopic NMR parameters （porosity， spectral area） and macroscopic damage indicators （mass loss rate， peak strength， elastic modulus）. As NMR porosity and spectral area increased， indicating microstructural damage accumulation， the mass loss rate increased， and the peak strength and elastic modulus decreased correspondingly. This confirmed that the continuous development of internal micro-pores was the fundamental cause of the macroscopic performance deterioration of sandstone under freeze-thaw cycles. This study provides a comprehensive understanding of the damage evolution in sandstone under high-cycle freeze-thaw conditions and clarifies the combined influence of cycle number and temperature. The identified critical thresholds （e.g.， significant changes around 100~150 cycles， heightened sensitivity below -10 ℃） and the established correlations between macro-meso damage indicators are crucial for predicting the long-term durability and performance of sandstone structures in cold regions. The findings offer valuable insights and experimental data for the design， construction， operation， and maintenance of rock engineering projects， such as tunnels and slopes， in extremely cold environments， contributing to enhanced safety and extended service life.

Freeze-thaw erosion represents a critical geomorphic process in the permafrost regions of the Qinghai-Xizang Plateau， where pronounced near-surface hydrothermal gradients and heterogeneous microenvironments jointly regulate soil thermal regimes， freeze-thaw cycling， and erosion intensity. However， existing regional-scale assessments， typically based on kilometer-scale datasets， fail to adequately capture microenvironmental variations relevant to engineering and ecological stability. Long-term observations from September 2016 to August 2024 were collected at nine sites in the source area of the Yangtze River， representing different local environmental conditions and covering a range of soil water contents， vegetation cover， and slope aspect. Based on this， the dynamic variations of air temperature， near-surface ground temperature， and freeze-thaw cycles were investigated， and the spatial heterogeneity and temporal evolution of freeze-thaw erosion intensity under different local environmental conditions were analyzed. The results showed that： （1） Air temperature exhibited a slight warming trend （0.08 ℃⋅a^-1）， while near-surface ground temperature increased more markedly （0.16 ℃⋅a^-1）， with freeze-thaw cycles occurring more frequently under higher soil water content and denser vegetation cover. （2） Freeze-thaw erosion intensity showed an overall weakening trend， shifting from severe erosion toward moderate erosion， and was strongly regulated by local environmental conditions. Specifically， high soil water content enhanced erosion， while high vegetation cover mitigated it. The most intense erosion occurred on sandy surfaces and south-facing slopes. （3） Soil erodibility and vegetation cover were identified as the key factors controlling near-surface freeze-thaw erosion intensity， exerting stronger effects than slope aspect. These findings highlight the regulating role of local environmental conditions in freeze-thaw processes and erosion intensity， providing scientific support for freeze-thaw erosion risk assessment and ecological conservation on the Qinghai-Xizang Plateau.

The Qinghai-Xizang Plateau， as one of the regions with the highest elevation and the most complex geological structures in the world， possesses unique geotechnical characteristics shaped by its distinctive geological environment and extreme climatic conditions. In this region， mudstone—a prevalent weak rock type—exhibits significant time-dependent and environment-sensitive mechanical behavior under the combined effects of intense physical and chemical weathering. This leads to frequent highway engineering disasters with complex failure mechanisms. Focusing on the strength degradation mechanisms of mudstone under weathering， this study aims to establish a quantitative response relationship from microstructural deterioration to macro-scale engineering disasters， thereby providing theoretical foundations and technical support for the stability assessment and disaster prevention of highway slopes in plateau regions. The study site was selected from the Kasuhu Bridge section of the under-construction G0611 Zhangye-Wenchuan National Expressway on the Qinghai-Xizang Plateau. This area had widely distributed weathered mudstone layers with distinct weathering gradients， making it highly representative. To comprehensively reveal the evolution of its mechanical properties， systematic laboratory testing and microscopic observation were conducted. In terms of macro-mechanics， conventional triaxial compression tests （CU） were performed to obtain stress-strain relationships， peak strength， and residual strength characteristics of mudstones at different weathering degrees under different confining pressures. Creep tests were conducted to investigate deformation progression and failure thresholds under long-term loading. For microstructural analysis， scanning electron microscopy （SEM） was employed to observe intergranular bonding states and microfracture development. High-resolution microscopy was used for pore structure quantification， while particle size distribution analysis was conducted to quantitatively describe compositional changes during weathering. The test results showed that as weathering intensified， the clay fraction （particles < 0.005 mm） in mudstone increased significantly. Fully weathered mudstone may contain over 40% clay， with marked deterioration in particle size distribution. The gradual depletion of coarse grains and enrichment of fine grains weakened intergranular interlocking and reduced structural cohesion， thereby affecting overall mechanical response. Further mechanical tests revealed that the strength properties of mudstone exhibited exponential decay with increasing degree of weathering. Unweathered mudstone had high cohesion and internal friction angle， exhibiting distinct peak strength. As weathering progressed， particularly upon reaching strongly weathered states， mudstone strength drastically decreased. Fully weathered mudstone almost lost effective bearing capacity， becoming prone to plastic flow and progressive failure. Creep test results indicated that under low stress levels， all weathered mudstones exhibited steady-state creep behavior， with deformation rates gradually stabilizing. However， under high stress conditions， mudstone entered an accelerated creep phase， ultimately leading to failure. This mechanical behavior was highly consistent with the shallow surface creep deformation commonly observed in field slopes， indicating that the rheological properties of weathered mudstone were a key factor controlling long-term slope stability. Building upon the macro-micro analysis， a comprehensive degradation factor D_m was defined using “clay content c” and “fracture group number n” as key indicators to quantify the degree of weathering damage in mudstone. This factor effectively captured the synergistic degradation effect of clay enrichment and fracture expansion on rock mechanical properties under shallow， low-confining pressure conditions， thereby establishing a mudstone damage evolution equation applicable to the unique geological context of high-altitude plateaus. This equation enabled dynamic assessment and prediction of the stable state of mudstone during weathering. More importantly， by tracing the chain process from particle size distribution reconstruction to microstructural damage and macro-scale mechanical property degradation， this study revealed the cross-scale catastrophic mechanisms of weathered mudstone and clarified the coupling relationship between shallow-surface catastrophic events and deep-seated failure. This study not only systematically elucidates the evolution patterns of strength and deformation characteristics in mudstones with different degrees of weathering but also provides a theoretical framework and practical guidance for slope design， disaster early warning， and prevention in highway engineering on the Qinghai-Xizang Plateau. The proposed degradation factor and damage model demonstrate excellent engineering applicability， offering valuable insights for rock mass stability assessment under similar geological conditions and holding significant importance for advancing disaster prevention and mitigation technologies in plateau transportation infrastructure development.

High-altitude highways are highly sensitive to environmental changes， and the evolution of their surface environments directly affects the long-term stability and safe operation of engineering projects. Under the background of intensified global warming and regional humidification， the fluctuation amplitudes and frequencies of environmental factors along highways in plateau regions have continuously increased， posing substantial challenges to the resilience and sustainable operation of high-altitude infrastructure systems. The Qinghai-Xizang Plateau， known as the “Third Pole” of the Earth， is characterized by its fragile ecosystem and complex frozen soil distribution， demonstrating significant responsiveness to climate change. The Lazi-Yecheng section of G219 is situated on the western boundary of the Qinghai-Xizang Plateau， featuring diverse landform types and significant elevation differences. It has the characteristics of permafrost， arid， and semi-arid regions， making it a typical representative area. Investigating the patterns of the surface environmental changes in this area is crucial for understanding plateau ecological processes and ensuring the safe operation of road engineering. Based on multi-source satellite remote sensing data， this study systematically analyzed the spatiotemporal variation characteristics of surface environmental factors within a 10 km range on both sides of the Lazi-Yecheng section of G219 by using data from surface water distribution， land surface temperature （LST）， normalized difference vegetation index （NDVI）， and temperature vegetation dryness index （TVDI） from 2000 to 2024. Through time-series statistics， spatial zoning， and trend and variability analyses， the variation patterns and the interrelationships among the major environmental factors were revealed. The results showed that the surface environment in the study area exhibited distinct spatial differentiation while maintaining overall temporal stability. Surface water bodies with an area less than 0.1 hm² accounted for more than 50% of the total number of water bodies， indicating a hydrological pattern dominated by small lakes and ponds. The total area of surface water increased from 648.3 km² in 2000 to 701.8 km² in 2020， with an average annual growth rate of approximately 1.46 km². LST was mainly distributed between 15~25 ℃， showing a slightly decreasing trend across most regions. The NDVI remained generally low， with values below 0.2 in most areas， indicating sparse vegetation cover in the study area. The TVDI exhibited a generally high level， and the proportion of the area with TVDI greater than 0.8 increased year by year， showing a trend of intensifying drought. Spatially， LST showed a gradual decrease with increasing distance from the highway， while vegetation cover showed minimal variation. Drought severity weakened with increasing distance， suggesting relatively poorer moisture conditions in the near-road areas. The results of trend and variability analyses demonstrated that， despite interannual fluctuations， land surface temperature， vegetation conditions， and drought severity remained generally stable overall. Comprehensive analysis showed that surface water played a primary role in regulating regional thermal and moisture conditions， while topographic factors such as elevation， slope， and aspect exerted significant influences on the spatial distribution of environmental factors. In summary， this study systematically reveals the spatial patterns and long-term evolution characteristics of surface environmental factors along the Lazi-Yecheng section of G219. The findings clarify the coupling relationships among hydrological， thermal， and ecological processes in plateau regions， offering a scientific basis for environmental monitoring， risk assessment， and sustainable management of high-altitude highways.

The Qinghai-Xizang Plateau （QZP）， known as the “Third Pole”， contains approximately 1.06×10⁶ km² of permafrost， accounting for over 70% of global high-altitude permafrost. Under accelerated climate warming ［0.3~0.4 ℃·（10a）^-1］， permafrost degradation has intensified dramatically， with retrogressive thaw slumps （RTSs） emerging as the most dramatic geomorphological response. These features， characterized by steep headwalls， extensive mudflows， and elevated frontal lobes， generate intensive freeze-thaw erosion processes that fundamentally alter regional hydrology， sediment transport， and ecosystem functioning. Despite their profound environmental impacts， the quantitative mechanisms linking ground ice content to freeze-thaw erosion intensity remain inadequately understood， significantly limiting the ability to predict environmental consequences and implement effective mitigation strategies.This study aims to establish quantitative relationships between ground ice content and freeze-thaw erosion processes through systematic laboratory experiments. Comprehensive thawing tests were conducted on 56 remolded ice-rich permafrost specimens from the Beiluhe River Basin， one of the most RTS-dense regions on the QZP. The experimental design employed dual complementary systems： a precision USTX-2000 triaxial apparatus for mechanical parameter monitoring and a modified vacuum filtration system for meltwater collection and analysis. Specimens were prepared with seven ice content levels （20%， 25%， 30%， 35%， 40%， 45%， and 50%） using standard compaction methods at -8 ℃， with dry density controlled at （1.35±0.02） g⋅cm^-3. During natural thawing at （18±1） ℃， thaw settlement displacement， drainage volume， and pore water pressure evolution were continuously monitored at 10-second intervals. Meltwater quality parameters including total suspended solids （TSS）， total dissolved solids （TDS）， and electrical conductivity were determined following suction filtration tests and water quality meters.The results revealed three distinct thawing phases， each showing markedly different erosion characteristics. During the accelerated settlement phase， 84.4% of total settlement occurred due to rapid failure of ice cement， creating extensive drainage networks. The thaw settlement coefficient demonstrated a strong quadratic relationship with ice content. Critical threshold behavior emerged at 35% ice content， above which all parameters showed nonlinear acceleration. As ice content increased from 20% to 50%， thaw settlement coefficient increased by 47.1%， cumulative drainage volume surged by 547%， and TSS concentration rose from 375 mg⋅kg^-1 to 866 mg⋅kg^-1 （131% increase）. Pore pressure fluctuations （0~3 kPa）， combined with strength loss， created conditions for catastrophic soil failure， directly explaining field observations of sudden headwall collapse in RTSs.The underlying mechanisms involved three synergistic processes driving freeze-thaw erosion. First， elevated pore pressures from ice-to-water phase transitions destabilized soil structure while ice cement loss eliminated particle bonding. Second， dynamic permeability evolution controlled fine particle transport， with initial high-permeability drainage networks facilitating rapid sediment mobilization before channel collapse limited flow. Third， enhanced hydrodynamic forces at higher ice contents exponentially increased erosion capacity， explaining the disproportionate environmental impacts of ice-rich permafrost thaw.These findings have profound implications for understanding and predicting RTS impacts on the QZP. The established relationships between ice content and erosion intensity provide quantitative tools for evaluating freeze-thaw erosion potential under the plateau’s diverse permafrost conditions. This is particularly critical given that 2 669 active RTSs pose threats to regional infrastructure， including the Qinghai-Xizang Railway， where some features are within 50 m of the track. The observed threshold effects indicate that permafrost with more than 35% ice content poses disproportionate risks， necessitating targeted monitoring and intervention. Furthermore， these results explain the observed changes in lake color and elevated sediment loads in permafrost watersheds， confirming that ground ice thaw is a primary driver of regional water quality degradation. This study advances the fundamental understanding of periglacial processes while providing essential data for climate adaptation strategies， infrastructure risk assessment， and ecosystem management in the rapidly warming Third Pole region of the Earth.

At the Beiting Ancient City earthen site in Xinjiang， abundant winter snowfall followed by rapid spring warming frequently induces snowmelt scouring and top-edge sliding. This study aims to elucidate the hazard chain of “abrupt melt → infiltration → weakening → sliding，” and to propose and validate a replicable full-scale， in-situ simulation and multi-parameter monitoring approach that provides parameterized evidence for mechanism identification and preventive conservation. An outdoor test area designed to replicate the original site in climate， radiation， and wind regime was utilized. A full-scale rammed-earth test wall was constructed， considering frost penetration depth， wall geometry， and sunny-shaded aspect. An integrated observation system—comprising an automatic weather station， time-lapse photography， and multisource UAV imagery—was combined with in-wall sensors for temperature， volumetric water content， pore-solution electrical conductivity， and strain deployed at multiple positions and depths. Continuous winter monitoring was conducted. The coupling between external forces and internal responses was characterized， and the locations and development paths of potential failure surfaces were identified by tracking the spatiotemporal evolution of the 0 ℃ isotherm， moisture belts， and high-conductivity zones， as well as by analyzing sectional and plan-view isopleths during representative periods. The results indicated a distinct seasonal pattern. Persistent snow accumulation occurred under low-temperature and high-humidity freezing conditions， followed by rapid spring warming that triggered accelerated melt infiltration—the key environmental driver of erosion and failure. The sun-facing surface exhibited large diurnal temperature amplitudes and frequent zero crossings， producing high-frequency freeze-thaw cycles and near-surface structural relaxation. In contrast， the shaded face remained stably frozen and thawed slowly with limited energy and moisture supply， leading to a pronounced directional tendency of sliding toward the shaded side. During snowmelt， water-salt migration was active in the sun-facing shallow layer， and cyclic fluctuations in water content and electrical conductivity coincided with rapid declines in surface strength. A mid-value moisture belt formed along the crest and migrated from sunny to shaded areas， while a progressively stabilized high-conductivity zone developed beneath the shaded top edge. These fields showed clear spatiotemporal co-location with the continuous 0 ℃ isotherm. In the mid-to-deep wall， thermo-hydraulic responses were low-amplitude， lagged， and slow， reflecting thermal inertia and hydraulic buffering that attenuated and delayed external perturbations. Under rapid melt， disturbances first concentrated within a shallow softening zone of around 0~7 cm. Potential slip planes preferentially propagated along rammed-layer interfaces and through shallow high-moisture/high-salinity weakened zones， resulting in small-scale yet time-sensitive directional sliding. Based on long-sequence joint indicators， a composite early-warning signal for entry into a rapid-instability window was proposed： two consecutive days with average air temperature above 0 ℃， an increase over 25% in shallow water content relative to baseline， and a concurrent rise in pore-solution electrical conductivity， with these signals co-located near the shaded crest along with the 0 ℃ isotherm and moisture/conductivity belts. The established full-scale in-situ framework enables synchronous acquisition of surface morphological evolution and internal thermo-hydro-saline mechanical responses， empirically revealing the ordered pathway “aspect-controlled energy input → crest-parallel water routing → salt reallocation → shallow interfacial weakening → shaded-side sliding.” The approach provides a transferable technical basis and quantitative indicators for risk identification， graded early warning， and mitigation design， and informs the parameterization of engineering measures such as crest snow interception and meltwater drainage， near-surface thermal regulation on shaded faces， and interface-oriented targeted reinforcement.

Frost resistance is a key factor affecting the structural stability and service life of subgrades in seasonally frozen soil regions. To systematically investigate the frost resistance of fly ash-based subgrade materials under freeze-thaw conditions， this study conducted freeze-thaw cycle tests and mechanical performance evaluations under different curing ages， compaction degrees， and cycle numbers. The evolution patterns of unconfined compressive strength （UCS）， splitting tensile strength， and microstructural characteristics were analyzed. Material degradation under freeze-thaw action was quantitatively assessed using indicators such as mass loss rate and strength retention rate. Microstructural evolution was further analyzed through X-ray diffraction （XRD） and scanning electron microscopy （SEM）， revealing the continuous formation of C-（A）-S-H hydration products and their microstructural mechanisms in pore filling， structural densification， and strength enhancement. The results indicated that the mechanical properties of fly ash significantly improved with curing age and compaction degree， while freeze-thaw cycles induced structural damage leading to progressive strength degradation， characterized by three stages： rapid deterioration， gradual decline， and eventual stabilization. The formation of hydration-induced bridging structures enhanced material compactness and frost resistance， which was macroscopically manifested in the progressive evolution of failure patterns from splitting to “X”-shaped and shear types. Furthermore， a multivariate nonlinear regression model was developed to accurately predict UCS under different conditions. These findings provide a theoretical basis and technical support for the design optimization and performance evaluation of fly ash-based subgrade materials in seasonally frozen soil regions of northwest China.

Tunnels constructed in seasonally frozen regions are particularly vulnerable to frost-related damages during the cold season， severely compromising their structural integrity and operational safety. These issues include freeze-thaw cracking of the lining concrete， freezing failure of the drainage system， icicle formation on lining surfaces， road icing， and even tunnel blockages due to excessive ice accumulation. These problems not only compromise the safe operation of tunnels but also lead to escalating maintenance costs， making the mitigation of frost damage a critical challenge. Among the various factors influencing frost damage， the maximum frozen depth of the tunnel surrounding rock is of paramount importance. This parameter reflects the radial temperature field within the tunnel structure. Consequently， accurately predicting the maximum frozen depth is essential for evaluating frost heave forces that may deform tunnel linings， determining the optimal burial depth of drainage systems to prevent freezing failures， and selecting appropriate frost-resistant materials to enhance durability. Thus， a reasonable and accurate prediction of the maximum frozen depth of the tunnel surrounding rock is crucial for guiding the anti-freezing design and construction of tunnels in seasonally frozen regions. Despite the existence of multiple approximate analytical methods for estimating the maximum frozen depth， conventional approaches often suffer from limited accuracy and practicality. As a result， accurately and efficiently determining the maximum frozen depth of the tunnel surrounding rock in seasonally frozen regions remains a significant challenge. To overcome these limitations， this study proposed a novel approximate analytical method for determining the maximum frozen depth of tunnel surrounding rock in seasonally frozen regions. Firstly， based on the theory of negative accumulated temperature， the inner surface temperature of the tunnel structure was equated to the sustained effect of an equivalent negative temperature over a specified period. Subsequently， under the assumption that the latent heat of the ice-water phase transition was significantly greater than the sensible heat， the heat transfer differential equation for the tunnel structure was approximately derived， thereby validating the application of the quasi-steady state theory. By applying the quasi-steady state theory， the radial temperature distribution of tunnels in seasonally frozen regions under this condition was incorporated into the heat flux balance equation at the freezing front. Through a variable separation process， an integral solution was derived to establish the relationship between the freezing development time and the freezing front radius under the prerequisite condition of a fixed ratio of the freezing front radius to the outer boundary radius of the tunnel structure. This approach yielded an approximate analytical solution for the maximum frozen depth of the tunnel surrounding rock. An approximate analytical method was then proposed to calculate the maximum frozen depth of the tunnel surrounding rock in seasonally frozen regions for tunnels with insulation layers. For case studies， the frozen depths of the surrounding rock in a high-latitude tunnel and a high-altitude tunnel in China were obtained through field measurements. The maximum frozen depths of the tunnel surrounding rock were subsequently calculated using both Stefan’s formula with the equal-thickness thermal resistance conversion algorithm and the method proposed in this study， and the results were compared with the field-measured data. The results showed that compared with the widely used Stefan’s formula， the approximate analytical solution proposed in this study aligned more closely with the measured data， thereby validating the rationality and effectiveness of the proposed method. The findings of this study have significant implications for the anti-freezing design and maintenance strategies of tunnel engineering in seasonally frozen regions.

Mountain permafrost refers to the frozen soil/rock layers in high-altitude mountainous areas that have remained frozen for more than two years. It is characterized by fragmented distribution， thin layers， sensitivity to climate， and strong influence from local environmental conditions. With the gradual increase in engineering construction in permafrost regions， engineering and climate effects have led to regional changes in permafrost， including permafrost degradation， a lowering of the upper limit of permafrost beneath engineering structures， and an increase in permafrost temperature. All these will cause vegetation desertification and increased risk of freeze-thaw disasters， thereby resulting in changes in engineering stability and a decrease in engineering service life. Ultimately， it leads to the decline in the sustainability of the permafrost environment. Evaluating the permafrost environment under the influence of engineering is crucial for ensuring the safe operation and use of linear projects in permafrost regions. To address key challenges in the current research on the environmental assessment of mountainous permafrost， such as the uncertainty of evaluation indicators and the randomness and fuzziness of grading boundary information， a comprehensive evaluation mathematical model of permafrost environment based on game theory and extension cloud theory was proposed. Starting from the intrinsic characteristics of mountainous permafrost， such as thermal stability， erosion sensitivity， and ecological vulnerability， a multi-level evaluation indicator system was systematically constructed， including 3 criterion layers and 12 specific indicators. The evaluation was conducted from the perspectives of seasonal melting depth， annual average ground temperature， thickness of permafrost， terrain slope， rock stability coefficient， annual temperature range， annual average precipitation， relative diversity of wetland species， vegetation coverage， wastewater treatment rate， and land reclamation rate. In terms of weight determination， the game theory-based comprehensive weighting method was employed. Through the Nash equilibrium principle， the subjective weights determined by the decision-making trial and evaluation laboratory （DEMATEL） method and the objective weights calculated by the entropy method were optimally combined， effectively solving the problem of unreasonable distribution of subjective and objective weights in traditional evaluations. Regarding the evaluation method， the unique advantages of the extension cloud model in dealing with uncertain problems and fuzzy boundary problems were fully utilized. By calculating the cloud membership function between the object element to be evaluated and the standard cloud， the comprehensive evaluation vector was obtained， and the final evaluation grade was determined based on the principle of maximum membership degree. To verify the reliability and validity of the model， a typical mountainous permafrost area along the Chaida’er-Muli Railway （DK101+500~DK102+400） section in the Qilian Mountains was selected as a case study. The evaluation results showed that the expected value Ex of the permafrost environmental quality in this area was 3.811， corresponding to the Grade IV （relatively good） state. The indicators of annual average ground temperature （C₁₂）， volumetric ice content （C₁₃）， and permafrost thickness （C₁₄） were evaluated as Grade III （medium） under the influence of disturbances in railway construction and operation and incomplete compensation of thermal imbalance by engineering repair measures， indicating signs of degradation in the permafrost in this area. This evaluation result was consistent with the actual situation. The game theory-based comprehensive weighting method not only reflected the subjective will of decision-makers， but also effectively avoided subjective randomness， thereby obtaining a more realistic and reasonable weight distribution. This approach could effectively solve the fuzziness and randomness of the grading boundaries in the environmental evaluation of permafrost. The research findings not only provide new ideas and methods for the environmental evaluation of mountainous permafrost， but also offer a scientific basis for decision-making in engineering construction and ecological protection in cold regions. In addition， the grading evaluation system constructed in this study is mainly applicable to the environmental evaluation of mountainous permafrost affected by engineering disturbances in the Qilian Mountains. When extending this model to other permafrost regions， it is necessary to reconstruct the grading system in combination with regional characteristics and adjust the model parameters accordingly. Future studies can attempt to adjust the selected evaluation indicators for different permafrost regions and consider the dynamics of external environmental condition disturbances during the selection process， thereby forming a more complete evaluation system.

The Ranwu-Tongmai section of the Parlung Zangbo serves as a critical strategic corridor connecting Xizang with inland China. This area is characterized by complex landforms， deeply incised valleys， widely distributed maritime glaciers and seasonally frozen soil， and high sensitivity to climate change. In the context of global climate change， the region experiences severe freeze-thaw erosion and frequent landslide hazards， which often damage roads， disrupt traffic， and cause significant losses. With China’s continuously increasing investment in Xizang’s infrastructure development， the exchange of personnel and goods through this corridor has undergone rapid growth. Consequently， there is an urgent need to systematically identify potential landslide hazards and implement specifically targeted risk prevention measures to effectively reduce regional landslide risks. This study utilized multi-temporal sub-meter high-resolution remote sensing images archived in Google Earth from 2015 to 2023. Based on interpretation indicators such as color variations， drainage patterns， landform characteristics， and vegetation changes， landslides were identified and delineated through visual interpretation. The final landslide inventory was determined through a field investigation conducted in April 2024， with remote sensing interpretations rigorously validated via on-site inspections to ensure accuracy. Twelve landslide-related characteristic variables， including the freeze-thaw intensity index （FTI） and summer rainfall， were selected. Correlation analysis was conducted to remove variables with high collinearity， and the remaining variables were used to construct the dataset. A 90 m grid cell was selected as the evaluation unit. The landslides involved 3 271 grid cells in total， which were set as positive samples with a label value of 1. An equal number of non-landslide grid cells were randomly selected within the study area as negative samples， resulting in a total of 6 542 samples. Seventy percent of both positive and negative samples were randomly allocated to construct the training dataset， while the remaining 30% were reserved exclusively for the testing dataset. Both training and testing datasets maintained an identical 1∶1 ratio of positive to negative samples. Three machine learning models—random forest （RF）， support vector machine （SVM）， and multilayer perceptron （MLP）—were trained and validated using these datasets to assess regional landslide susceptibility. The importance of each characteristic variable obtained from the RF algorithm and correlation analysis was used to analyze the main controlling factors. The results showed that the study area had abundant rainfall and contained 296 landslides in total， predominantly small to medium in size. These landslides were mainly distributed within the elevation range of 3 500~4 500 m， on sunny slopes （67.5°~202.5°）， and in regions with FTI values greater than 0.6. All three algorithms—RF， SVM， and MLP—demonstrated robust landslide prediction performance. Among these， RF achieved superior performance indicators： Accuracy （0.82）， Recall （0.85）， F1-score （0.83）， Jaccard Index （0.70）， and AUC （0.90）， outperforming SVM （AUC = 0.87） and MLP （AUC = 0.86）. Feature importance analysis revealed four main controlling factors on landslide susceptibility： slope aspect （25.63%）， elevation （18.82%）， summer rainfall （14.05%）， and freeze-thaw intensity index （9.83%）. Among them， elevation and slope aspect determined the material basis and energy conditions for slope landslides， while rainfall and freeze-thaw cycles were external triggering factors for landslide development. The susceptibility assessment results from the RF， SVM， and MLP models showed that the distribution of high and very high susceptibility zones was basically consistent， and the areas of very high susceptibility zones were nearly identical. The study area was divided into two target zones for landslide hazard mitigation. For the Guxiang-Tongmai section， special attention should be given to the impact of rainfall， with landslide mitigation focusing on improving water-catching structures and drainage systems. Meanwhile， for the Ranwu-Yupu section， landslide mitigation should prioritize the effects of freeze-thaw cycles， implementing measures such as insulation and seepage prevention to enhance the frost resistance of the geotechnical materials. These findings not only contribute to the risk prevention and control of potential landslide hazards along the Ranwu-Tongmai section of the Parlung Zangbo in the high-altitude alpine mountainous areas， but also provide a reference for subsequent engineering decision-making in disaster mitigation and prevention for regional landslide hazards.

Environmental factors such as extreme climate conditions and harsh topography significantly affect the planning， design， construction， operation， and maintenance of highway engineering in cold regions. A comprehensive understanding of the regional distribution patterns of environmental factors such as climate， topography， and frozen soil types， followed by targeted research on the natural zoning of highway engineering in cold regions， can provide a basis for determining the technical measures of construction and maintenance as well as design parameters for highway engineering in cold regions. This study took Gansu Province， located at the intersection zone of the Qinghai-Xizang Plateau， Loess Plateau， and Inner Mongolian Plateau， as an example. Based on the natural zoning principles （including the principle of combining comprehensiveness with dominance， principle of hierarchy， principle of combining qualitative indicators with quantitative indicators， principle of practicability， principle of relative uniformity， principle of intra-regional similarity and inter-regional difference and principle of regional genesis） and methods （such as the sequential division method， merging method， GIS multi-source data layer overlapping method， and three-segment nomenclature method） for highway engineering， a two-level natural zoning indicator system for highway engineering was established. Firstly， the study area was divided into five first-level zones based on the mean annual air temperature isotherm of -2 ℃， mean monthly air temperature isotherm of 0 ℃ in January， and frozen soil types in Gansu Province. Furthermore， the five first-level zones were divided into twelve second-level zones based on the difficulty index of highway engineering， incorporating environmental factors including elevation， slope， relief amplitude， surface cutting depth， surface cutting density， geotechnical characteristics， and land cover. Among these twelve second-level zones， three were classified as areas with moderate difficulty in highway engineering construction， five as areas with high construction difficulty， and four as areas with extremely high construction difficulty. Meanwhile， the specific construction difficulty of highway engineering in each second-level zone can be reflected by differences in frozen soil types and environmental factors. The natural zoning results for highway engineering in Gansu Province showed that the highway engineering in the short-duration frozen soil region of the Qinling Mountains （Zone Ⅴ） was mainly affected by complex topographic conditions. Challenges of highway engineering construction such as the difficulty in compacting high-fill subgrades， severe water damage risk to highways， and unfavorable geological conditions need to be overcome. Therefore， it was recommended to employ more bridge and tunnel projects in Zone Ⅴ. In the seasonally frozen soil region on the plateau （Zone Ⅰ） and the seasonally frozen soil region in the Sugan Lake Basin （Zone Ⅱ）， the spatial distribution patterns of highway damage varied due to their different geographical units. Meanwhile， as the difficulty index of highway engineering increased within the same geographical unit， the occurrence frequencies of highway frost damage became higher. In mountain frozen soil region of the A-erh-chin Mountains-Qilian Mountains （Zone Ⅲ） and sporadic frozen soil region of the Aemye Ma-chhen Range （Zone Ⅳ）， due to the extreme climate conditions and harsh topography in these regions， highways suffered from severe damage. Highways within these zones were prone to severe pavement icing during cold seasons and slope failures and landslides during the wet season. These types of highway damage significantly disrupt traffic operations. Therefore， route selection is critical for the construction， operation， and maintenance of highway engineering in these two first-level zones. Conducting natural zoning of highway engineering in cold regions is conducive to reducing construction， operation， and maintenance costs， improving highway quality， and promoting cost-efficiency and high-quality development of highway engineering.

Dike seepage hazards in cold-region channels during the ice flood season occur frequently， characterized by prominent abruptness， cascading effects， and uncertainty. Predicting their evolution process is a key technology for preventing dike breaches. This study reviews the research progress both domestically and internationally on the driving factors， evolution mechanisms， and prediction methods of dike seepage hazards during the ice flood season. It points out that while current studies have made progress in areas such as the patterns of sin-gle-factor effects and numerical simulation， the evolution mechanisms driven by multi-factor coupling remain insufficiently understood. Existing prediction methods have limitations in simultaneously improving accuracy and timeliness. On this basis， two key scientific and technical challenges are summarized： the evolution mechanism of dike seepage hazards during the ice flood season driven by multi-factor coupling in complex environments， and the pre-diction method based on physics-data dual drive. By integrating artificial intelligence tech-nologies， this study proposes new interdisciplinary research ideas， highlighting that integrating physical mechanisms with data-driven approaches is the primary technical pathway for precise and rapid prediction， and establishing a “Four-Preventions” linkage model as a new paradigm for scientific prevention and control. The research findings can provide a scientific basis for im-proving the theoretical and methodological system for preventing and mitigating dike hazard disasters during the ice flood season.

Polycyclic aromatic hydrocarbons （PAHs） are a critical class of persistent organic pollutants （POPs） that can be transported over long distances by atmospheric circulation and are present in low-temperature environments such as the North and South Poles and the Qinghai-Xizang Plateau. Global warming accelerates the melting of snow and ice， leading to the exposure of PAHs and affecting human health. Microorganisms serve as the primary degraders of PAHs， with extensive research reported on their degradation. However， studies on PAHs-degrading microorganisms in low-temperature environments have mainly emerged in the past two decades. In this study， the diversity of PAHs-degrading microorganisms in low-temperature environments， their degradation mechanisms， and the responses of microbial biodegradation to low temperatures are systematically summarized and reviewed. Additionally， further research directions and prospects for microbial degradation of PAHs in low-temperature environments are proposed， providing theoretical support for the source prevention and ecological remediation of PAHs contamination in low-temperature environments in the future.

Functioning as a pivotal link between the atmosphere and the pedosphere， vegetation is essential for regulating surface energy exchanges， maintaining ecosystem processes， and facilitating global biogeochemical cycling. Peak vegetation growth serves as a critical indicator of how terrestrial ecosystems respond to climate change， directly influencing their carbon sequestration capacity and the seasonal dynamics of atmospheric CO₂ concentrations. Permafrost regions in the Northern Hemisphere have been undergoing significant warming， leading to deepening of the active layer and shortened soil freezing durations. These thawing processes indirectly regulate vegetation function and carbon sink potential by altering subsurface thermal and hydrological conditions. However， a comprehensive quantitative assessment of the combined effects of climate drivers and permafrost thaw on peak vegetation growth remains limited in these regions. In this study， multi-source remote sensing products， reanalysis meteorological data， and permafrost parameters， including active layer thickness （ALT） and the start of spring soil thaw （SOT） were integrated to systematically investigate the spatiotemporal dynamics of peak vegetation growth and its response mechanisms to climate and permafrost thaw across the permafrost regions of the Northern Hemisphere from 2001 to 2018. The results revealed widespread and significant upward trends in both peak normalized difference vegetation index （NDVI_max） and peak near-infrared reflectance of terrestrial vegetation （NIRv_max）. The annual average increase rates were 0.11% for NDVI_max and 0.05% for NIRv_max. Notably， 75% of the study region exhibited synchronous increases in both peak indicators， indicating a widespread enhancement of peak ecosystem function in permafrost-affected landscapes. Using an explainable machine learning analysis， it was found that temperature was the primary driver of changes in both NDVI_max （60%） and NIRv_max （54%）， followed by moisture （15% and 17%） and radiation （11% and 13%）. In contrast， the influence of permafrost thaw was relatively weaker （8% for NDVI_max and 12% for NIRv_max）， as was that of atmospheric CO₂ concentrations （4% and 6%）. Across different vegetation types， peak growth indicators were generally positively correlated with temperature， precipitation， and CO₂ concentrations， but negatively correlated with radiation. However， in broadleaf forests， mixed forests， and shrublands， NIRvₘₐₓ exhibited a negative correlation with precipitation and a positive correlation with radiation， contrary to the overall trends. The ALT in permafrost regions exhibited an overall increasing trend during the study period， with 77% of the area showing thickening and a regional mean rate of 0.53 cm per year. The SOT generally advanced， with 65% of the area experiencing earlier thawing and a regional average advancement rate of 0.11 days per year. Furthermore， piecewise structural equation modeling was applied to identify the pathways through which permafrost thaw affected vegetation peak growth. The results showed that permafrost thaw primarily exerted indirect effects by altering soil thermal and moisture regimes. Earlier SOT had a positive effect on peak vegetation growth. In contrast， the impact of ALT varied by permafrost type. It promoted peak growth in continuous and discontinuous permafrost regions， but suppressed growth in sporadic and isolated permafrost zones. The negative effects were attributed to enhanced soil water percolation and drainage， which led to root-zone moisture deficits during the growing season. Moreover， mediation pathway analysis indicated that the dominant variable mediating the influence of permafrost thaw differed across regions. In continuous permafrost zones， soil temperature was the primary mediator linking permafrost thaw to peak vegetation growth. In contrast， in other permafrost types， particularly sporadic and isolated zones， the dominant mediation pathway was through soil moisture. These findings underscore the complex and spatially heterogeneous effects of permafrost thaw on high-latitude vegetation dynamics. Such variability highlights the importance of incorporating soil thermal-hydrological feedbacks and permafrost heterogeneity into evaluation of ecosystem responses to climate change. By capturing these interactions， the results provide a scientific basis for improving predictions of carbon cycling processes and for guiding policy decisions related to ecosystem conservation and climate mitigation in permafrost-affected regions.

Driven by the dual forces of global climate warming and intensified human activities， the degradation rate of permafrost on the Qinghai-Xizang Plateau is accelerating. This trend is particularly pronounced within the Qinghai-Xizang Engineering Corridor， where retrogressive thaw slumps （RTS） occur frequently and continue to expand. The development of RTS poses serious threats to the ecological environment and the safety of infrastructure within the corridor. Currently， methods for extracting RTS predominantly rely on visual interpretation of optical remote sensing imagery or deep learning models. However， these approaches demonstrate limited capability in recognizing potential hazard points that are in the early stages of development and lack conspicuous morphological characteristics. Research in this domain remains relatively insufficient. This study utilized Sentinel-2A satellite imagery acquired in 2023 to accurately delineate the boundaries of RTS in Beiluhe and the section extending from the hilly and mountainous areas west of Wudaoliang Town to the Cuodarima within the Qinghai-Xizang Engineering Corridor. By integrating multi-factorial disaster-inducing factors—including topography， vegetation， and permafrost—the developmental characteristics of RTS were analyzed， and thresholds for developmental conditions of RTS were established. Meanwhile， InSAR （interferometric synthetic aperture radar） data from Sentinel-1A， spanning 2018 to 2023 and collected via ascending and descending orbital passes in multi-temporal phases， were employed to analyze the deformation characteristics across the study area. By integrating disaster-inducing factors for RTS development， potential hazard points were identified. Typical sites of existing RTS and potential hazard points were then selected to analyze their deformation characteristics. The results showed that there were 622 RTS occurrences in the study area， with a total area of 12.507 km² （the maximum area was 0.29 km² and the average area was 0.02 km²）. The disaster-inducing conditions for RTS development were north-facing slopes in alpine steppe regions at an elevation of 4 700 to 4 900 meters， with slope gradients between 2° and 12°， normalized difference vegetation index （NDVI） values from 0 to 0.3， fractional vegetation cover （FVC） between 0.4 and 0.6， mean annual ground temperature （MAGT） spanning -3 to -1.5 °C， volumetric water content （VWC） of 30% to 40%， and active layer thickness （ALT） measuring 1 to 2 meters. In the study area， 124 potential hazard points were identified， with a total area of 3.12 km² （the maximum area was 0.22 km² and the average area was 0.0252 km²）. Among them， 47 potential hazard points were located in the Beiluhe area， and 77 within the Wudaoliang zone. Smaller hazard points exhibited relatively higher annual average settlement rates， indicating a greater likelihood of these areas evolving into active RTS in the near future. Deformation curve analysis demonstrated that RTS in an active expansion phase showed an annual average vertical displacement rate of approximately （-18±2.3） mm， with cumulative settlement exceeding 100 mm. Additionally， a high-velocity settlement zone was observed at the crown of the headscarp， with rates between 11 and 15 mm⋅a^-1. The frontal accumulation zone exhibited uplift deformation ranging from 4 to 7 mm⋅a^-1. Potential hazard points， in contrast， exhibited relatively high localized deformation rates while the surrounding terrain remained stable. Their cumulative surface settlement was about 80 mm—less than that observed at established RTS sites， and the surface showed periodic settlement， indicating a high-risk potential of evolving into RTS. The findings can provide a scientific basis for early warning systems for RTS disasters within the Qinghai-Xizang Engineering Corridor， contribute to ecological environment protection efforts， and inform the safe operation and maintenance of engineering infrastructure.

Pavement cracks are one of the earliest manifestations of road structure degradation， particularly prevalent in regions subject to extreme climatic variability. In permafrost regions， frequent freezing-thawing cycles and drastic ground temperature fluctuations induce mechanical stresses in roads that differ from those in ordinary regions， leading to significantly distinct crack formation mechanisms. Understanding these differences is crucial for developing robust crack detection systems adapted to diverse environmental conditions and formulating reasonable road maintenance strategies. This study aims to establish an intelligent and high-precision framework for crack detection capable of accurately identifying crack locations and extracting their geometric structural features that reflect regional differences. To achieve this objective， this study established an improved deep learning detection model based on the Faster Region-Based Convolutional Neural Network （Faster-R-CNN）， integrating multi-feature channels and multi-scale information. The model integrated three major innovations： （1） the introduction of a split-attention network to improve inter-channel feature discrimination and suppress background interference； （2） the integration of convolutional block attention module （CBAM） with ResNeXt50 in the feature pyramid network （FPN） to strengthen the representation of cracks at different scales； and （3） the use of the soft non-maximum suppression （Soft-NMS） algorithm to replace traditional NMS to retain more valid bounding boxes for slender or densely distributed cracks. This model was trained and validated on two independent datasets： UAV-based orthophoto data from the permafrost sections of the Qinghai-Xizang Highway that represented the permafrost regions， and urban pavement crack images from the widely used VOC-2007 dataset that represented the ordinary regions. All images were annotated manually， and training samples were expanded through data augmentation. The end-to-end optimization was performed using a joint classification and regression loss function. During evaluation， model performance was comprehensively measured using indicators including precision， recall， mean average precision （mAP） at both IoU=0.5 and IoU=0.5~0.95 thresholds. On the test set of the ordinary regions， the model achieved a precision of 92.20%， a recall of 89.25%， and a mAP （IoU=0.5~0.95） of 81.14%. On the permafrost dataset， although the overall accuracy was slightly lower due to complex backgrounds， the performance improvement over the baseline was more significant， demonstrating the model’s adaptability and stability in complex environments. Ablation experiments indicated that Soft-NMS increased mAP by approximately 5% and the integrated multi-module mechanism provided an 11%~14% improvement. Based on crack detection， this study developed a crack geometric parameter extraction module， capable of extracting indicators such as crack length， width， area ratio （R_c）， length density （L_c）， connectivity index （K）， and curvature to quantitatively characterize the spatial structure of cracks. The results showed that cracks in permafrost regions exhibited greater scale variability and geometric expansion， with typical transverse cracks exceeding 2.5 m in length， widths often over 10 mm， and connectivity indices generally above 0.6. The statistical distributions of L_c and R_c in permafrost regions were right-skewed， indicating high crack density and coverage. Further Pearson correlation analysis showed stronger coupling among geometric parameters in permafrost regions. For example， the correlation coefficient between L_c and R_c exceeded 0.65， reflecting a synergistic expansion of crack networks likely driven by thermal and hydraulic forces. In contrast， cracks in ordinary regions were more isolated， with weaker coupling between parameters. The contributions of this study are mainly reflected in two aspects. First， a highly accurate， generalizable crack detection model is proposed， suitable for automated inspection tasks across different road surface environments. Second， a standardized workflow of crack geometric analysis is developed， which can be applied to evaluate crack morphology and reveal deterioration mechanisms under specific climatic and geological conditions. These findings can inform maintenance agencies in prioritizing repair tasks based on indicators such as crack length density and connectivity and provide crack control parameters for road structure design in permafrost regions， thereby promoting the construction of more resilient road infrastructure. Moreover， the geometric parameters can serve as input variables for predictive models， facilitating intelligent evaluation of road performance and long-term early warning systems. In conclusion， this study integrates computer vision technology with environmental pavement science， representing significant progress in both automated detection technology and the mechanistic understanding of pavement deterioration processes.