In the context of climate warming and humidification on the Qinghai-Xizang Plateau， the Qinghai-Xizang Railway （QXR）， as a railway on plateau crossing the frozen soil regions， faces significant stability challenges-particularly the frozen soil subgrade-due to the presence and dynamics of surrounding surface water bodies. Based on remote sensing data， this study obtained the distribution of surface water bodies within a 1 000 m buffer zone along the QXR from 2006 to 2021 and analyzed their spatiotemporal distribution characteristics. Combined with case studies， this study investigated the impact of water ponding along the sides of the railway on the deformation stability of the frozen soil subgrade and its macroscopic manifestations. The results revealed distinct patterns in the distribution of surface water near the QXR. Quantitatively， water bodies smaller than 0.5 hm² dominated the landscape， accounting for the majority. Spatially， the distribution of these water bodies showed a pronounced orientation， primarily aligned along a southeast-northwest direction relative to the railway. Temporally， significant changes were observed during the study period. The total number of identifiable surface water bodies exhibited a consistent upward trend， increasing from 1 809 in 2006 to 2 170 in 2021， representing an increase of 19%. Concurrently， the area of these water bodies expanded. Notably， the proportion of larger-area water bodies （exceeding 2 hm²） within the total count increased from 6.91% to 7.74%， indicating a trend toward both more numerous and larger-area surface water features along the railway. Furthermore， detailed case analyses provided compelling evidence that water ponding adjacent to the railway subgrade profoundly influenced the deformation stability of the underlying frozen soil subgrade. A critical feedback mechanism-an interaction loop among the frozen soil layer， surface ponding， and the railway subgrade-was identified. Both engineering disturbances （primarily associated with the construction and presence of the railway） and ongoing climate change served as key drivers triggering the degradation of the subgrade’s supporting frozen soil. Crucially， this degradation of frozen soil caused depressions and altered local hydrological conditions， directly leading to the formation of surface water ponding. The presence of this ponded water served as a significant secondary driver， accelerating further thermal erosion and degradation of the adjacent and underlying frozen soil through enhanced heat transfer into the ground. This positive feedback loop， where frozen soil thaw led to water ponding that in return accelerated thaw， inevitably exacerbated subgrade deformation. The case studies highlighted a particularly hazardous scenario： when water ponding was concentrated predominantly on one side of the railway subgrade. This asymmetrical saturation induced a thermal imbalance across the subgrade. The side exposed to water experienced significantly accelerated frozen soil thaw and ground weakening compared to the drier side. This disparity led to pronounced uneven subsidence or differential settlement of the railway track， posing a serious threat to track geometry， alignment， and overall operational safety. Such deformation represented a macroscopic manifestation of the destabilizing influence of unilateral water ponding. Consequently， this study emphasizes that the effective management and mitigation of water ponding near the subgrade must become a priority for ongoing engineering maintenance and operational safety protocols of the QXR. Proactive measures to monitor， drain， or prevent the formation of such water bodies are essential to interrupt the identified frozen soil-degradation feedback loop and ensure the long-term stability of the infrastructure. In summary， this study provides critical empirical data derived from remote sensing on the evolving distribution of surface water near the QXR. Through case studies， it elucidates the mechanistic link between roadside ponding and the instability of frozen soil subgrade. The findings provide substantial data support and case references to directly inform and enhance engineering maintenance strategies crucial for preserving the structural integrity and operational reliability of this vital high-altitude railway in a changing climate.
Under ongoing global warming， the frequency and intensity of extreme weather and climate events have increased significantly， exerting a profound impact on the cryosphere. Compared with changes in mean climate conditions， extreme events often act on glacier systems in an intense and short-term manner， triggering rapid and non-linear responses in glacier mass balance. As highly sensitive indicators of climate change， glaciers respond directly to extreme variations in temperature and precipitation， particularly manifested in the significant enhancement of ablation processes. However， existing studies have mostly focused on glacier responses under mean climatic conditions or at the regional scale， while assessments of the impact of extreme weather and climate events on glacier mass balance at the interannual scale remain relatively limited， especially for continental glaciers in arid and semi-arid regions. The Qiyi Glacier， a typical continental glacier located in the Beida River Basin of the central Qilian Mountains， is selected as the study object. This glacier is a typical continental glacier and possesses relatively abundant long-term observational data， providing favorable conditions for investigating glacier-climate interactions under both mean and extreme climate conditions. The primary objectives of this study are： to reconstruct the mass balance series of the Qiyi Glacier over the past five decades， to characterize the long-term evolution of extreme temperature and precipitation events in the study area， and to quantitatively assess the relative contributions and potential non-linear impacts of extreme weather and climate events， particularly extreme high-temperature events， on glacier mass balance. To achieve these objectives， the Open Global Glacier Model （OGGM） was employed to simulate the annual mass balance of the Qiyi Glacier for the period 1974—2024. Glacier outlines were derived from the Randolph Glacier Inventory （RGI 6.0）， and historical climate forcing data were obtained from the bias-corrected W5E5 reanalysis dataset. Model parameters， including temperature sensitivity， precipitation correction factor， and temperature bias， were calibrated using geodetic data and measured mass balance data. Based on daily meteorological observational data， extreme temperature and precipitation events were quantified using indices recommended by the Expert Team on Climate Change Detection and Indices （ETCCDI）. In addition， the degree-day factor approach was applied to distinguish the relative contributions of extreme high-temperature days and non-extreme days to glacier ablation. Statistical methods， including trend analysis， abrupt change detection， correlation analysis， first-difference analysis， and regression modeling， were employed to systematically explore the relationships between extreme climate indices and glacier mass balance. The results indicated that： （1） over the past five decades， the mass balance of the Qiyi Glacier exhibited an overall negative trend， with an overall rate of -0.102 m w.e.⋅a-1. During 1974—1993， the glacier experienced a state of net mass accumulation， with a rate of +0.259 m w.e.⋅a-1. An abrupt change occurred in 1993， after which the glacier transitioned into a phase of rapid mass loss， with a loss rate of -0.336 m w.e.⋅a-1. （2） During the study period， the climate in the Qiyi Glacier region experienced a pronounced warming and wetting trend. Specifically， warm extreme indices increased markedly， while cold extreme indices declined significantly. The diurnal temperature range （DTR） showed an overall weak decreasing trend， with a rate of -0.08 ℃⋅（10a）-1， and the decreasing rate accelerated markedly after the 1990s， reaching -0.29 ℃⋅（10a）-1. The regional wetting trend was evident， with annual total precipitation increasing significantly at a rate of 34.73 mm⋅（10a）-1. （3） Extreme high-temperature events were identified as the dominant driver of the accelerated mass loss of the Qiyi Glacier by enhancing ablation processes. Their annual average contribution to glacier ablation reached 21.3%， and approached nearly 50% in years with a high frequency of extreme high-temperature events. Preliminary regression analyses based on short-term observational data indicated a pronounced non-linear impact of extreme high-temperature events on glacier mass balance. In contrast， the regulating effect of precipitation on glacier ablation was relatively limited and gradually weakened with the increasing frequency of extreme high-temperature events. This study provides a preliminary quantitative assessment of the impact of extreme weather and climate events， particularly extreme high-temperature events， on glacier mass balance， offering a useful reference for deepening the understanding of glacier response mechanisms under extreme climate conditions and for future research on glacier evolution under climate change scenarios.
As a vital component of the terrestrial water cycle， snow is one of the most critical freshwater resources in arid and semi-arid regions， playing an irreplaceable role in regulating surface hydrological processes， maintaining ecosystem stability， and ensuring regional water supply. Located in the arid interior of Northwest China， Xinjiang features a typical arid and semi-arid climate， with a high dependence on snowmelt. In recent years， as climate change intensifies， Xinjiang has experienced significant increases in temperature， leading to a marked reduction in snow cover extent and duration， with the phenomenon of “snow drought” becoming increasingly prominent. However， a reasonable snow drought classification system based on Xinjiang’s unique snow and climate characteristics has not yet been established. Based on this， the ERA5 reanalysis dataset， China’s daily precipitation dataset， and China’s 25 km daily snow water equivalent （SWE） product were utilized. By comprehensively considering the key climatic conditions and characteristics during snow formation and ablation， four climate indicators—snow water equivalent， precipitation， thawing degree， and radiation—were selected to classify snow drought in Xinjiang.Given that snow accumulation primarily occurred in autumn and winter and ablation primarily in spring， this study focused on the meteorological conditions during the cold season and their impact on snow changes in the classification of snow drought. Therefore， the cold season （from November 1 to April 1 of the following year） of each hydrological year was selected as the study period to systematically analyze the snow distribution characteristics and drought evolution processes during this period. Based on this， snow drought during the cold season in Xinjiang was categorized into three types： warm type， dry type， and warm-dry type. The spatiotemporal evolution characteristics of snow drought during the cold season in Xinjiang from 1980 to 2020 were systematically analyzed. The results indicated that snow drought during the cold season in Xinjiang exhibited significant spatial variability， with a distribution pattern of higher occurrence frequency south of the Tianshan Mountains and lower frequency north of the Tianshan Mountains. Specifically， the average occurrence frequency of snow drought in the area south of the Tianshan Mountains was approximately 38%， significantly higher than the 18% in the area north of the Tianshan Mountains， demonstrating a pronounced pattern of “higher in the south， lower in the north”. Among them， the Tarim Basin and its surrounding areas experienced the highest occurrence frequency of snow drought， reaching up to 47%. In contrast， regions such as the Ili River Valley and the northern margin of the Junggar Basin， influenced by moisture conditions and orographic lifting， exhibited relatively favorable snow conditions and the lowest occurrence frequency of snow drought， at only about 7%.Different types of snow drought also showed significant differences in their long-term trends. Over the past four decades， the maximum area proportions of warm-type and dry-type snow drought significantly decreased from 53.91% and 59.04% to 17.61% and 6.98%， respectively， with their occurrence ranges continuously shrinking at rates of 275 km2⋅a-1 and 87.5 km2⋅a-1. In contrast， the area proportion of warm-dry snow drought steadily increased from 22.41% to 47.11%， expanding by approximately 31.25 km2 per year， gradually becoming the dominant type of snow drought during the cold season in Xinjiang.Further analysis indicated that the formation mechanisms of different types of snow drought were significantly different， with distinct dominant driving factors. Warm type snow drought was driven by the combined effects of multiple climatic factors， exhibiting pronounced multi-factor driving characteristics. Dry type snow drought was primarily governed by precipitation conditions and was more sensitive to anomalies in cold-season precipitation. Warm-dry type snow drought， however， was mainly driven by rising temperatures， which shortened snow cover duration and triggered earlier snowmelt， making it a representative compound snow drought type under climate warming.Overall， this study develops a snow drought classification indicator and system tailored to the snow and climate distribution characteristics of Xinjiang. The findings provide a scientific basis for an in-depth understanding of the evolution patterns and formation mechanisms of snow drought in Xinjiang under climate change， and also offer crucial theoretical support for the rational allocation of water resources， snow disaster prevention and control， and ecological security in arid and semi-arid regions.
In the context of global warming， the degradation of permafrost on the Qinghai-Xizang Plateau has altered the hydrothermal dynamics of the active layer and the surface energy balance， exerting significant impacts on regional ecology， engineering infrastructure， and global climate. Conducting relevant research is of great significance for the safety and stability of alpine ecosystems on the plateau. Due to the complex terrain of the Qinghai-Xizang Plateau and the limited number of monitoring stations， numerical modeling has become a commonly used research approach. However， existing models still have room for improvement in simulating soil moisture. GEOtop is a process-based distributed hydrological model that enhances the simulation of water movement by solving three-dimensional flow equations， yet its application in the permafrost regions of the Qinghai-Xizang Plateau remains relatively limited. Taking the Tanggula station， located in the permafrost region of the Qinghai-Xizang Plateau， as the research object， this study systematically analyzed the variation characteristics of active layer hydrothermal processes and surface energy fluxes based on observational data from 2005 to 2006， and conducted numerical simulation experiments using the GEOtop model. The results showed that soil temperature in the active layer in the Tanggula region exhibited a sinusoidal pattern， with decreasing amplitude and increasing phase lag with depth. Soil moisture displayed pronounced seasonal variations that were closely synchronized with precipitation. The freeze-thaw process was characterized by “unidirectional thawing and bidirectional freezing”. All surface energy fluxes varied seasonally， and sensible and latent heat showed distinct seasonal alternation patterns. In addition， the GEOtop model could well simulate the dynamic variations of soil temperature and moisture， with average correlation coefficients （r） above 0.94. The simulation of the active layer thawing process was also relatively accurate， with a deviation of only 4% in the simulated active layer thickness. Regarding surface energy fluxes， the GEOtop model performed well in simulating net radiation， sensible heat， and latent heat， but its simulation of surface soil heat flux had certain limitations due to the simplified model assumptions. Overall， the GEOtop model can accurately describe the surface-active layer hydrothermal processes in the Tanggula permafrost region， demonstrating certain applicability in permafrost regions on the Qinghai-Xizang Plateau.
Engineering construction in permafrost regions inevitably alters the thermal regime of the underlying permafrost， thereby threatening the bearing capacity of the permafrost foundation and the long-term stability of overlying artificial structures. Despite increasing infrastructure demands in cold regions， hydraulic engineering projects in permafrost regions remain scarce， with limited field monitoring data available. Consequently， the thermal impact of engineering construction activities， especially in combination with flowing water， on permafrost foundations is still not well understood. This study focused on a hydraulic channel project located in a continuous permafrost region in the hinterland of the Qinghai-Xizang Plateau. To investigate the thermal response of permafrost to engineering disturbance and hydrological processes， continuous ground temperature monitoring was conducted at four strategically selected boreholes—natural ground， channel slope， channel slope toe， and channel center—to a depth of 20 meters over a four-year observation period from 2020 to 2024. A systematic analysis was performed on variations in active layer thickness， shallow and deep permafrost temperatures， and near-surface energy budget processes under different surface conditions and proximity to the water flow. The results showed that， under the influence of heat carried by the flowing water， the onset of freezing at the channel center and channel slope toe boreholes was significantly delayed compared with that at the natural ground and channel slope boreholes. Channel excavation and the heat carried by the flowing water induced degradation of the underlying and surrounding permafrost. During the monitoring period， the active layer thickness at the natural ground borehole increased from 3.9 to approximately 4.5 m， whereas those at the channel slope， channel slope toe， and channel center boreholes increased year by year. Permafrost within a 10-meter depth range at the channel slope， channel slope toe， and channel center boreholes exhibited different degrees of warming during the monitoring period， and the closer a borehole was to the center of the flow section， the more pronounced the thermal erosion effect on the underlying soil. In terms of near-surface soil temperature and energy budget， the annual average shallow ground temperature at the channel center borehole was 3.4 to 5.2 ℃ higher than that at the natural ground borehole.The channel center borehole consistently remained in a state of heat absorption， with the maximum annual heat absorption reaching 7.9507×10⁴ kJ⋅m-2. These findings， based on long-term field monitoring data， provide essential references for future analyses of the thermal stability of channel foundations in permafrost regions.
The construction and maintenance of transportation infrastructure in permafrost regions often face significant challenges due to the complex morphological changes of frozen soil under specific thermal conditions. These engineering challenges primarily stem from three types of frozen soil-related distresses： seasonal freeze-thaw cycles， frost heave， and thaw settlement. Among these damaging phenomena， thaw settlement is the primary mechanism causing road damage， such as excessive subgrade deformation， differential settlement， and subsequent pavement distresses， including cracking and structural failure. Rising ground temperatures accelerate permafrost degradation， making thaw settlement-related damages increasingly prevalent in cold-region engineering projects. Consequently， these distresses are exacerbated under the warming climate. Therefore， development projects in Northwest， Northeast， and the Qinghai-Xizang regions of China are increasingly concerned with the vulnerability of transportation infrastructure in these thermally sensitive environments. This study investigated several representative permafrost sections along the Qinghai-Xizang Engineering Corridor， and the relevant field investigation was conducted through systematic drilling of 139 permafrost samples at various depths and geographically distributed locations along the highway. In terms of the sampling method， this study employed specialized equipment and preservation techniques to maintain the original structure and ice content of the permafrost samples， which was essential for obtaining representative experimental data. Laboratory analysis followed rigorous testing procedures， beginning with detailed particle size distribution tests and soil sample classification. Based on standardized classification criteria， samples were categorized into three typical soil types： coarse-grained soil， silt， and silty clay， thereby enabling a systematic comparison of the thaw settlement properties of different soil types. The laboratory thaw settlement tests were conducted using a specialized， standardized thaw consolidation apparatus， following established permafrost testing methods. The test results established fundamental relationships between thaw settlement coefficient and basic soil parameters. Statistical analysis demonstrated that the thaw settlement coefficient exhibited a strong positive correlation with water content and a significant negative correlation with dry density. Among the three soil types， silty clay demonstrated significant sensitivity to changes in both parameters， with a significantly larger regression slope. The study identified critical thresholds for silty clay， in which the boundary water content for strong thaw settlement was approximately 44.2%， and the boundary dry density for strong thaw settlement was approximately 1.15 g·cm⁻³. These values， determined through statistical analysis of the experimental data， provided practical guidance for engineering design and construction control. Subsequently， a comparative analysis of the empirical formula derived from this study with those obtained from existing literature and the Chinese national standard revealed significant discrepancies. The formulas in the current standard generally exhibited steeper slopes and yielded larger estimated values of the thaw settlement coefficient. The deviation suggested that the thaw settlement coefficient may be overestimated， particularly under conditions of high water content and low dry density that were common in ice-rich permafrost areas. Such overestimation could lead to excessive mitigation measures in projects along the Qinghai-Xizang Engineering Corridor， potentially resulting in overly conservative engineering design and increased construction costs. Additionally， the current standard’s practice of classifying silt and silty clay into a single category and applying a unified empirical formula to both represented a major limitation. This study clearly demonstrated that silty clay exhibited fundamentally distinct thaw settlement properties from silt， thereby necessitating separate predictive models. Based on these findings， corresponding recommendations were proposed for engineering practice. To accurately predict thaw settlement， field or laboratory testing should be used as the primary method for determining site-specific parameters， particularly for critical engineering projects with low settlement tolerance. Furthermore， due to the unique geotechnical characteristics of silty clay， there is an urgent need to establish an independent predictive model specifically for calculating its thaw settlement coefficient.
Statistical analysis conducted by the maintenance department of the China Railway Lanzhou Group Co.， Ltd. reveals that frost heave and thaw settlement in the existing Lanzhou-Xinjiang Railway roadbed， located in deep seasonally frozen soil regions， are particularly severe. Among these issues， the roadbed-culvert transition sections exhibit a high frequency of frost damage and significant frost heave displacement. This situation seriously compromises the operational safety of the Lanzhou-Xinjiang Railway， reduces transportation efficiency， and substantially increases maintenance costs. To address the frost damage problem in existing railway roadbed-culvert transition sections within （deep） seasonally frozen soil regions， field monitoring of roadbed temperature and moisture changes was conducted at the K411+860~K411+880 roadbed-culvert transition section along the Jiling-Shandan section of the Lanzhou-Xinjiang Railway. The spatiotemporal evolution of the hydrothermal environment in the transition section during seasonal freeze-thaw processes was investigated to reveal the hydrothermal coupling mechanisms behind the formation of frost heave damage. This study also clarified the thermophysical and mechanical indicators， along with the frost heave susceptibility classification， of roadbed fill in typical frost damage sections of the existing Lanzhou-Xinjiang Railway under long-term train loading and freeze-thaw cycles. Based on the identified causes of frost damage， targeted remediation measures were proposed. The research results showed that the presence of the culvert and the forced convection effect induced by train operation caused significant spatial heterogeneity in the temperature field of the roadbed-culvert transition section. The closer to to the culvert sidewall， the greater the frost depth and the longer the freezing period. The frost depth on the culvert side reached 2.10 m， with a freezing period of 175 days. Similarly， the closer to the track， the greater the frost depth and the longer the freezing period. At 1.50 m outside the protective net， the frost depth was 1.50 m and the freezing period was 130 days. The frost depth at the natural ground was 1.20 m， with a freezing period of 105 days. The roadbed soil was mainly low-liquid-limit silty clay with a silt content exceeding 70%， and the actual degree of compaction was insufficient. At the initial monitoring stage， the average volumetric water content of the roadbed soil was higher than the threshold water content for frost heave， indicating high susceptibility to frost heave and thaw settlement under the combined effects of soil properties， low temperatures during the cold season， and high water content. In addition， the porous overhead structure within the soil layer and the temperature field heterogeneity induced by the culvert strongly affected the moisture distribution within the roadbed-culvert transition section. During the monitoring period， the average fluctuation amplitude of the unfrozen water content in the homogeneous clay layer at 1.50 m on the culvert side was 1.76 times that in the porous overhead structure layer within the transition section， indicating that the combined effects of the porous overhead structure and the temperature field weakened moisture migration and redistribution during freeze-thaw cycles. Based on the identified causes of frost damage， engineering measures combining thermal insulation and waterproofing are recommended for this transition section. The findings not only enrich the theoretical understanding of frost damage in roadbed-culvert transition sections within deep seasonally frozen soil regions but also provide theoretical and technical guidance for preventing and mitigating such damage in railway infrastructures in these regions.
The development of resources in permafrost regions has led to seepage issues through overburden rock fractures， causing groundwater system imbalances and accelerating permafrost degradation. In remediation practices， grouting technology is widely applied for sealing overburden fractures， reinforcing surrounding rock， and preventing seepage. However， its effectiveness is constrained by adverse factors in extreme cold regions—including harsh climatic conditions， poor transportation access， low construction efficiency， and risks of oxygen deprivation for personnel—making it difficult to achieve ideal remediation goals. Leveraging the pronounced “thermal semiconductor” effect of block rock structures—which promotes permafrost development in cold regions—this study designed an experimental apparatus to simulate frozen ground improvement of overburden fractures. The setup comprises a simulated test chamber， simulated fracture layer， data acquisition system， simulated rainfall system， and temperature control system. Coal gangue was used as the test material， with boundary conditions set based on actual meteorological data from high-altitude mining areas： Temperature boundaries were set by fitting annual daily average temperatures to simulate a complete freeze-thaw cycle； rainfall boundaries were determined based on the median rainfall intensity of the region’s annual precipitation. Tests were conducted under both rainfall and non-rainfall conditions to investigate the cooling performance and fracture sealing effectiveness of coal gangue layers at three thicknesses： 70 cm， 80 cm， and 90 cm. Results indicate that all three coal gangue layer thicknesses （70 cm， 80 cm， and 90 cm） exhibit excellent heat dissipation capabilities. As layer thickness decreases， heat absorption per cycle increases， yet the layers maintain overall net heat dissipation. The 80 cm layer demonstrates the highest net heat release per cycle and the best cooling performance. Simultaneous rainfall effectively reduced heat infiltration into the bottom of the coal gangue layer. The active thermal effect of the coal gangue layer counteracted rainfall's impact on bottom temperatures， maintaining stable temperatures below 0 ℃ at the base. All three thicknesses demonstrated effective cooling performance. The 80 cm thick coal gangue layer demonstrated superior convective heat transfer during cooling， enabling its bottom periodic average temperature to reach -3.57 ℃—outperforming the 70 cm and 90 cm variants. The 70 cm， 80 cm， and 90 cm thick coal gangue layers all effectively promoted permafrost development and achieved effective sealing of overburden fractures in the high-altitude mining area. Specifically， the 80 cm thick coal gangue layer achieved complete fracture sealing within the third cycle， whereas the 70 cm and 90 cm thick layers required five and seven cycles， respectively， to complete sealing. These findings provide valuable reference for addressing overburden fissures in approximately 3 970 abandoned mines across high-altitude regions of the Qinghai-Xizang Plateau， offering theoretical support for near-natural restoration in frigid mining areas.
Accurate characterization and prediction of the ground temperature field are essential for evaluating the thermal stability of highway subgrades in permafrost regions under a warming climate. In high-altitude permafrost areas， the combined effects of rising air temperatures， complex surface energy exchange， and pronounced sunny-shady slope effect lead to significant spatial heterogeneity of boundary conditions. Conventional numerical models often adopt uniform or single-point calibrated upper boundary conditions， potentially neglecting transverse thermal differences across the embankment and consequently reducing simulation reliability， especially when monitoring data are limited. To address these limitations， this study proposed a coupled physical and data-driven framework that explicitly incorporated the spatial heterogeneity of boundary conditions and enabled time-dependent prediction of key boundary parameters for permafrost subgrade temperature field analysis. The governing equation of transient heat conduction with phase change was established based on Fourier’s law and the principle of energy conservation. The enthalpy method was introduced to account for the ice-water phase transition， thereby avoiding explicit tracking of the moving phase interface and improving numerical robustness. The model was implemented in COMSOL Multiphysics using the coefficient-form partial differential equation module. Thermophysical parameters， including thermal conductivity， density， and temperature-dependent heat capacity， were determined according to soil composition and phase state. A two-dimensional subgrade-foundation model was constructed for the Xieshuihe section of the Qinghai-Xizang Highway （China National Highway 109）， where long-term ground temperature monitoring data were available. The lateral boundaries were treated as adiabatic， and a constant geothermal gradient was imposed at the bottom boundary. Initial conditions were calibrated using in-situ temperature profiles to ensure consistency between steady-state initialization and subsequent transient simulation. To represent the spatial heterogeneity of boundary conditions induced by the sunny-shady slope effect， a linear interpolation scheme was adopted to quantify spatial variations of annual average ground temperature， temperature amplitude， and phase angle along the embankment surface. This approach enabled a continuous representation of boundary heterogeneity under limited monitoring data， while avoiding over-parameterization associated with higher-order interpolation or complex geostatistical models. For temporal prediction of boundary conditions， a long short-term memory （LSTM） neural network was developed to predict surface ground temperature evolution based on multi-year meteorological and ground temperature time series. The predicted annual average temperature increment and temperature amplitude were then incorporated into the upper boundary formulation to simulate future thermal responses. Model performance was evaluated using monitoring data from 2010 to 2015 at multiple locations and depths， including the subgrade center， shoulders， berms， and natural ground. Statistical indicators demonstrated satisfactory agreement between simulated and measured values， with RMSE ranging from 0.32 ℃ to 1.23 ℃ and R² up to 0.98 at representative depths. The model effectively reproduced the asymmetric freezing and thawing processes across the embankment， confirming that accurately representing the spatial heterogeneity of boundary conditions was critical for capturing transverse thermal differences. The shady slope exhibited faster freezing rates and a greater freezing depth， whereas the sunny slope showed delayed freezing and enhanced thawing intensity， consistent with field observations. For the period 2015—2020， the LSTM model achieved an RMSE of 2.189 ℃ and an R² of 0.784 in surface temperature prediction. When the predicted boundary parameters were applied to the numerical model， the simulated temperature fields at different depths maintained consistent phase characteristics and magnitude trends compared with measured data. Although deviations increased slightly with depth near the permafrost table， the overall differences remained within an acceptable range. The integrated framework effectively captured the short- to medium-term thermal evolution of the subgrade under climate warming conditions. The proposed methodology demonstrates that combining physics-based heat transfer modeling with data-driven boundary prediction provides a practical and reliable solution for simulating ground temperature fields in permafrost regions with limited monitoring data. By explicitly addressing the spatial heterogeneity of boundary conditions and the temporal variability of upper boundary parameters， the approach enhances predictive capability and supports risk assessment， design optimization， and maintenance planning for transportation infrastructure in cold regions.
Soil salinization is the result of the combined effects of salt migration， water evaporation， and salt precipitation. Due to the characteristics of saline soil such as collapsibility， salt expansion， and corrosiveness， construction projects in saline soil areas are prone to damage including subsidence and deformation， severely compromising the stability and durability of construction projects. Among them， the salt expansion caused by moisture content changes in sulfate saline soil has long been a focus and challenge of research. This study employed aeolian sand to improve the water stability of sulfate saline soil. Specimens were prepared through the dry-wet cycle method， and the effects of sulfate content， number of dry-wet cycles， and aeolian sand content on the physical and mechanical properties of the specimens were systematically investigated through triaxial and microscopic tests. At the macroscopic level， the impact of aeolian sand on the mechanical properties of sulfate saline soil was analyzed based on strength changes of specimens under dry-wet cycles. Additionally， the height changes of specimens under dry-wet cycles and the effects of different variables on the porosity of specimens were explored. At the microscopic level， scanning electron microscopy （SEM） and energy-dispersive X-ray spectroscopy （EDS） were utilized to analyze the micromorphology and elemental composition of specimens， revealing the mechanisms of physical and mechanical changes of sulfate saline soil specimens under dry-wet cycles and clarifying the microscopic mechanisms of aeolian sand in sulfate-aeolian sand composite specimens. The experimental results indicated that the stress-strain curves of sulfate saline soil specimens all exhibited strain-hardening behavior under dry-wet cycles. Strength degradation occurred in three stages： rapid decay during the first three dry-wet cycles， stabilization between the third and seventh cycles， and slow decline from the seventh to eleventh cycles. By comparing the structural degradation rates of different sulfate-aeolian sand composite specimens after dry-wet cycles， it was found that the optimal aeolian sand content was 5%， with a 37.6% reduction in structural degradation rate. However， when the content increased to 10% and 20%， the degradation rates rose to 61.4% and 65.5%， respectively. Dry-wet cycles significantly affected the porosity of sulfate saline soil specimens. During the drying stage， pore water evaporation increased the salt concentration to supersaturation， causing the liquid sodium sulfate within the pores to absorb water and crystallize into sodium sulfate decahydrate. Salt crystals accumulated at the contact points between soil particles， generating local compressive stress that disrupted the original cemented structure and particle arrangement， macroscopically manifesting as salt expansion. During the wetting stage， the ion concentration in pore water decreased， the thickness of the electric double layer increased， and the repulsive forces between particles enhanced， leading to an increase in inter-particle spacing. Furthermore， the bound water film adsorbed on clay particle surfaces thickened， weakening the physical connections between particles. The bound water between particles facilitated particle sliding and rearrangement， transforming the particle contacts from face-to-face to a loose face-to-edge arrangement， resulting in a macroscopic volume increase of the specimens. The effect of aeolian sand content on the strength of sulfate-aeolian sand composite specimens exhibited nonlinear characteristics. With low aeolian sand content， the dry-wet cycles broke down clay particles into fine particles， and moisture migration drove these particles to fill the pores. This temporarily optimized the gradation to form a dense structure， effectively suppressing moisture intrusion and expansion deformation， thereby increasing specimen strength. However， with slightly higher aeolian sand content， continuous sand addition led to significant gradation degradation. Clay bonding broke to form continuous permeable channels， and rapid moisture migration along particle interfaces caused severe specimen deformation and a sharp decrease in strength. With high aeolian sand content， particles became extremely uniform， and the porosity and average pore size of the specimens increased synergistically. The loss of structural stability caused a further decrease in strength. Finally， the trends of sulfate content， number of dry-wet cycles， porosity， aeolian sand content， and confining pressure on strength were analyzed. Based on these trends， the Duncan-Chang model was modified， and the modified model could accurately predict the stress-strain curves of specimens under dry-wet cycles. This study analyzes the degradation mechanisms of sulfate saline soil specimens under dry-wet cycles and proposes that an appropriate amount of aeolian sand can improve the water stability of saline soil， providing a theoretical basis for enhancing the water stability of saline soil.
To investigate the mechanical properties of frozen clay under different testing conditions， uniaxial compression tests and graded loading creep tests were conducted. The effects of salt contents （0%， 0.5%， 1.0%， and 2.0%） at temperatures of -5 ℃， -10 ℃， -15 ℃， and -20 ℃ on the unconfined compressive strength and uniaxial compressive creep deformation characteristics of frozen clay were analyzed. By introducing the internal state variable theory and treating strain as an internal state variable， an internal variable-based creep model was developed to describe the creep behavior of saline frozen clay. Its reliability was verified through quantitative comparison between model predictions and experimental results. The test results indicated that salt content had a significantly adverse effect on the mechanical performance of frozen clay. As the salt content increased， the unconfined compressive strength and deformation modulus decreased linearly. The strength of specimens with a salt content of 2.0% degraded by approximately 40% compared to those with 0% salt content. The relationship between strength reduction and salt content followed an exponential decay pattern， with the most significant strength deterioration occurring within the salt content range of 0.5% to 1.0%. This degradation mechanism was attributed to the depression of the pore water freezing point， which led to an increase in unfrozen water content， thereby weakening the cementation between ice and soil particles. Conversely， decreasing temperature significantly enhanced the mechanical properties due to strengthened ice cementation， partially counteracting the adverse effects of salt. A temperature drop from -5 ℃ to -20 ℃ increased the strength of specimens across all salt contents by 55%~60%， indicating that temperature played a crucial role in controlling mechanical behavior. A noticeable transition in failure mode was observed. Under low salt content conditions， failure was predominantly brittle with distinct shear planes， while it gradually shifted to plastic failure characterized by bulging and distributed cracking as the salt content increased. The transition threshold occurred at approximately 1.0% salt content. Beyond this value， the soil behavior changed from brittle failure to plastic failure. Analysis of creep behavior revealed typical three-stage creep curves under all testing conditions. The creep process was significantly influenced by both stress level and environmental factors. Under different stress and temperature conditions， the creep curves of the clay exhibited similar shapes， undergoing decelerated （primary）， steady-state （secondary）， and accelerated （tertiary） creep stages， with the duration of each stage varying due to differences in stress and temperature. The creep rate curves consistently showed the characteristics of “deceleration-stable fluctuation-acceleration to failure”. Under identical temperature and salt content conditions， higher stress levels resulted in greater creep rates， while low stress levels significantly prolonged the duration of the steady-state stage. At lower stress levels， specimens primarily exhibited decelerated and steady-state creep， whereas at higher stress levels， the accelerated creep stage developed rapidly， leading to structural failure. Additionally， the study found that increased salt content accelerated creep deformation and shortened the time to failure， particularly under high-stress conditions. Under identical temperature and pressure conditions， the steady-state creep rate increased with rising salt content. Conversely， lower temperatures effectively suppressed creep development and extended the durability of frozen structures. Under the same stress and salt content conditions， creep strain decreased as temperature dropped. Based on the experimental results， an internal variable creep model considering the effect of salt content was established by introducing salt content as a key parameter. This model expressed the creep rate as a function of stress， accumulated strain， and salt concentration via a power-law relationship. A salt content correction factor was introduced to quantitatively account for the plasticizing effect of the unfrozen water film at the ice-particle interface. Model parameters were systematically optimized using experimental data， demonstrating strong predictive capability at low to medium stress levels， with goodness-of-fit （R 2） all above 0.89. The model was validated against tests on three typical frozen soils： silt， mine clay， and sulfate saline soil， showing consistent trends and good agreement between calculated and experimental values across different soil types. The study concludes that salt content promotes creep， while low temperature inhibits deformation. This model is only applicable for validating creep under relatively low load levels. Under high load levels， the soil experiences damage and quickly enters the accelerated creep stage， where the model becomes no longer applicable. The research findings are applicable to engineering projects involving saline frozen soil， where salt content significantly affects ground freezing performance and the long-term stability of frozen soil structures.
In recent years， global climate has shown a warming trend， leading to the thawing of shallow frozen soil in some regions. During frozen soil thawing， the solid water within the soil is converted into liquid water and discharged， which substantially reduces the stiffness of the soil around the pile. Therefore， this process inevitably and significantly affects the bearing capacity of existing engineering structures in frozen soil regions. To investigate the effects of shallow frozen soil thawing on the bearing performance of pile foundations， this study conducted indoor model tests on frozen soil pile foundations. These tests compared two scenarios—frozen soil pile foundations and thawed soil pile foundations—and analyzed bearing characteristics such as pile strain， pile-side freezing stress， pile foundation bearing capacity， and pile-top displacement. The results were intended to provide a reference for pile foundation design in regions affected by frozen soil thawing. The results indicated that under all test conditions， the pile strain gradually decreased with increasing depth and increased with increasing load. Under the same load， at any burial depth， the strain of the pile in thawed soil was consistently greater than that of the pile in frozen soil. Within the thawed soil zone， the lower stiffness of the soil around the pile resulted in a gradual change in pile strain， mainly increasing compressive deformation in the lower part of the pile. The pile-side freezing stress along the pile first increased and then decreased with burial depth， with larger values in the middle of the pile and smaller values at both ends， and it increased as the load increased. Frozen soil thawing significantly affected the pile-side freezing stress. Compared with frozen soil pile foundations， the burial depth of maximum stress increased. The stress values in the upper and middle parts of the pile decreased significantly， while those in the lower part slightly increased. This indicated that the thawing of frozen soil weakened the freezing strength of the nearby pile-soil interface， causing an increase in pile-side freezing strength in the lower part of the pile. The bearing capacity of the pile foundation was mainly provided by the pile-side freezing force， with the pile-end resistance accounting for a relatively small proportion. Both increased with the increasing load， and the contribution of the pile-side freezing force became increasingly significant as the load increased. Overall， frozen soil thawing reduced the pile-side freezing force， increased the pile-end resistance， and thus significantly altered the bearing characteristics of the pile foundation. The pile-top displacement continuously increased with the increasing load. Furthermore， as the load increased， both the incremental displacement and the time needed to stabilize at each load level also gradually increased. At the same load level， the displacement of the thawed soil pile foundation was greater than that of the frozen soil pile foundation. In this experiment， frozen soil thawing reduced the bearing capacity of the pile foundations by 16.9%. Therefore， frozen soil thawing is highly detrimental to the bearing capacity of pile foundations， and thermal insulation measures for frozen soil should be fully considered in engineering practice.
Saline soils are widely distributed in China’s seasonally frozen soil regions. In recent years， the construction scale of major infrastructure projects such as airport runways， highway subgrades， and transmission tower foundations in saline soil areas has continuously expanded， often partially or completely covering the saline soil foundation. Under the repeated action of seasonal freeze-thaw cycles， water and salt within the saline soil foundation gradually migrate downward beneath the cover layer and continuously accumulate， thereby inducing the cover effect. This effect is the primary cause of engineering distresses such as frost boiling， differential settlement， and salt heave， posing a serious threat to the long-term stability and service safety of engineering structures in cold regions. Therefore， investigating the evolution patterns of water and salt migration in unsaturated sulfate saline soil under the cover effect during freeze-thaw cycles holds significant scientific research value and engineering practical significance. Such investigation is crucial for deeply revealing the mechanisms underlying these engineering distresses and for refining the theoretical framework for preventing and controlling engineering hazards in saline soils in cold regions. To investigate the water and salt migration patterns in unsaturated sulfate saline soil under the cover effect during freeze-thaw cycles， soil column tests were conducted under sealed conditions to analyze the characteristics of water and salt migration within the saline soil subjected to multiple freeze-thaw cycles. Subsequently， based on the principles of heat and mass conservation and incorporating a salt crystallization kinetics model， a coupled water-vapor-heat-salt mathematical model for unsaturated sulfate saline soil considering the cover effect was developed. The numerical solution of this model was obtained using the COMSOL Multiphysics software. By comparing the simulation results with the experimental data from the soil column tests， the reliability of the established model was validated. Then， the numerical model was employed to further reveal the variation patterns of the moisture field， temperature field， and salt field within the soil column during the freeze-thaw process. The results indicated that under the action of freeze-thaw cycles， both the liquid water content and salt concentration in the surface soil layer increased with the number of cycles. By the 10th freeze-thaw cycle， the liquid water content and salt concentration in the surface soil layer at a depth of 0.35 m increased by 2.5% and 5.5 kg·m-³， respectively， compared to their initial values. During the freezing period， the liquid water flux in the surface soil layer was zero， while the liquid water in the underlying unfrozen zone migrated upward. During the thawing period， the liquid water in the surface soil layer migrated downward. During the freezing period， the water vapor flux in the surface soil layer was significantly greater than the liquid water flux， indicating that vapor migration was the dominant mechanism for moisture accumulation in the surface soil. The matric suction of the surface soil layer generally decreased with an increasing number of freeze-thaw cycles， while that of the deeper soil layer increased overall. The matric suction gradient formed between the surface and deeper soil layers served as the key driving force for water and salt migration. Through a combination of laboratory experiments and numerical simulations， this study revealed the patterns of water and salt migration in unsaturated sulfate saline soil under the cover effect during freeze-thaw cycles， and clarified the dominant role of vapor migration and the driving effects of the matric suction gradient. The research findings not only deepen the understanding of the mechanisms underlying engineering distresses in saline soils in seasonally frozen soil regions， but also provide an important theoretical basis and data support for developing effective mitigation measures， such as optimizing drainage systems， installing barrier layers， or regulating the properties of cover materials.
The long-term stability of biochar-stabilized heavy metals in seasonally frozen soil regions faces challenges from freeze-thaw cycles and moisture migration. To evaluate the long-term stability of biochar-stabilized heavy metal-contaminated soil under different freezing conditions， lead （Pb） and cadmium （Cd） co-contaminated soil stabilized with biochar was taken as the research subject. First， open-system unidirectional freeze-thaw tests and unconfined compressive strength tests were conducted， with biochar content （1%， 5%， and 10%） as the variable. The effects of biochar content on the water absorption， frost heave， pH， electrical conductivity， and compressive strength of the stabilized soil were comprehensively analyzed. Based on the results， the optimal biochar content was selected as the specimen parameter for subsequent freeze-thaw cycle tests. Solidified specimens were prepared with the optimal biochar ratio determined from the preliminary tests. With the number of freeze-thaw cycles （0， 1， 3， 5， 7， and 11） as the variable， freeze-thaw cycle tests were conducted under both open-system unidirectional freezing （FTO） and closed-system constant-temperature freezing （FT） conditions. Subsequently， a toxicity characteristic leaching procedure （TCLP） test was performed on the soil after testing to analyze the deterioration effect of freeze-thaw cycles on the stabilized bodies under the respective conditions. X-ray diffraction （XRD） and Fourier transform infrared （FTIR） spectroscopy were used to analyze the changes in the mineral composition of the specimens and the surface functional groups of biochar before and after freeze-thaw cycles. The results showed that biochar effectively suppressed soil frost heave. With increasing biochar content， both frost heave and water absorption of the soil decreased， while the unconfined compressive strength of the stabilized soil first increased and then decreased. The reduction in soil frost heave and water absorption was attributed to the porous structure and hydrophobicity of biochar， which filled soil pores and reduced water infiltration paths. Biochar at low contents improved soil pore connectivity and uniformity， enhanced interparticle bonding forces， and promoted aggregate stability. High biochar content （exceeding 5%） reduced the effective contact area between soil particles， weakened structural continuity， and decreased compactness due to excessive pores， thereby exerting a negative impact on strength. The addition of biochar increased both pH and electrical conductivity of the stabilized bodies. With increasing freeze-thaw cycles， the pH of the stabilized bodies gradually decreased， while the electrical conductivity gradually increased. The inherent alkalinity of biochar led to an increase in the pH of the stabilized bodies. However， intensified oxidation of biochar under frequent freeze-thaw cycles generated more acidic functional groups， further reducing the soil pH. The increase in electrical conductivity could be attributed to the structural damage to the soil caused by freeze-thaw cycles， which enhanced ion mobility and dissolution. The deterioration degree of the biochar-stabilized specimens under the open-system condition （considering a shallow groundwater environment） was more pronounced than under the closed-system condition （without considering a shallow groundwater environment）. After 11 freeze-thaw cycles in the open system， the average TCLP leaching concentrations of Pb and Cd in the biochar-stabilized bodies reached 5.99 mg·L-1 and 1.31 mg·L-1， respectively， which were 178.60% and 125.86% higher than those in the closed system. In the open system， the leaching concentrations of Pb and Cd from the biochar-stabilized bodies partially exceeded the limits specified by the national soil standard after 7 freeze-thaw cycles and fully exceeded the limits after 11 cycles， whereas under the closed-system constant-temperature freezing condition， the limits were not exceeded. Moreover， the TCLP leaching concentrations of Pb²⁺ and Cd²⁺ were positively correlated with the number of freeze-thaw cycles. XRD and FTIR analyses indicated that although biochar was oxidized during freeze-thaw cycles， resulting in increased oxygen-containing functional groups on its surface and enhanced adsorption capacity for heavy metals， the negative impact on the solidification was aggravated by segregation frost heave induced by moisture migration and temperature gradients. This effect intensified with the increasing number of freeze-thaw cycles. Therefore， when evaluating the effectiveness of biochar solidification， full consideration must be given to the deterioration impact of frost heave resulting from moisture migration.
As an important water system in the inland river basin of the Hexi Corridor， the Shiyou River originates from the northern foot of the Qilian Mountains， with numerous glaciers in its upper reaches and a unique geographical environment in its runoff formation area. It serves as a typical basin for exploring the evolution of inland river water resources in cold and arid regions. The river provides essential water resources for industrial and agricultural development in Yumen and other regions in the middle and lower reaches， and plays a critical role in maintaining regional ecological security. This study aims to investigate the evolution patterns of runoff under changing environments， thereby providing theoretical support for the sustainable management of water resources in the Shiyou River. Based on monthly runoff data from the Yumen hydrological station and monthly-scale meteorological data， methods including concentration degree， Mann-Kendall trend test， Pettitt change-point test， wavelet analysis， and Budyko model were employed to analyze the intra-annual distribution characteristics， interannual trends， abrupt changes， and periodic variation of river basin runoff， and to quantify the driving effects of climate and underlying surface changes on runoff. The results showed that the intra-annual distribution of runoff in the Shiyou River became more uniform， the concentration period was delayed， and runoff exhibited a pattern of high flow in summer and low flow in winter. From 1978 to 2024， variations in runoff and precipitation in the Shiyou River differed. Runoff exhibited an insignificant increasing trend at a rate of 0.013×108 m3∙10a-1， while precipitation showed a significant decreasing trend. The abrupt change year for runoff was 2005， after which runoff increased by 25.48%. The abrupt change for precipitation occurred in 1984. The first dominant periods for annual runoff and precipitation were 10 years and 21 years， respectively. Correlation analysis showed that the correlation coefficient of precipitation and runoff before runoff abrupt change was 0.45， and the correlation coefficient after runoff abrupt change was 0.69. The fitting relationship between runoff and precipitation after runoff abrupt change was significantly improved. During the change period of 2005—2024， the elasticity coefficients of runoff to underlying surface parameter， precipitation， and potential evapotranspiration were -1.86， 1.56， and -0.56， respectively. Analysis of the elasticity coefficients indicated that runoff was most sensitive to changes in the underlying surface parameter， followed by precipitation， and least sensitive to potential evapotranspiration. Attribution analysis based on the Budyko model and the double mass curve method yielded consistent results. The contribution rates of underlying surface changes to runoff changes were 126.84% and 110.18%， respectively. In addition， temperature in the Shiyou River Basin showed a significant increasing trend at a rate of 0.31 °C∙10a-1， and glacier area in the upper reaches of the river basin continued to shrink in response to rising temperatures， decreasing from 17.75 km2 in the 1980s to 7.71 km2 in the 2020s—a reduction of 10.04 km2 over 40 years. These findings indicate that changes in the underlying surface are the dominant factor driving the increase in runoff in the Shiyou River. Specifically， these underlying surface changes are manifested as the conversion of grassland to bare land and the reduction of glacier area caused by rising temperatures. Therefore， future water resource management in the river basin should give high priority to the hydrological effects of human activities.
Due to its unique geographical and climatic conditions， the Ningxia-Inner Mongolia reach of the Yellow River is the region most frequently affected by ice flood disasters in China. Accurate prediction of the ice flood peak flow is crucial for ice flood control scheduling and disaster prevention. However， traditional forecasting methods are increasingly limited in applicability due to changes in river channel conditions and human activities. Meanwhile， existing machine learning forecasting studies mostly focus on algorithm optimization， with insufficient analysis of the coupling mechanisms and influence pathways of multiple factors， which restricts model interpretability and forecasting stability. To address these issues， this study focuses on the Toudaoguai hydrological section and couples random forest （RF） with the Decision-Making Trial and Evaluation Laboratory （DEMATEL） method to construct an RF-DEMATEL model. This model systematically identifies key factors influencing the ice flood peak flow and quantifies their interaction intensity and direction. Through path analysis， the coupling mechanism of multiple factors is revealed， and a machine learning forecasting model is established based on the top 80% of variables ranked by importance. The results show that the release of stored channel water （contributing 31.26%） and the incoming flow （contributing 23.83%） are the dominant factors affecting the ice flood peak flow. Air temperature indirectly promotes the formation of the ice flood peak by driving increased incoming flow and channel water release， making it a core variable in the coupling pathway. Compared with the Model-Full model， the Model-6F model constructed based on the selected variables performed better overall in forecasting during the ice periods from 2022 to 2024， indicating that removing redundant factors helps reduce noise interference and improve predictive performance. This study provides an interpretable factor screening framework and modeling pathway for ice flood forecasting， offering important practical significance for enhancing the forecasting capability of ice flood events in the Yellow River.
As a critical component of the cryosphere， snow profoundly alters the surface energy balance， hydrological cycle， and atmospheric circulation patterns. The Qinghai-Xizang Plateau is characterized by extensive lake distribution and abundant snow and ice resources. Snow cover， with its distinct spatial variability， not only regulates the stability of lake ecosystems but may also exert feedback effects on the regional climate by influencing lake water temperature and ice regime. However， the impact of snow cover during the ice-covered period on the heat transfer processes， lake ice phenology， and the evolution of lake ice thickness in alpine lakes remains unclear. This study utilized in-situ observational data from Ngoring Lake， MODIS land surface temperature data and remote sensing imagery， observational data from surrounding meteorological stations， and the China Daily Surface Climate Dataset （V3.0） to conduct a comparative analysis of under-ice water temperature variations and ice regime characteristics under low-snow （2015—2016） and high-snow （2022—2023） conditions. The LAKE2.3 model was employed to investigate the effects of snowfall and its optical parameters. The observational results indicated that under-ice water temperature in Ngoring Lake exhibited an overall increasing trend during the ice-covered period under both low- and high-snow conditions. However， the water temperature， rate of water warming， and average ice growth rate were higher in the low-snow year than in the high-snow year. Additionally， compared with the high-snow year， the onset of ice ablation occurred earlier and the ice-covered period was shorter in the low-snow year. The numerical experiment results demonstrated that the LAKE2.3 model effectively simulated the magnitude and variability of lake water temperature and ice regime characteristics. Snow was identified as the main factor causing the differences in lake water temperature and ice regime between low- and high-snow years in Ngoring Lake. A decrease in either the albedo or the extinction coefficient of snow could lead to an increase in incident radiation at the ice-water interface， thereby causing differences in lake water temperature and ice regime characteristics between low- and high-snow years. Overall， the effect of albedo variation was more significant. This study clarifies the regulatory mechanisms of snow cover and its optical parameters on water temperature and ice regime of lakes on the plateau. This insight not only deepens the understanding of the interaction between the cryosphere and lake ecosystems， but also provides scientific support for ecological conservation and climate strategy formulation on the Qinghai-Xizang Plateau.
Global warming has profoundly altered carbon cycling processes in peatlands of permafrost regions. Against the dual backdrop of climate change and China’s strategic goals of carbon peaking and carbon neutrality， soil organic carbon （SOC） loss from peatlands in permafrost regions have become a critical research focus. However， current studies on SOC in permafrost peatlands of the Da Xing’anling Mountains， Heilongjiang Province， China， remain limited by the lack of high-precision and region-specific SOC databases and predictive models calibrated based on local data. Moreover， research on the coupled carbon-nitrogen-water cycle and its biological regulatory mechanisms in peatland ecosystems remains incomplete， and in situ controlled field experiments remain largely absent. On this basis， this study systematically reviews the spatiotemporal characteristics and primary driving mechanisms of SOC loss from permafrost peatlands in the Da Xing’anling Mountains under climate warming. The results indicate significant spatial heterogeneity in SOC loss across different permafrost regions of the Da Xing’anling Mountains. （1） In continuous permafrost regions， SOC storage is generally high. However， with climate warming causing significant thickening of the active layer and progressive permafrost degradation， SOC mineralization and loss rates have increased significantly. The deepening of the active layer exposes previously frozen carbon pools to microbial decomposition， which may turn these vast carbon reservoirs into enhanced carbon sources. （2） In discontinuous permafrost regions， SOC loss are jointly regulated by air temperature， hydrological dynamics， and vegetation changes. Rising air temperatures directly stimulate microbial activity， while variations in soil moisture critically control anaerobic and aerobic decomposition pathways. Concurrently， changes in vegetation composition and productivity further influence carbon loss. （3） In island permafrost regions， ground temperatures rise more rapidly and permafrost patches are shrinking， and SOC decomposition and lateral transport rates increase significantly. Intense hydraulic erosion and thermokarst development promote both physical migration and biogeochemical processes of SOC. Current studies suggest that this region is particularly vulnerable and may become a substantial carbon source in the future. The temporal variation characteristics of SOC in this region are the core mechanisms controlling its decomposition， mineralization， and migration. Pronounced peaks of carbon dioxide （CO2） and methane （CH4） loss occur during the spring thaw period. Snowmelt and changes in surface hydrodynamics increase the export of dissolved organic carbon （DOC） from peatlands. Freeze-thaw cycles affect soil microbial community structure， thereby affecting SOC stability and degradation rates. Freeze-thaw cycles physically disrupt soil aggregates， release protected SOC， and significantly alter soil microbial community structure and function， ultimately influencing SOC stability and subsequent decomposition processes. Therefore， understanding SOC dynamics across temporal scales is essential. SOC loss are driven by the coupled effects of multiple factors， including temperature， soil moisture variability， vegetation type， microbial processes， and soil physicochemical properties such as texture， cation exchange capacity （CEC）， and mineral composition. The complex interactions among these factors ultimately determine the magnitude and pathways of SOC loss. Future research should prioritize the establishment of long-term monitoring networks， the development of high-resolution datasets， and the construction of mechanistic models integrating hydrological， thermal， and biogeochemical processes. The incorporation of isotope tracing techniques and the strengthening of multi-method experimental integration are also needed to systematically elucidate the coupled carbon-nitrogen-water processes and their microbial regulatory mechanisms in peatlands of permafrost regions， thereby providing a more comprehensive understanding of SOC loss dynamics. It is crucial to conduct more in situ controlled field experiments to predict SOC loss patterns under projected future climatic and hydrological changes. This study aims to provide a scientific basis for carbon loss mitigation and climate response research in high-latitude regions under China’s carbon peaking and carbon neutrality strategy.
Permafrost ecosystems store approximately 1 700 Pg of organic carbon—equivalent to 50% of global subsoil stocks and double the atmospheric inventory—yet are undergoing accelerated degradation on the Qinghai-Xizang Plateau， where warming rates exceed the global average by 40%. Dissolved organic matter （DOM） constitutes a biogeochemically reactive fraction of organic carbon released during permafrost thaw， acting as a critical vector for terrestrial carbon mobilization and microbial processing. The molecular signature of DOM fundamentally controls the potential risk and pathways of carbon release to the atmosphere and hydrosphere. Additionally， permafrost degradation induces significant vegetation succession. These distinct vegetation types exert strong control over DOM formation and transformation dynamics by modulating both litter input characteristics and the soil microenvironment. Although the study of DOM in permafrost is currently a hot topic， a systematic understanding of how vegetation types specifically affect the composition and characteristics of DOM in permafrost is still lacking. This study investigated how vegetation regulated the molecular signature and stability of DOM across four alpine vegetation types （alpine swamp meadow， SM； alpine meadow， AM； alpine steppe， AS； alpine desert， AD） in the Beiluhe basin. Through an integrated analysis of 47 active-layer soil using dissolved organic carbon （DOC） concentrations， ultraviolet-visible （UV-Vis） spectroscopy， and three-dimensional fluorescence excitation-emission matrix spectroscopy coupled with parallel factor analysis （EEM-PARAFAC）， this study identified four fluorescent components， comprising two terrestrial humic-like components （C1， C2）， one common humic-like component （C3）， and one protein-like component （C4）. The results demonstrated that DOM across all ecosystems was microbially dominated but paradoxically enriched in terrestrial humic-like components （SM： 50%； AM： 69%； AS： 50%； AD： 33%）. This phenomenon may result from incomplete microbial degradation of plant residues， generating low-molecular-weight humic substances and microbial metabolites. Soil moisture regulated DOM composition， as evidenced by the unimodal dynamics of the protein-like component C4 along the gradient （SM—AM—AS—AD）. The contribution of C4 decreased from 24% in SM to a nadir of 15% in AM， then peaked at 35% in AD. This pattern pinpointed AM as the key transition zone， corresponding to an anaerobic-to-aerobic shift. Molecular characterization revealed divergent stability pathways. The extreme aridity in AD ecosystems induced a state of biological stasis or “metabolic arrest.” The combination of depleted DOC concentrations and a high proportion of protein-like components （C4=35%） suggested that the DOM pool was more susceptible to rapid microbial mineralization. Furthermore， the low degree of aromatic condensation indicated a labile carbon pool prone to decomposition. The convergence of these characteristics indicated a highly unstable carbon reservoir in AD， suggesting a heightened sensitivity to carbon loss compared to other vegetation types. AM DOM showed plant polysaccharide/aliphatic signatures （highest E₂/E₃=4.88； SUVA₂₅₄=8.32 L·mgC-1·m-1） and optimal humification （HIX=0.85）， promoting organo-mineral associations that enhanced stability. In terms of concentration， composition， and properties， SM closely resembled AM. AS DOM displayed the highest aromaticity （SUVA₂₅₄=12.71 L·mgC-1·m-1）. Carbon stock vulnerability manifested as sharp DOC declines and α₂₅₄ reduction along the vegetation gradient （SM / AM—AS—AD）. A pronounced dichotomy was observed between AM and AD in terms of DOM concentration， properties， and sources. AM represented a stable carbon stock with high storage， whereas AD exhibited high biological lability， minimal carbon storage， and overall instability. Therefore， the shift from AM to AD can be regarded as an indicator of enhanced carbon release. These findings establish vegetation-driven DOM dynamics as key predictors of permafrost carbon-climate feedbacks and call for immediate attention to the changing spatial patterns of vegetation under climate warming on the Qinghai-Xizang Plateau.
The Qinghai-Xizang Plateau is an ideal site for investigating climate convergence effects and material transport cycles. Studying the physicochemical characteristics of its natural surface soil contributes to understanding the climatic and environmental conditions as well as evolutionary processes of this region. Based on X-ray fluorescence （XRF） spectroscopy， this study analyzed the composition and spatial differences of major elements and their compounds in the surface soil of the Qinghai-Xizang Plateau， aiming to reveal the distribution of surface material composition and its implications for dust provenance. The results showed that the major elements in the surface soil of the Qinghai-Xizang Plateau were Si， Al， Ca， Fe， Mg， K， Na， Ti， P， and Mn. Among them， Si， Al， and Ca were the most abundant， accounting for 56.01%， 17.41%， and 10.12%， respectively. The major compounds mainly included SiO2， Al2O3， CaO， Fe2O3， MgO， K2O， Na2O， TiO2， P2O5， and MnO， among which SiO2， Al2O3， and CaO accounted for 62.42%， 16.65%， and 7.64% of the soil compounds， respectively. Spatially， the contents of SiO2 and CaO were higher in the southeastern Qinghai-Xizang Plateau and lower in its northeastern part， while the distribution of Al2O3 was generally more uniform. The surface soil of the Qinghai-Xizang Plateau generally exhibited moderate chemical weathering. The chemical weathering intensity across regions was in the order of south > northeast > southeast， and the average weathering indices （CIA： 69.60， BA： 97.58） were significantly higher than those of the surrounding desert regions and the average values of the upper continental crust （UCC） （CIA： 47.93， BA： 145.24）. This indicated a distinct difference in surface weathering processes between the Qinghai-Xizang Plateau and its surrounding environments， primarily due to its unique climatic environment that led to the formation of distinct pedogenic weathering processes. This study provides new geochemical insights into major elements regarding the intensity of surface chemical weathering and environmental differences on the Qinghai-Xizang Plateau， revealing the potential sources of dust materials in the plateau’s surface soil.
Traditional laboratory tests for determining the shear strength of frozen soil are time-consuming， labor-intensive， and unable to meet the dynamic prediction requirements of cold-region engineering. To address these limitations， this study aims to develop an efficient and interpretable prediction framework for frozen soil shear strength based on ensemble learning techniques. Specifically， four ensemble strategies—Boosting， Bagging， Stacking， and Voting—were employed to integrate multiple base learners and to investigate the effects of different ensemble mechanisms on model performance. Silty sand collected from the Inner Mongolia region was selected as the research material. Through a series of laboratory direct shear tests， a dataset containing 768 samples was established， with freezing time， freezing temperature， moisture content， axial pressure， and shear strength serving as key features. A total of 27 models， including single base models， Bagging ensembles， and Boosting ensembles， were constructed and compared in terms of predictive accuracy. The nine best-performing models were then selected as base learners for constructing Stacking and Voting ensemble models. Model performance was comprehensively evaluated using the coefficient of determination （R 2）， mean absolute error （MAE）， and mean squared error （MSE）， and the influence of each feature was examined through feature importance analysis. The results demonstrate that the Stacking model achieved superior overall performance （R 2=0.998， MAE=1.7091， MSE=6.4233）， outperforming the Voting model （R 2=0.9979， MAE=1.6864， MSE=6.7903）. Both models exhibited consistent feature importance rankings， namely： moisture content > freezing time > freezing temperature > axial pressure. These findings indicate that moisture content and freezing time are the dominant factors controlling the shear strength of frozen soil. In conclusion， this study confirms that an appropriately designed ensemble learning strategy can achieve high-accuracy predictions using a small number of easily measurable parameters. The proposed approach not only reduces the cost and time of traditional testing but also enhances model interpretability. It provides an effective， economical， and scientifically robust method for predicting the shear strength of frozen soil， offering valuable insights for stability assessment and design optimization in cold-region geotechnical engineering.
The direct-shear mechanical characterization of fine-grained frozen soils has long been constrained by insufficient load range， low measurement resolution， and lack of convenient temperature control in conventional small direct shear apparatuses. In this study， a localized temperature-controlled modification was proposed and developed for the shear box assembly of the existing ShearTrac-II high-precision conventional small direct shear apparatus， constructing a high-precision temperature-controlled direct shear system suitable for fine-grained frozen soils. Without altering the original loading frame， automatic control functions， or testing precision， two independent cold-liquid circulation channels were embedded within the loading box and the upper shear box， combined with a composite thermal insulation structure， to construct a temperature-controllable shear-box assembly， and the repeatability and accuracy of specimen temperature control were verified. This addressed the issue that the ShearTrac-II high-precision conventional small direct shear apparatus could not control temperature. Based on this upgraded platform， systematic direct shear tests were conducted on frozen silty clay， enabling the acquisition of more accurate test data. The test conditions covered temperatures of -1.0 to -5.0 °C， water contents of 12% to 26%， and normal stresses of 100 to 700 kPa. The results showed that the silty clay specimens were predominantly strain-softening， and that with decreasing temperature and increasing water content， the shear strength generally increased and the brittleness was enhanced. Further multi-factor analysis of variance （ANOVA） and response surface methodology （RSM） indicated that the quadratic term of temperature contributed the most. Interaction term analysis revealed that a significant composite regulatory effect was observed among temperature， water content， and normal stress， indicating that in practical engineering， attention should be paid to the regulatory mechanism of their synergistic effect on the mechanical properties of frozen soils. Overall， the modified small temperature-controlled direct shear system possesses high load capacity， high measurement resolution， and precise temperature control. It can provide a reliable experimental platform for investigating the low-temperature shear mechanisms of fine-grained frozen soils and for obtaining design parameters for cold-region engineering.
In the context of global climate change， rapid changes in the cryosphere directly affect human survival and development， attracting international attention. As cryosphere elements are mainly distributed in alpine and remote regions， remote sensing technology has become an essential tool for acquiring cryosphere information. Offering cryosphere remote sensing courses is essential for cultivating talent in Earth sciences. This initiative serves not only to systematically cultivate interdisciplinary professionals， but also to broaden students’ global perspectives and foster innovative thinking through the integration of cutting-edge technologies such as satellite remote sensing and numerical modeling. Consequently， it holds important educational value and practical relevance. Offering cryosphere remote sensing courses is crucial for cultivating talent in Earth sciences and thus deserves considerable attention. Northwest China is one of the first regions to introduce such courses and has played a pivotal role in cultivating expertise in this field. The changes in the cryosphere are directly relevant to water resource security in Northwest China， which is vital for regional survival and development. This region is also among the first to introduce cryosphere remote sensing courses， accumulating substantial teaching resources and practical experience. These resources not only serve as valuable references for graduate students and young scholars but also provide a model for other institutions seeking to offer similar specialized courses. This study investigated the current status of cryosphere remote sensing-related programs in Northwest China， analyzed regional characteristics and disciplinary advantages， and examined the status of related course offerings both domestically and internationally. Based on this analysis， the necessity and implementation pathways for a “Cryosphere Remote Sensing” course were discussed. The results revealed that cryosphere remote sensing-related programs were widely distributed in Northwest China with high educational standards， whereas programs in remote sensing science and technology were less common. Overall， the region featured a dense distribution of relevant programs and high educational standards， providing a solid foundation for talent development in cryosphere remote sensing. Northwest China exhibited prominent research capabilities in cryosphere science， supported by leading institutions such as the Northwest Institute of Eco-Environment and Resources， Chinese Academy of Sciences， and Lanzhou University. These institutions produced a considerable number of Chinese-language articles on cryosphere remote sensing and trained numerous high-level national experts， making significant contributions to the advancement of cryosphere science in China. The establishment of the State Key Laboratory of Cryospheric Science and Frozen Soil Engineering， along with multiple national scientific observation stations， provided valuable data sources and practical training venues for course instruction. Furthermore， the books of the cryosphere science series led by Academician Qin Dahe offered a rich repository of teaching materials for cryosphere remote sensing courses. Many domestic and international universities and organizations offered cryosphere remote sensing-related courses in various forms， often as dedicated modules within other courses. While these models offered valuable references， several challenges remained in course implementation， including ambiguous course positioning， uneven distribution of instructors， and repetitive teaching models. To address these problems， this study took the development of the “Cryosphere Remote Sensing” course at Northwest Normal University as an example and proposed specific solutions in terms of course orientation， content design， teaching resources， practical components， and ideological and political education. Additionally， the course increased student engagement by incorporating scientists’ stories， expanding literature reading， and integrating online resources. However， the current course faced challenges such as insufficient student initiative and weak practical components， requiring further improvements by optimizing assessment methods， strengthening experimental teaching， and regularly updating course materials. This study summarizes the experience of developing cryosphere remote sensing courses in Northwest China， offering a valuable reference for more universities. It contributes to the cultivation of interdisciplinary talent in cryosphere science and supports the continued advancement of this critical field.
