The rapid formation， expansion， and outburst of glacial lakes are the prominent consequences of global warming and glacier recession. Large-scale glacial lake mapping， glacial lake outburst flood （GLOF） compilations， and risk assessments are essential for understanding regional environmental changes and implementing effective disaster prevention and management strategies. Centered on the Tibetan Plateau， the Asian Water Tower is one of the major global glacial lake regions and has experienced rapid glacial lake expansion and significant disaster impacts over the past few decades. This study focuses on glacial lake changes， outbursts， and associated risks in the Asian Water Tower and summarizes recent advances in glacial lake research. The results show that： （1） The overall rate of glacial lake area expansion in the Asian Water Tower from 1990 to 2018 ［（31.6±4.0） km²·（5a）^-1］ exceeded the average rate observed during 2018—2022 ［（22.7±8.5） km²·（5a）^-1］. Proglacial lakes， accounting for 83% of the total area increase during 2018—2022， emerge as the primary contributor to the current expansion of glacial lakes in the Asian Water Tower， with changes mainly concentrated in the eastern Himalayas and southeastern Xizang. Under the future ice-free scenarios， a total of 15 826 potential glacial lakes with areas exceeding 0.02 km² are projected to emerge， covering an area of （2 253.95±1 291.29） km² with a water volume of （60.49±28.94） km³. The western glacier-rich region is projected to experience a greater emergence of new and rapidly expanding lakes， although with a time lag compared to eastern regions such as the eastern Himalayas and southeastern Xizang， due to different climate patterns. （2） From 1900 to 2020， 145 credible moraine-dammed GLOFs were recorded in the region， with a marked increase in frequency after the 1980s. Specifically， the average annual number increased from 1.5 GLOFs during 1981—1990 to 2.7 GLOFs during 2011—2020. Spatially， GLOF activity has intensified in southeastern Xizang and China-Nepal border area. Additionally， 183 GLOFs stemming from ice-dammed lakes are documented， mainly concentrated in the Karakoram and western Tianshan Mountains， demonstrating watershed-scale clustering and periodic outbursts. （3） The ongoing expansion and frequent outbursts of glacial lakes have significantly increased disaster risk across the region. A systematic hazard assessment has identified 1 499 glacial lakes classified as high or very high hazard level， potentially affecting a flood inundation area of 6 353 km². In terms of disaster-bearing bodies， approximately 55 808 buildings， 105 existing or planned hydropower projects， 194 km² of farmland， 5 005 km of roads， and 4 038 bridges are vulnerable to potential GLOFs. Approximately 190 000 peoples are directly exposed along GLOF-prone pathways. Subsequently， 85 lakes are classified as very high risk level and 113 as high risk level. The eastern Himalaya and southeastern Xizang exhibit the highest levels of GLOF hazard， exposure， and risk. This situation is exacerbated by the rapid expansion of glacial lakes， the increased frequency of GLOFs， the density of dangerous glacial lakes， and the significant potential disaster volume in the Asian Water Tower. These findings highlight the urgent need for more proactive GLOF prevention and mitigation efforts and the necessity of constructing a regionally coordinated and dynamic disaster management system. Overall， while glacial lake research in the Asian Water Tower has progressed rapidly over the past decade， critical challenges remain. These include the need for standardized and high-precision lake mapping， improved understanding of GLOF-climate relationships， and more refined and watershed-scale risk assessments.

Understanding the impact of snow phenology on vegetation carbon sequestration is crucial for evaluating ecosystem responses to climate change. This is particularly important in arid and semi-arid regions of Xinjiang， where snowmelt serves as one of the region’s major water resources. Based on global daily carbon flux simulation data from 2001 to 2018 and the AVHRR China snow phenology dataset， this study analyzed the spatiotemporal variation characteristics of vegetation carbon sequestration indicators such as gross primary production （GPP） and net primary production （NPP） in Xinjiang， and snow phenology parameters such as snow cover days （SCD）， snow cover start dates （SCS）， and snow cover melt dates （SCM） in Xinjiang using Theil-Sen Median trend analysis， Mann-Kendall test， geodetector， and partial correlation analysis. Additionally， the impact of snow phenology in Xinjiang on the spatial differentiation and temporal variation of vegetation carbon sequestration was explored. The results showed that： （1） From 2001 to 2018， the carbon sequestration of vegetation in Xinjiang showed an overall increasing trend. Among them， from 2001 to 2007， the GPP and NPP of vegetation in Xinjiang decreased at a temporal rate of 9.83 gC⋅m^-2⋅a^-1 and 5.1 gC⋅m^-2⋅a^-1， respectively， and the areas with significant spatial decreasing trends accounted for 38.78% and 36.50%， respectively. From 2007 to 2018， the GPP and NPP of vegetation in Xinjiang increased at a temporal rate of 6.62 gC⋅m^-2⋅a^-1 and 3.26 gC⋅m^-2⋅a^-1， respectively， with areas showing significant spatial increasing trends accounting for 48.05% and 49.84%， respectively. From 2001 to 2018， the SCS in Xinjiang showed a trend of first delaying and then advancing， the SCM showed a trend of first advancing and then delaying， and the SCD showed a trend of first decreasing and then increasing. （2） The results of the geodetector showed that the interaction between any two driving factors had a greater impact on GPP and NPP than a single factor， showing nonlinear enhancement or two-factor enhancement. Among them， the interactions between snow phenology and elevation， vegetation type， and temperature were mainly nonlinear enhancement， while the interaction with precipitation and solar radiation exhibited mainly two-factor enhancement. The interaction between topography and snow phenology had the highest explanatory power for the spatial differentiation of vegetation carbon sequestration. （3） Partial correlation results showed that the response of vegetation carbon sequestration in Xinjiang to snow phenology exhibited a “positive and negative coexistence” characteristic. Overall， vegetation GPP and NPP were mainly negatively correlated with the SCS， with the proportions of pixels showing significant negative correlation being 10.01% and 11.27%， respectively. Vegetation GPP and NPP were mainly positively correlated with the SCM， with the proportions of pixels demonstrating significant positive correlation being 13.73% and 10.86%， respectively. Additionally， vegetation GPP and NPP were primarily positively correlated with the SCD， of which the proportion of pixels with significant positive correlation was 11.14% and 13.35%， respectively. This indicated that an earlier SCS， a delayed SCM， and an increased SCD were more conducive to vegetation growth and carbon absorption. These findings help deepen our understanding of the impact of snow phenology on vegetation carbon sequestration under climate warming. They provide a reference for the evaluation of terrestrial carbon sinks and ecological support capacity， as well as for the formulation of ecologically sustainable development policies， offering a theoretical basis for ecological protection and sustainable development in Xinjiang.

Potentilla parvifolia is a typical alpine shrub widely distributed in the Qilian Mountains. In the context of climate change， it has accelerated its migration trend toward higher altitudes in recent years. Soil microorganisms， as crucial biological communities with transformation potential， are strongly influenced by the root activities of P. parvifolia. Therefore， this study aims to use Illumina Miseq high-throughput sequencing technology to analyze rhizosphere microbial communities and their functional transformation characteristics across different altitudinal habitats and identify the key driving factors， thereby providing an important basis for in-depth investigation of the mechanisms of soil ecological function evolution in alpine regions due to plant migration. The study found that at the low-altitude site （3 204 m）， the coverage of P. parvifolia significantly increased soil total carbon （TC）， available phosphorus （AP）， and nitrate nitrogen （NO₃^--N） contents， while enhancing the activities of sucrase （SUC）， urease （URE）， and cellobiohydrolase （CBH） （P<0.05）. The coverage of P. parvifolia increased the diversity and richness of soil microbial communities， with a more pronounced response observed in fungal communities compared to bacterial communities. Furthermore， P. parvifolia increased the relative abundances of Proteobacteria and Ascomycota. Analysis based on microbial community assembly revealed that stochastic processes dominated bacterial community assembly at low （3 204 m） and middle （3 550 m） altitudes， whereas deterministic processes prevailed at the high-altitude site （3 650 m）. In contrast， fungal community assembly was governed by deterministic processes across all three altitudes. A total of 23 conserved COG functional categories were identified through PICRUSt functional prediction. The coverage of P. parvifolia significantly increased the relative abundance of nitrogen cycle-related functional genes （gudB/rocG， nirK， narH/narY/nxrB） in the soil （P<0.001）， while the abundance of these genes generally decreased with increasing altitude. These findings indicate that P. parvifolia coverage positively affects soil ecological functions at high altitudes in the Qilian Mountains by improving soil physicochemical properties， enhancing key enzyme activities， altering microbial community structure， and regulating functional gene expression. The microbially mediated nitrogen cycling reinforcement may serve as a key driver for the successful migration and niche occupation of P. parvifolia.

Against the backdrop of the ongoing implementation of the Belt and Road Initiative， infrastructure construction in cold regions of China has entered a phase of rapid development. Influenced by the unique climatic environment in cold regions， rock mass engineering faces severe challenges. Diurnal and seasonal freeze-thaw cycles cause repeated ice-water phase transitions in water contained within the pores and fractures of rock masses， generating frost heave stress accompanied by complex moisture migration. This leads to the initiation， propagation， and interconnection of micro-cracks inside the rocks， resulting in significant degradation of their macroscopic mechanical properties. Investigating the evolution mechanisms of mesoscopic structural damage in sandstone under the coupled action of freeze-thaw cycles and mechanical loading holds significant theoretical value and practical engineering implications. Taking freeze-thaw sandstone as the research object， in-situ CT monitoring tests on sandstone under freeze-thaw cycles were systematically conducted. Based on deep learning algorithms， a fully convolutional network （FCN） architecture was integrated with the representative elementary volume （REV） theory to develop a multiscale characterization method linking mesoscopic and macroscopic damage throughout the entire uniaxial compression process of freeze-thaw sandstone. This method accurately extracted the actual mesostructures of internal fractures and their geometric parameters in sandstone. It revealed the controlling mechanisms of freeze-thaw cycles on the anisotropic deterioration of the rock mass， and clarified the cross-scale correlation between micropore reorganization and macroscopic mechanical response. The main contributions and conclusions were as follows. （1） The freeze-thaw rock damage identification algorithm based on the FCN achieved high-precision automatic segmentation of the internal pore （fracture） network in freeze-thaw rocks， enabling quantitative identification of mesoscopic damage. （2） A voxel size of 350×350×350 was selected as the minimum REV. The variations in connected porosity across scanning layers of freeze-thaw sandstone under uniaxial compression were obtained. The sharp increase in internal connected pores of freeze-thaw rocks led to sudden failure under compressive loading. （3） Under identical loading conditions， sandstone subjected to a greater number of freeze-thaw cycles exhibited faster growth in bulk porosity and more rapid internal damage development. The sudden increase in bulk porosity was directly related to the loss of rock strength， serving as a sensitive indicator for predicting failure. （4） A pore-throat network model was established using the maximal ball method. Quantitative analysis of the three-dimensional reconstructed REV of freeze-thaw sandstone indicated that the number of pore-throats inside the rock samples increased significantly under freeze-thaw cycles. Pore-throats were the main pathways for the transmission of frost heave force and damage propagation. The pore-throat system evolved dynamically， transitioning from an increase in frost-induced small pores to an increase in load-assisted medium pores， and finally to large pores before peak stress. （5） The failure process of freeze-thaw rocks under uniaxial compressive loading was essentially the result of the synergistic evolution of pore structure expansion and throat network reorganization. This process induced progressive damage accumulation. Ultimately， dominated by the pore coalescence effect， the percolation channel network formed rapidly， leading to the instability and failure of the rock samples. The coupling action of freeze-thaw cycles and loading profoundly influenced the evolution path of pore structures and failure modes. These findings provide a scientific basis for the stability assessment of rock engineering in cold regions.

The Laptev Sea， as a typical marginal sea of the Arctic Ocean， occupies a pivotal position within the Arctic climate system and marine environment. This region functions as a critical zone for Arctic sea ice formation， freshwater input， and land-ocean heat exchange. Its unique geographical location not only regulates the surface freshwater flux and heat budget of the North Atlantic but also modulates key biogeochemical processes in the Arctic， such as nutrient cycling， primary productivity， and carbon transport. Furthermore， as a strategic hub of the Arctic Northeast Passage， the dynamic characteristics of its sea ice directly determine the navigational potential of the route. Consequently， a comprehensive analysis and systematic review of the multi-dimensional characteristics of sea ice are imperative for fully understanding the mechanisms of sea ice change in the Laptev Sea. This study integrated multi-source observational data to systematically investigate the spatiotemporal evolution of total sea ice area， fast ice area， floating ice area， sea ice thickness， and sea ice age in the Laptev Sea from 1979 to 2024. Additionally， by incorporating model data， this study elucidated and summarized the driving mechanisms of sea ice change in this region and their ecological and environmental effects. The results showed that from 1979 to 2024， the total sea ice area in the Laptev Sea exhibited a significant decreasing trend during the melting season （June-October）， with the largest decline rate observed in October， reaching 0.95×10⁴ km² a^-1. Additionally， the fast ice area exhibited a continuous shrinking trend， with the most pronounced decrease in July， reaching 0.14×10 ⁴ km² a^-1. The fast-ice-free period extended from 78.4 days in the 1980s to 111.2 days in the past decade， and the date of complete ablation advanced by an average of 22.8 days. The floating ice area exhibited a significant decreasing trend during the melting season， whereas it showed an increasing trend during the freezing season （December-January）， reflecting an enhanced conversion from fast ice to floating ice. The floating ice thickness showed thinning trends of 0.13 m per decade in April and 0.23 m per decade in August， and the ice age structure exhibited a trend toward younger ice. Sea ice change in the Laptev Sea was jointly driven by coupled atmosphere-ocean forcing. Among atmospheric factors， offshore wind fields， the Arctic Oscillation （AO）， and the Arctic Dipole （AD） mode influenced sea ice dynamics through momentum and heat transport. At the oceanic level， the enhanced heat flux induced by Atlantification （with a 400% increase in winter over the past two decades） significantly altered the thermodynamic balance during the freezing season. The synergistic effects of meridional winds and warm， humid air masses during the melting season significantly accelerated sea ice ablation， whereas the freezing season was dominated by thermal processes. Due to its unique functions in sea ice export， entrainment of terrestrial sediments， and regulation of permafrost， the Laptev Sea has become a key node connecting the Arctic and the global climate system. The persistent retreat of sea ice in this region laid the foundation for the commercial operation of the Arctic Northeast Passage. Finally， this study summarized the key future research directions for sea ice in the Laptev Sea， including critical scientific issues such as the analysis of sea ice dynamic mechanisms， the feedback effects of sea ice anomalies on the Arctic climate system， and the prediction of route navigability. This study provides theoretical references and directional guidance for interdisciplinary research in the context of rapid Arctic change.

The Qinghai-Xizang Plateau （Tibetan Plateau， TP）， renowned as one of the most distinctive geographical landmarks globally， stands as a critical tipping point within the climate system. As the “Asian Water Tower”， this region is highly sensitive to the effects of climate change and could significantly impact the regional climate patterns. Its influence extends profoundly， affecting the climate of East Asia and even the broader Northern Hemisphere. Therefore， it is imperative to comprehend the historical climate patterns and current dynamics of TP. As a proxy for reconstructing past climates， tree rings have been widely utilized in TP climate reconstruction studies. Dendrochronological works in TP have demonstrated a significant potential for developing long tree-ring chronologies over 1 000 years from living trees， archaeological samples， and in situ timber remnants. These records contain temperature and precipitation variability information at multidecadal scales with perfect annual resolution. While numerous literature reviews have focused on different tree-ring parameters， a comprehensive review of dendroclimatological research on TP is still lacking. Therefore， it is necessary to systematically review the research progress of tree ring climate studies in this region. This study retrieved tree-ring chronology data from 1 436 sampling sites across the TP， outlining the progress of dendroclimatology research since 1990. The analysis revealed that most studies utilized tree-ring width （TRW）， followed by maximum latewood density （MXD）， stable oxygen isotopes （δ¹⁸O）， and stable carbon isotopes （δ¹³C）. TRW studies were widely distributed， encompassing nearly 70 tree species， with the length of the chronologies ranging from 300 to 600 years. MXD studies were concentrated in the Hengduan Mountains， focusing mainly on Picea balfouriana， with shorter chronologies and no millennial records. MXD is predominantly used to reconstruct regional temperatures during the growing or late growing season. Studies of δ¹⁸O are primarily located in the surrounding areas of the Hengduan Mountains， Qilian Mountains， and Himalayas. The studied species is substantial， with the longest δ¹⁸O chronology dating back to 4680 B.C. δ¹⁸O mainly records hydroclimatic signals. Studies of δ¹³C are relatively weak， mainly concentrated near the cities of Linzhi and Chamdo and the surrounding areas of the Qilian Mountains. The species most commonly used for δ¹³C studies are Picea crassifolia and Sabina przewalskii. Most of the δ¹³C chronologies are shorter than 200 years， with the longest chronology reaching 1 171 years. Tree-ring δ¹³C is used to reconstruct changes in temperature and hydroclimatic signals. In the future， the accuracy of climate reconstructions can be improved by extending the length of MXD and isotope chronologies， conducting multi-parameter comprehensive analyses， and refining the detrending methods. Expanding sampling sites， particularly in the western and southern TP， is essential to address these geographic gaps. Future research should also explore the climate responses of shrubs and non-coniferous species to enhance the regional dendroclimatological database. Extending the length of tree-ring density and isotope chronologies—especially in northern and western regions—is critical. Given the current reliance of most studies on a single parameter from a single species for climate reconstruction， multi-species and multi-parameter approaches remain uncommon. Therefore， future research should prioritize integrated analyses， combining diverse dendrochronological indicators with climate models to improve the accuracy and temporal depth of multifactor climate reconstructions. Researchers can gain deeper insights into this region's changing frequency and intensity of extreme climate events by integrating tree-ring records and modern meteorological data and combining tree-ring ecology and physiology studies. This study offers a comprehensive overview of advancements in dendroclimatological research in TP， thus providing readers with a clear understanding of the latest developments and the persistent challenges in these studies. Furthermore， this review also summarizes the shortcomings of existing research and lays a theoretical foundation for future in-depth investigations in this field.

Snow Water Equivalent （SWE） is a critical hydrological variable for assessing the water content stored in snowpacks， particularly in alpine and high-altitude regions like the Qinghai-Xizang Plateau. Given the region’s complex topography， harsh climatic conditions， and the scarcity of in-situ snow measurements， SWE estimation remains a major scientific challenge. This study presents a novel hybrid framework that combines physical modeling and deep learning to simulate daily SWE across the Qinghai-Xizang Plateau， offering a new technical pathway for SWE estimation under data-scarce conditions. The proposed methodology integrates two core models. First， the Factorial Snow Model （FSM）， a physically based process model， is employed to simulate daily snow depth. FSM uses meteorological inputs including air temperature， precipitation， radiation， humidity， wind speed， and pressure to simulate key snowpack processes such as accumulation， compaction， energy exchange， and melt. Second， snow density is estimated using a CNN-BiLSTM-Attention model， which leverages Convolutional Neural Networks （CNN） to extract local spatiotemporal features， Bidirectional Long Short-Term Memory networks （BiLSTM） to capture forward and backward temporal dependencies， and an attention mechanism to dynamically emphasize the most influential features across time steps. Meteorological and snow density data were obtained from ERA5 reanalysis datasets spanning 1979 to 2014. Six key input variables were selected via Pearson correlation analysis： longwave radiation， snowfall， rainfall， temperature， wind speed， and relative humidity. The CNN-BiLSTM-Attention model was trained on data from 1979 to 2003 and tested on data from 2004 to 2014. The model achieved strong predictive performance， with MSE=71.66 kg⋅m^-3， RMSE=8.465 kg⋅m^-3， MAE=6.378 kg⋅m^-3， MAPE=4.556， and R²=0.732， indicating its high accuracy in modeling snow density over long timescales. SWE was calculated by multiplying simulated snow depth from FSM with snow density predicted by the deep learning model. The daily SWE time series from 2006 to 2014 revealed clear seasonal patterns. SWE begins accumulating in October， peaks between December and February， and melts rapidly from March to May. The average daily SWE across the historical period was 0.278 cm， with a maximum of 0.838 cm observed in late December， reflecting the seasonal snow accumulation and melt dynamics typical of the region. The modeled SWE was further validated against two reference datasets： a high-resolution 0.01° SWE dataset and a 0.25° national fused SWE product. Comparisons showed that the proposed model closely tracked observed seasonal and interannual SWE trends， particularly during the critical accumulation and melt periods. It exhibited better agreement with high-resolution data than with coarser products， especially in representing peak values and transitional dynamics. This study introduces an effective and scalable method for SWE estimation in regions lacking dense observational networks. By decoupling the estimation of snow depth and snow density and applying specialized models to each， the framework combines the physical interpretability of FSM with the pattern recognition strength of deep learning. This hybrid modeling approach captures both the mechanistic and statistical characteristics of snowpack evolution， providing a reliable basis for snow resource evaluation. The CNN-BiLSTM-Attention model， which has rarely been applied to snow density modeling before， demonstrated a strong ability to model complex spatiotemporal interactions. When integrated with FSM， it forms a robust and adaptable modeling system that can be generalized to other alpine or cryospheric environments. The results provide valuable support for snow hydrology， water resource planning， and climate change impact assessment on the Qinghai-Xizang Plateau and similar high-mountain regions.

Soil carbon pools are among the largest carbon reservoirs in the Earth’s surface ecosystems and play a key role in the global carbon cycle due to their large storage capacity. This study took the northern Xizang （Tibet Autonomous Region） as the study area and obtained measured data of soil organic carbon stock and physicochemical properties in the 0~30 cm soil layer through field experiments. Combined with meteorological station data in the study area and meteorological and soil data from sampling points from 2014 to 2023， the DNDC model was driven， calibrated， and validated. Subsequently， the model was employed to simulate soil organic carbon stocks across the study area of northern Xizang. Statistical methods including the coefficient of variation， spatial interpolation， correlation analysis， and partial least squares regression analysis were used to analyze the spatiotemporal characteristics of soil organic carbon stocks in different soil layers of northern Xizang， and quantitatively assess the influence of climatic and environmental factors on soil organic carbon stocks. The results showed that： （1） Based on the results of previous studies and the actual data of the grasslands on the Qinghai-Xizang Plateau， the model parameters were iteratively adjusted through repeated experiments. The R² value between the simulated and measured values from the DNDC model was greater than 0.8， with a root-mean-square error （RMSE） of 5.16 and a Nash-Sutcliffe efficiency coefficient （NSE） of 0.61. The localized DNDC model demonstrated strong applicability for simulating soil organic carbon stocks in the grasslands on the Qinghai-Xizang Plateau， though its simulation results required simple correction. （2） On the time scale， from 2014 to 2023， the soil organic carbon stock in the 0~10 cm soil layer in northern Xizang showed a fluctuating decline trend. The soil organic carbon stock in the 10~20 cm soil layer remained relatively stable， with a slight increase in 2023. In the 20~30 cm soil layer， the soil organic carbon stock was stable from 2014 to 2022 but declined significantly in 2023. On the spatial scale， the soil organic carbon stock exhibited a distribution pattern characterized by low values in the central area and high values around the periphery. The soil organic carbon stock was concentrated in the eastern part of the study area， such as Baqen County， Sêni District， and Lhari County， while it was lower in the central and western parts of the study area， such as Nyima County， Xainza County and Gar County. （3） Average precipitation and soil water content were the main factors positively correlated with carbon storage. In contrast， soil bulk density， pH value， sand-to-gravel ratio， and elevation were negatively correlated factors， and the annual average temperature had a weak influence on the soil organic carbon stocks in northern Xizang. The results of partial least squares （PLS） regression showed that elevation， soil bulk density， sand-to-gravel ratio， and soil water content had relatively strong explanatory importance for the dependent variable （soil organic carbon stocks）. This study holds significant implications for grassland management and ecosystem conservation in northern Xizang. In future carbon pool management， attention should be paid to environmental factors， land use， and ecological restoration measures， especially the importance of precipitation and elevation in the dynamic changes of soil organic carbon. This will promote the stabilization or growth of soil organic carbon stocks. This study provides data support for the carbon cycling processes of grassland ecosystems and offers a practical basis for ecological environmental protection and construction in northern Xizang.

Reconstructing glacier retreat from the Last Glacial Maximum （LGM） to the early Holocene using cosmogenic ¹⁰Be exposure dating of moraines is essential for understanding the climatic transition between glacial and interglacial periods. This period， spanning approximately 26.5~10 ka， is marked by major reorganizations in the global climate system， including rapid temperature shifts， fluctuations in atmospheric CO₂， changes in ocean circulation， and variations in orbital forcing. These processes collectively shape ice sheet dynamics and influence cryospheric feedbacks in the Northern Hemisphere. However， the fragmented preservation of moraine sequences at individual sites has hindered efforts to resolve the spatiotemporal patterns of glacier retreat on a continental scale. Single-site records often fail to capture the timing and magnitude of broader deglacial trends. To overcome this limitation， this study compiled a comprehensive dataset of 4 003 published ¹⁰Be exposure ages from 802 well-dated moraines spanning the period between 26.5 and 10 ka. The dataset encompassed a wide range of glacial settings and climates， offering a rare opportunity to conduct integrated regional and hemispheric-scale analyses. The moraines were grouped into four major regions—Asia， Europe， North America， and Greenland—to enable comparative analysis. All ¹⁰Be exposure ages were recalculated using a consistent， globally averaged ¹⁰Be production rate and the LSDn （time-dependent） scaling scheme. This recalibration ensured inter-study comparability and minimized biases caused by inconsistent production rates or scaling methods across different studies. To resolve the temporal distribution of glacier retreat events， the Probabilistic Cosmogenic Age Analysis Tool （P-CAAT） was employed based on probability density estimation （PDE） of exposure ages. This method identified statistically significant clusters of moraine ages by fitting the composite age distribution with multiple Gaussian components， enabling the identification of peak periods of moraine formation or abandonment. Gaussian component analysis was conducted independently for each region and for a combined Northern Hemisphere dataset excluding Greenland. The Greenland moraines were analyzed separately due to preservation patterns heavily constrained by the spatial extent and temporal evolution of the ice sheet margin， differing from other regions. This analytical framework enabled the identification of eight statistically significant hemispheric-scale glacier retreat episodes， centered at （23.9±1.0） ka， （20.5±1.3） ka， （18.3±0.7） ka， （16.6±0.8） ka， （14.7±0.8） ka， （13.0±0.7） ka， （11.4±0.4） ka， and （10.3±0.6） ka. The first two episodes occurring before 19 ka were primarily attributed to increases in Northern Hemisphere summer insolation due to orbital forcing. In contrast， the third through seventh episodes occurred after 19 ka and broadly coincided with abrupt increases in atmospheric CO₂， indicating that greenhouse gas forcing became the dominant driver of glacier retreat. The final retreat episode around 10.3 ka aligned with the peak in summer insolation during the early Holocene. Regional comparison showed a high degree of synchronicity in retreat timing， although the earliest phase （~24 ka） was absent from the Greenland and North American records， likely due to differing responses between ice sheet and mountain glacier systems. Some variability in the number or prominence of retreat episodes across regions could also be attributed to differences in sample density and uncertainties in exposure ages. This study represents one of the most extensive syntheses of ¹⁰Be moraine chronologies in the Northern Hemisphere to date. By integrating a large， globally distributed dataset and applying a standardized analytical workflow， this study improves the temporal resolution of deglacial patterns and provides critical geological constraints for climate-cryosphere interaction models. The findings enhance the understanding of the timing and drivers of glacier retreat and have implications for refining ice sheet reconstructions， glacial isostatic adjustment models， and past sea-level estimation. Moreover， the observed regional differences highlight the importance of glacier type， geographic context， and climate forcing in shaping the heterogeneous responses of ice masses to global climate change.

Dust microparticles， as key components of atmospheric aerosols， have significant impacts on climate change and atmospheric environments. Based on continuous observations from November 2022 to January 2024， this study systematically analyzed the deposition characteristics of dust microparticles in snow and ice， glacial meltwater， as well as atmospheric precipitation and meltwater-fed Mingyong river water in the Mingyong Glacier， Meili Snow Mountain， southeastern Qinghai-Xizang Plateau. The results showed that： （1） The number concentration of microparticles in glacier meltwater runoff had pronounced seasonal variations， and the monsoon season was higher than in the non-monsoon season. （2） During the period of intense glacier melting （May—October）， the number concentration of microparticle in the Mingyong river water showed notable diurnal variations， nighttime concentrations were higher than daytime values. This phenomenon was mainly attributed to a decrease in the rate of glacier ablation at night， which slowed down meltwater runoff and thus prolonged the retention time of suspended particles in the river water. Through the continuous observation of meltwater runoff during day-night， it was found that the peak concentration of microparticles appeared around 20：00 CST， which further proved that there was a responsive relationship between the diurnal variation of microparticles in meltwater runoff and the strong ablation process of the glacier. （3） Fine particles （0.57~2 μm in diameter） dominated the number concentration of microparticles in water bodies. The volume-diameter distribution of microparticles in different water bodies generally exhibited a unimodal pattern， with small median particle sizes， indicating that dust microparticles in this glacier region predominantly originated from long-range atmospheric transport and deposition. This study revealed the deposition characteristics of dust microparticles in snow， ice and water bodies in the Meili Snow Mountain glacier area， providing significant insights for analyzing the mechanisms of the rapid ablation of the cryosphere and its response to regional climate change in the context of global warming.

River is a channel for carbon transport between land and ocean and an important reactor for the metabolism of aquatic ecosystems. It plays an important role in the migration and transformation of carbon in terrestrial and marine aquatic continuums and the carbon budget of watershed ecosystems. The Qinghai-Xizang Plateau is the largest high-altitude cold region in the mid-latitude region. As an Asian water tower， it has developed many major rivers and plays an important role in the regional carbon cycle. In the context of climate change， the process of riverine carbon cycle in the Qinghai-Xizang Plateau has received extensive attention in recent years. This study focuses on the process of river carbon cycle in the Qinghai-Xizang Plateau， and systematically summarizes the progress of riverine carbon cycle in many source areas such as the Yellow River， the Yangtze River， the Lancang River， the Nujiang River and the Yarlung Zangbo River in the past decade. It is found that： （1） The dissolved carbon in the rivers of the Qinghai-Xizang Plateau is dominated by inorganic carbon， and the concentration of dissolved organic carbon is relatively low. The factors such as thawing of frozen soil， enhanced weathering， changes in hydrological processes， and increased erosion and sediment production caused by future climate change will further increase the lateral carbon transport flux in the rivers of the Qinghai-Xizang Plateau. （2） Current observations show that rivers on the Qinghai-Xizang Plateau are important sources of carbon dioxide and methane emissions， and their processes are also affected by climate change and have an increasing trend； （3）The river carbon flux in the Qinghai-Xizang Plateau has an important impact on the carbon budget of the basin ecosystem. The large river carbon horizontal and vertical carbon fluxes offset some terrestrial carbon sinks and are an important part of the ecosystem carbon process. With the warm and humid climate of the Qinghai-Xizang Plateau， its river carbon cycle process will inevitably change significantly， which in turn affects regional carbon sinks， but the extent of its impact is still unclear. In the future， in-situ observation， remote sensing inversion， machine learning and other means should be further used to accurately describe the mechanism of river carbon process， enhance the in-depth understanding of the microscopic mechanism of river carbon cycle， and develop a carbon cycle model suitable for alpine rivers on the Qinghai-Xizang Plateau based on relevant mechanisms， so as to strengthen the simulation and prediction ability of carbon cycle in plateau rivers. This study can provide a scientific basis for further understanding the role of rivers in the carbon neutrality of the Qinghai-Xizang Plateau in the context of changing environment.

Snow depth is a fundamental parameter in hydrology， cryosphere science， weather forecasting， and climate modeling. Accurate monitoring of snow depth is essential for water resource management， natural hazard assessment， and climate change prediction. Passive microwave remote sensing， owing to its strong penetration capability， enables all-weather and all-time observation of the land surface， providing significant advantages for snow depth estimation. The first passive microwave snow depth retrieval algorithm was proposed by Chang et al. in 1987. Since then， numerous snow depth and snow water equivalent products based on passive microwave data have been developed. However， due to differences in retrieval algorithms， results from these snow products often show significant discrepancies. The Fengyun-3 （FY-3） satellite series， China’s first system to provide multi-frequency passive microwave remote sensing data， has played a vital role in improving the autonomy and reliability of climate monitoring. The operational satellites in this series currently include FY-3B， FY-3C， and FY-3D， each of which operates in both descending and ascending orbits. Snow depth retrieval algorithms have been developed using FY-3B and FY-3D microwave brightness temperature data， but their results are inconsistent. Furthermore， the compatibility of other snow depth retrieval algorithms with FY-3 series satellite data requires further investigation and validation. To explore the compatibility of different snow depth algorithms in China， this study applied three typical remote sensing algorithms—KELLY， CHE， and JIANG—to retrieve snow depth from FY-3 MWRI data. The accuracy of these algorithms was evaluated against in situ snow depth measurements from meteorological stations between 2010 and 2019， and the causes of discrepancies were analyzed. The performance and applicability of the KELLY， CHE， and JIANG algorithms were evaluated in three regions—Inner Mongolia-Northeast China， Qinghai-Xizang Plateau， and northern Xinjiang—using root mean square error （RMSE）， bias （Bias）， mean relative error （MRE）， and correlation coefficient （r）. Overall， the KELLY algorithm showed the lowest accuracy， significantly overestimating snow depths between 0 and 60 cm compared to the other two algorithms， with RMSE values ranging from 3.99 cm to 8.23 cm. The CHE and JIANG algorithms demonstrated comparable performance， with RMSEs of 2.78~5.48 cm and 2.88~4.99 cm， respectively. When in situ snow depth exceeded 60 cm， all algorithms tended to underestimate the depth， highlighting a limitation of brightness temperature gradient methods for snow depth retrieval. Regionally， in northern Xinjiang where snow cover was primarily distributed over mountainous terrain， all three algorithms exhibited underestimation in the ascending orbits of FY-3B and FY-3D because of daytime overpass. However， the KELLY algorithm showed relatively smaller underestimation in this region. For other orbits， the KELLY algorithm consistently demonstrated the lowest accuracy across all subregions. In contrast， the CHE and JIANG algorithms demonstrated comparable performance and achieved the highest accuracy in the Inner Mongolia-Northeast China region and the Qinghai-Xizang Plateau. Temporally， the CHE algorithm outperformed the JIANG algorithm during the shallow snow period （November to January）， while the JIANG algorithm outperformed the CHE algorithm during the deep snow period （January to March）. Both the CHE and JIANG algorithms achieved better performance in the Qinghai-Xizang Plateau and Northeast China than in northern Xinjiang. This was attributed to larger interannual snow variability and deep snow causing signal saturation in northern Xinjiang. Overall， the local algorithms （CHE and JIANG） were more suitable for snow depth estimation in China. However， due to variations in snow characteristics， these methods could not fully capture the seasonal patterns of snow depth. Additionally， although cross-platform calibration was conducted to reduce the systematic bias in brightness temperature， snow depth derived from different platforms still showed obvious disparities. In summary， these findings offer valuable insights and technical support for snow depth algorithm improvement and for the reasonable application of passive microwave remote sensing data.

With the continuous expansion of global natural resource exploitation and infrastructure development into cold and high-altitude regions， the stability of geotechnical engineering under extremely low-temperature conditions has become a growing concern within the engineering community. Among the various challenges， frost heave damage is recognized as a critical factor affecting the service performance and operational safety of underground engineering. Its occurrence mechanism is closely related to the pore structure of rock mass and the state of pore water. Although previous studies have confirmed that the mechanical properties of frozen rocks are significantly influenced by temperature， pore size distribution， and water saturation， systematic investigations into the unfrozen water content during the freezing process remain limited. In particular， the mechanisms by which water-ice phase transitions at different pore scales contribute to frost heave damage are still unclear. In this study， low-field nuclear magnetic resonance （NMR） technology was employed as the primary analytical technique. Six representative sandstone samples from diverse geological backgrounds were selected and subjected to temperature-controlled freezing experiments from room temperature to -50 ℃. The dynamic evolution of pore water states was monitored in real time， with a focus on analyzing the variation of transverse relaxation time （T₂ spectra） with temperature， thereby revealing the transformation trends and patterns of unfrozen water and ice content across different pore size ranges. The results showed that all samples exhibited typical bimodal T₂ spectral distribution characteristics， with both peak positions and areas decreasing significantly as temperature dropped， indicating the progressive freezing of pore water. Water in larger pores froze almost completely at around -5 ℃， while in smaller pores， particularly those with diameters less than 0.1 μm， the freezing point was significantly depressed due to pore size effects. As a result， a considerable amount of water remained unfrozen in the form of bound water， even at temperatures below -20 ℃. Moreover， the rate of decrease in unfrozen water content and the freezing behavior varied significantly among the sandstone samples， which was closely related to their pore structure characteristics such as pore size distribution and connectivity. These results underscored the critical role of pore-scale features in governing the controlling phase transition process and the associated frost heave risk. This study not only deepens the understanding of the evolution of pore water states in frozen rocks but also elucidates the microscopic physical mechanisms underlying frost heave damage under low-temperature conditions. The findings provide a theoretical basis for evaluating the stability of rock masses in cold regions. Furthermore， the results offer valuable references for risk identification and structural optimization in the design phase of major cold-region infrastructure projects， including polar railways， highways， tunnels， and hydropower stations， and provide a theoretical basis for the development of materials and technologies for frost damage prevention and control.

High-altitude glacial ecosystems serve as sensitive indicators of global climate change. Within these systems， aquatic microbial communities， acting as core carriers of biodiversity and ecological function， are crucial for maintaining ecosystem stability. Selin Co， the largest glacier-fed endorheic lake on the Qinghai-Xizang Plateau， provides a unique environment that serves as an ideal platform for investigating the ecological patterns of microbial communities under extreme conditions. This study focused on analyzing the diversity patterns and assembly mechanisms of bacterial communities in the waters of Selin Co Lake and assessing their potential response dynamics to environmental changes. Systematic sampling was conducted in June 2022 at the south shore， north shore， and main inflow of Selin Co. Key physicochemical parameters， including pH， total dissolved solids （TDS）， turbidity （TUR）， and nutrients， were measured concurrently. Using 16S rRNA gene high-throughput sequencing， the bacterial community composition and structure were deeply analyzed， yielding a total of 1 327 612 high-quality sequences clustered into 8 587 operational taxonomic units （OTUs）， spanning 28 phyla and 98 genera， thereby revealing exceptionally high taxonomic diversity. Community structure analysis showed that Proteobacteria， Bacteroidota， and Actinomycetota were the absolutely dominant phyla， with Loktanella， Belliella， and Aquiluna as the predominant genera at the genus level. Although α-diversity indices exhibited no statistically significant differences among sampling regions， richness and diversity showed an increasing spatial trend from the south shore to the north shore and inflow， indicating potential influence of local environmental heterogeneity or hydrological inputs. β diversity decomposition （based on Bray-Curtis dissimilarity） demonstrated that differences in bacterial community composition across the regions in Selin Co were primarily driven by species turnover rather than nestedness， reflecting strong environmental filtering or dispersal limitation. Microbial co-occurrence network analysis showed that bacterial taxa were mainly significantly positively correlated， indicating widespread cooperative interactions or niche overlap within the community. This pattern suggested that implementing “a simultaneous multi-region conservation strategy without prioritization” might be more effective for protecting aquatic microbial diversity than prioritizing specific areas. To further analyze community assembly mechanisms， this study applied both the neutral community model （NCM） and checkerboard score （C-score） test. The results showed that deterministic processes predominated in structuring bacterial communities in Selin Co， while stochastic processes played a relatively minor role. Further linear regression modeling of the relationship between bacterial OTU niche breadth and environmental factors identified pH， TDS， and TUR as key drivers significantly influencing bacterial niche breadth. Using multiple community ecological analytical approaches， this study systematically revealed the high diversity， spatial pattern characteristics， and assembly mechanisms of bacterial communities in Selin Co， Qinghai-Xizang Plateau. The findings provide an important case for understanding microbial biogeography in high-altitude glacial lakes， highlighting the central role of environmental selection in shaping microbial communities within extreme aquatic environments. More importantly， the identified key environmental driving factors （pH， TDS， TUR） are highly susceptible to regional climate changes （e.g.， accelerated input of debris from glacial retreat， altered precipitation patterns， and rising water temperatures）. Therefore， the community-environment relationship model established in this study provides a solid scientific foundation for predicting the dynamic responses of microbial community structure and function in Selin Co and similar high-altitude glacial lake ecosystems under future climate change scenarios. This study holds significant scientific value for assessing the impact of global change on vulnerable alpine aquatic ecosystems and for formulating conservation strategies.

Thermokarst lakes are typical representatives of severe degradation of permafrost， where dissolved organic matter （DOM） from permafrost enters thermokarst lakes. Due to intense solar radiation on the Qinghai-Xizang Plateau， DOM in thermokarst lakes undergoes significant photodegradation processes. However， few studies have reported the photodegradation characteristics of DOM in thermokarst lakes on the Qinghai-Xizang Plateau， which may lead to bias in understanding carbon cycle and carbon-climate feedback under permafrost degradation. This study sampled water from thermokarst lakes under four distinct vegetation types on the Qinghai-Xizang Plateau， including alpine wet meadows， alpine meadows， alpine steppes and alpine deserts. For each vegetation type， dark control and light-exposed groups were established to conduct in-situ photodegradation experiments. By measuring dissolved organic carbon （DOC） concentrations， ultraviolet-visible absorption spectra， and three-dimensional fluorescence spectra—coupled with parallel factor analysis （3D-EEM-PARAFAC）， to investigates the effects of solar radiation on DOM content， optical properties， and composition in thermokarst lakes on the Qinghai-Xizang Plateau. Results show that after 10 days of in-situ experiments， DOM in thermokarst lakes under dark conditions exhibited limited changes across all four vegetation types. Although no significant variation in DOM content was observed under light exposure， significant alterations occurred in DOM optical properties and composition under light treatment. Chromophoric dissolved organic matter （CDOM） decreased by 23.6% to 36.7%， indicating that sunlight radiation significantly degrades the CDOM content in DOM. Under the same ultraviolet radiation intensity， the degree of CDOM photodegradation is greater in systems with low DOC concentrations. The decrease in surface water CDOM content may lead to enhanced photodegradation in deep water， resulting in more organic carbon being released into the atmosphere. The aromatic index （SUVA₂₅₄） decreases by 18.9% to 37.1%， indicating that sunlight radiation degrades aromatic compounds in DOM； while the spectral slope ratio （S_R） increases by 45.5% to 124.2%， indicating that sunlight radiation converts high molecular weight DOM to low molecular weight DOM. The significant decrease in humification index （HIX） indicates that solar radiation substantially reduced the humification degree of DOM， while the significant increase in freshness index （BIX） suggests that solar radiation promoted the production of more fresh DOM. These findings are consistent with the decreasing trend of humic-like components and the increasing trend of protein-like components under light exposure. Humic-like substances （C1， C3） exhibit greater photosensitivity than protein-like substances （C2， C4）， with C3 being lost faster than C1， and C2 accumulating faster than C4. Moreover， allochthonous DOM component C3 shows greater photoreactivity than the autochthonous DOM component C2. The study supports the idea that a portion of the protein-like component C4 is a photodegradation product of terrestrial humic component C3. The above findings demonstrate that during the 10-day in-situ observation period， the photochemical mineralization quantum yield of DOC was evidently low. This may be attributed to the relatively static conditions of thermokarst lakes in permafrost regions during summer， combined with prolonged water residence times， which collectively constrain the photomineralization of DOM in these water bodies-otherwise， DOM mineralization would be significantly more pronounced. The photobleaching effect led to a reduction in the absorption coefficient of CDOM， thereby increasing the maximum depth of light penetration in the water column. This phenomenon consequently exerts significant influence on the structure and function of aquatic ecosystems. Solar radiation preferentially degrades terrestrially-derived humic substances in Tibetan Plateau thermokarst lakes， thereby promoting the photodegradation of DOM in these water bodies. These observations collectively indicate that solar radiation plays a crucial role in the migration and transformation of DOM in thermokarst lakes. Under future climate warming and permafrost degradation scenarios， substantial quantities of DOM from thermokarst lakes will be introduced into aquatic ecosystems and exposed to solar radiation， creating favorable conditions for DOM photodegradation and significantly altering its migration and transformation behaviors.

In the context of global warming， the ice and snow in the source regions of the Yangtze River and the Yellow River are melting at an accelerated rate. Quantitatively evaluating the runoff effect of glacier melting is crucial for the management of water resources in high-altitude cold regions of the Qinghai-Xizang Plateau. The meteorological station data （daily precipitation and daily temperature） provided by the National Climate Data Center were used in combination with the 90 m×90 m digital elevation model （SRTM DEM） and the vector data of the second glacier inventory. Subsequently， a degree-day factor model was employed to reconstruct the multi-year glacial mass balance and the historical evolution characteristics of glacial meltwater runoff and its components in the source regions of the Yangtze River and the Yellow River from 1958 to 2022 （including rainfall runoff， snowmelt runoff， and ice melt runoff from glacial areas）. The extent of the impact of climate change on the melting of glaciers in the source regions was explored. The main findings were summarized as follows： （1） over the past 60 years， the glacier mass balance in the source regions of the Yangtze River and the Yellow River showed a significant negative equilibrium. The annual average glacier mass balance was -117.2 mm and -84.3 mm， respectively， and the cumulative mass balance was -7.03 m and -5.48 m， respectively. （2） The glacier equilibrium line altitudes （ELAs） in these two source regions showed a significant upward trend. The upward rates were 5.57 m·a^-1 and 3.93 m·a^-1， respectively， and the ELAs of the source regions of the Yangtze River and the Yellow River increased by 334.2 m and 313.6 m， respectively. （3） The total runoff of glacial meltwater in these source regions generally showed an increasing trend. The multi-year average total runoff of meltwater in the two source regions was 18.43×10⁸ m³ and 1.87×10⁸ m³， respectively. The variation trend of the ice melt runoff in the source regions was consistent with that of the total meltwater runoff， and its proportion showed an increasing trend year by year. The proportions of ice melt runoff in summer reached 91.88% and 90.95%， respectively. Snowmelt runoff showed a slight increase in the source region of the Yangtze River and a slight downward trend in the source region of the Yellow River. The seasonal distribution characteristics indicated that summer （June to August） was the primary period for glacial meltwater runoff， with its runoff volume accounting for 90.05% and 88.23% of the total annual runoff volume of the Yangtze River source and the Yellow River source， respectively. The proportions of runoff in spring and autumn decreased significantly. The spring runoff volumes of the Yangtze River source and the Yellow River source were 3.79% and 1.95%， respectively， and those in autumn were 6.17% and 9.82%， respectively. In winter （December to February）， there was basically no runoff generation. （4） The sensitivity of glacier mass balance to temperature was much higher than that to precipitation. The sensitivity of glaciers in the source region of the Yellow River to climate change was higher than that in the source region of the Yangtze River， which was related to the scale of glaciers in the source regions. In conclusion， this study systematically analyzes the variation patterns of glacial mass balance and meltwater runoff components under climate change， and quantifies the contribution of glacial ablation to streamflow. The findings provide critical insights into the implications of cryospheric changes for water security.

In the fields of national defense and civil engineering， such as the explosion-proof structure design of tunnels， railways， highways， and pipeline networks in cold regions， and the vertical shaft excavation using the freezing method in coal mines， frozen soil is often subjected to impact loading during activities like drilling and blasting construction， weapon damage， and seismic events. Investigating the strength characteristics， deformation and failure mechanisms， and stress wave propagation patterns of frozen soil within a high strain rate range is of great theoretical and practical significance for improving the efficiency of frozen soil excavation and fragmentation and for analyzing the safety and stability of frozen soil masses. This study summarizes the research status of the dynamic characteristics of frozen soil under impact loading from three aspects： frozen soil SHPB test system and data processing， laboratory SHPB tests on frozen soil， and dynamic constitutive relationship of frozen soil. First， the advantages and disadvantages of the developed temperature-controlled SHPB test system for frozen soil are analyzed. It is found that the current temperature control system has disadvantages such as large temperature fluctuations， cumbersome test process， and low refrigerant utilization efficiency. Second， the effects of parameters such as temperature， strain rate， stress state， moisture content， and fracture distribution on the dynamic strength， deformation modulus， and failure characteristics of frozen soil are systematically summarized. It is found that frozen soil exhibits typical characteristics of “freezing brittleness” and “dynamic brittleness”. Confining pressure and axial pressure help enhance the dynamic compressive strength of frozen soil. The presence of prefabricated fissures significantly reduces the bearing capacity of frozen soil specimens. Finally， the methods for establishing constitutive models of frozen soil and their advantages and disadvantages are summarized and evaluated. It is concluded that the Z-W-T model can better characterize the relationship between the strength and deformation of frozen soil under impact loading. Based on the summary of existing research， prospects are provided for the urgent problems to be solved and future research directions in frozen soil dynamics.

The accelerated glacier melting under the intensification of global warming has significantly altered glacial erosion and weathering processes within the cryosphere， with unprecedented enhancement observed in recent decades. In warm glacier catchments， these enhanced processes may exert substantial impacts on the global carbon balance and climate change. Specifically， chemical weathering regulates the carbon cycle by consuming atmospheric CO₂， playing a crucial regulatory role in global climate change over geological timescales. However， with the accelerated melting of glaciers， the chemical weathering rate may change， and its specific impact on the carbon cycle requires further investigation. Mingyong Glacier catchment， located on the southeastern margin of the Qinghai-Xizang Plateau， was selected as the research area. A two-year monitoring campaign （from October 2018 to October 2020） of river water hydrological indicators and daily sampling was conducted， ultimately acquiring 731 river water samples. Multiple analytical approaches were employed， including mass balance calculations， Gibbs diagram analysis， mathematical statistics， and hydrochemical characterization. The hydrochemical characteristics of the river water in Mingyong Glacier catchment were investigated， and the chemical weathering rate of rocks and the carbon sink/source rate in the catchment were quantified. The results showed that river water exhibited significant seasonal variations in Mingyong Glacier catchment， with higher anion and cation concentrations in river water during the non-monsoon periods and lower values during the monsoon periods； the water hydrochemical type was classified as HCO₃-Ca. The anions and cations were predominantly derived from carbonate rock weathering， with participation of sulfuric acid in carbonate weathering contributing the largest proportion （62.1%） to chemical composition in river water. Other sources include carbonic acid in carbonate rock weathering （32.4%）， silicate weathering （4.5%）， and atmospheric deposition （1.0%）. Quantitatively， the silicate weathering rate and atmospheric CO₂ consumption flux averaged 0.16 t·km^-2·a^-1 and 0.31×10³ mol·km^-2·a^-1， respectively. In contrast， sulfuric acid-driven carbonate weathering exhibited higher rates （3.34 t·km^-2·a^-1） and corresponding CO₂ release （4.00×10³ mol·km^-2·a^-1）， yielding a net carbon source flux of 3.70×10³ mol·km^-2·a^-1. This indicated that CO₂ release rates through chemical weathering in the study area exceeded its consumption rates by up to 13 times， exhibiting pronounced seasonal variability with monsoon periods showing significantly higher values than non-monsoon periods. Further analysis of the influencing factors for the above characteristic changes revealed that runoff was identified as the primary controlling factor affecting chemical weathering rates in the Mingyong Glacier catchment， with air temperature exerting greater impact than precipitation. The carbon fluxes in the Mingyong Glacier catchment were directly controlled by chemical weathering rates， while runoff and rainfall indirectly controlled these fluxes by modulating the weathering processes. In conclusion， chemical weathering in warm glacier catchments plays a pivotal role in regulating both local hydrochemical characteristics and ecosystems， as well as global carbon balance and climate change. Consequently， research in this field holds substantial scientific and social value， and deserves widespread attention. Future research should focus on the interaction mechanism between chemical weathering process and climate change to provide a scientific basis and decision-making support for addressing global climate change.

In the context of global climate change， the persistent reduction of Arctic sea ice has created more favorable conditions for the opening of Arctic shipping routes. The Northeast Passage， in particular， has garnered the interest of shipping companies due to its potential distance and cost benefits. However， the passage presents significant navigational safety challenges stemming from its complex and variable climatic and sea ice conditions. Therefore， there is an urgent demand for an intelligent path planning method to optimize the utilization of Arctic routes. This study introduces an intelligent route planning approach that integrates the Polar Operational Limit Assessment Risk Indexing System （POLARIS） with deep reinforcement learning. POLARIS assesses navigational risks by evaluating sea ice conditions along potential routes. The researchers integrate POLARIS with traditional A* algorithms and Deep Reinforcement Learning （DRL） to enhance the path planning process. This integration is crucial as it improves the capacity to manage dynamic environments—typical of Arctic conditions—more effectively than static algorithms such as A*. Experimental results indicate that DRL significantly surpasses the traditional A* algorithm in computational efficiency， achieving approximately 50 times faster processing. This efficiency is vital for real-time route planning， necessary to adapt to the rapidly changing Arctic environment. The article explores the mechanics of DRL， elucidating its superiority in managing complex and dynamic conditions. Unlike traditional methods that struggle with large state spaces and require predefined heuristic functions， DRL employs neural networks to learn optimal strategies through trial and error， making it adept at navigating the unpredictable Arctic environment. The study highlights DRL’s capacity for learning and adaptation， providing a practical solution for real-time decision-making in shipping route planning. Moreover， the study offers a detailed explanation of the employed methodologies. It categorizes DRL approaches into value-based， policy-based， and actor-critic methods， with the research selecting a value-based approach. The Deep Q-Network （DQN） algorithm is employed for its ability to handle large state spaces and efficiently learn optimal policies. The research describes the environmental setup， where Arctic sea ice and weather data are modeled on a grid to simulate navigational scenarios. The DQN model is trained on this data to predict optimal routes， taking into account navigational risk and efficiency. The researchers conducted case studies using historical sea ice data to validate the model， comparing the predicted routes against actual shipping paths taken by vessels such as the ship “Yong Sheng” in 2013. The results show a strong alignment， with the model’s predictions reflecting practical and safe routes that avoid areas of high ice concentration， akin to decisions made manually by ship captains. In conclusion， the article establishes that DRL serves as a suitable core algorithm for Arctic route planning systems. It provides a robust and efficient solution for managing the challenging and variable conditions of the Arctic， facilitating the development of intelligent， automated route planning systems. The study underscores the importance of integrating advanced AI techniques like DRL into maritime navigation to enhance safety， efficiency， and adaptability， thereby contributing to the development of the Arctic as a viable international shipping corridor. The findings suggest a promising future for automated decision-making in maritime logistics， particularly in regions where environmental conditions are critical and volatile.

The chain reaction of glacial lake outburst floods in the alpine and high altitude mountains area poses a significant threat due to its extensive impact range， large flood scale， and high flow velocity. Traditional prevention and control measures downstream of the disaster chain are often difficult to implement and economically unfeasible. Moreover， the primary triggering factors， such as ice collapses， are usually located at altitudes above 5，000 m， making access and engineering interventions extremely challenging. To address these issues， this study proposes a novel prevention and control approach focusing on the overflow outlet of the moraine dam， based on field investigations and statistical analysis. A new integrated protective system， combining rigid and flexible reinforcement layers， is developed to enhance the stability of the moraine dam’s armor layer at the overflow section. The system consists of scaffold steel pipes， high-strength metal flexible nets， and anchor rods， forming a “rigid-flexible integrated plate-like” protective structure. Laboratory physical model experiments were conducted to verify its effectiveness under simulated flood conditions. The results indicate that as protection measures are strengthened， the initiation flow velocity of the armor layer increases significantly. Compared to the unprotected state， a single-row rigid connection provides a minimal increase in stability， whereas a double-row rigid connection increases the initiation flow velocity by approximately 2%， a triple-row rigid connection by about 5%， and the combined rigid-flexible system by approximately 8%. The effectiveness of the protection system improves with an increase in the protected area， demonstrating its potential for mitigating moraine dam breaches. This research provides a feasible and technically viable solution for preventing moraine dam failures in high-altitude， cold environments. The proposed system offers advantages in terms of adaptability， cost-effectiveness， and ease of implementation in remote mountainous regions. It has the potential to serve as an effective first line of defense against glacial lake outburst floods and contribute to disaster risk reduction in alpine and high-altitude areas.

Slope instability and collapse due to freeze-thaw cycles are significant challenges in infrastructure development in permafrost regions. To explore and analyze the causes and stability variation patterns of slope instability in seasonally frozen soil areas in Southwest China， this study focuses on the slopes of the Batang region. A research approach combining theoretical analysis， numerical simulation， and laboratory experiments is employed. A permeability coefficient model for the negative temperature zone， considering the unfrozen water and pore ice states， is applied within a soil hydrothermal coupling model. Through direct shear tests， the variation patterns of shear strength parameters for remolded slope soils were further obtained. Using the COMSOL finite element software， a customized slope stability model was developed. The slope height was set to 10 m with a gradient of 1∶1.5， and both the slope top and bottom extended 50 m and 60 m， respectively， with the slope base extending 10 m downward. This model was used to analyze the changes in temperature and seepage fields， as well as the slope instability mechanisms induced by freeze-thaw cycles affecting the soil’s shear strength parameters. Results show that the modified hydrothermal model， which incorporates the adjusted permeability coefficient， has higher accuracy， as verified by comparing the calculated values with those from one-dimensional soil column experiments. Temperature fluctuations within 1 m of the slope surface are significant. From January to March， air temperatures remain low， with deeper soil temperatures being relatively high， and soil temperature increases monotonically with depth. Between April and September， surface soil temperatures rise faster than the deeper layers， which maintain relatively high temperatures. From October to December， surface temperatures decrease more rapidly than in deeper layers， where temperatures remain relatively stable. The seepage field shows minimal variation beyond 2.5 m depth. From November to March， the slope surface freezes first， leading to a decrease in water content， although internal moisture migration is limited. Water content initially decreases with depth， then increases. Between April and June， during the thawing period， shallow soil melts， with water content decreasing with depth. From July to October， the water content initially increases with depth before decreasing again. In March， water content variation with depth is most pronounced， and due to the surface thawing while the subsurface remains frozen， excess water above the frozen layer cannot drain effectively， resulting in the lowest slope stability with a safety factor of only 1.59. The mechanism of slope failure in seasonally frozen regions is attributed to saturation above the thaw-freeze interface during the thawing period， where water cannot drain， reducing soil friction and leading to slope collapse. By dynamically adjusting the soil strength parameters to calculate slope stability， it is determined that the most significant strength degradation occurs at the thaw-freeze interface， primarily resulting in shallow slides within 1.2 m of the surface. The sliding surface is relatively flat and arc-shaped， with the primary failure mode being slumping collapse. This study provides valuable insights for slope design and disaster prevention in Southwest China， particularly for engineering projects in seasonally frozen regions， offering significant practical implications.

The Qinghai-Xizang Plateau experiences frequent geological disasters due to its unique geographical environment and intense tectonic activity. Among these， disaster chains such as glacial lake outburst floods triggered by landslides and collapses are particularly prominent. Therefore， systematic research on the formation mechanisms of such disaster chains holds significant theoretical and practical value for regional disaster prevention and mitigation. This study focused on the landslide on the north side of Tangzhen Co in Damxung County on the Qinghai-Xizang Plateau. Based on detailed field geological surveys and remote sensing interpretation， the basic characteristics， deformation-failure mechanisms， and main influencing factors of the landslide were comprehensively analyzed. Using the Massflow numerical simulation platform and considering the actual conditions of the disaster chain， the calculation program was secondarily developed. This enabled dynamic simulation and quantitative prediction of the entire process from landslide initiation， movement into the lake， glacial lake outburst， to flood propagation. The study area is located on the northwestern margin of the Ningzhong Basin within the Yangbajain-Damxung-Gulu graben system in the hinterland of the Qinghai-Xizang Plateau. The region is tectonically active， with frequent occurrences of collapse and landslide disasters. The landslide on the north side of Tangzhen Co is an ancient landslide. In plan view， it exhibits an “armchair” shape， with an average slope gradient of about 30° and a main sliding direction of 185°. The sliding body has an average thickness of approximately 11 m and a total volume of about 11.55×10⁴ m³， classifying it as a medium-sized landslide. The landslide material is mainly glacial till， and the underlying bedrock consists of feldspathic quartz sandstone. Investigations show that the landslide is currently in an accelerated creep stage. The main cracks at the rear and on both sides are connected， shear feather cracks have formed on both flanks， and bulging cracks and discontinuous radial cracks appear at the front. The overall stress environment of the slope is gravitational stress. Due to stress redistribution， the maximum principal stress is nearly parallel to the slope surface， forming a tensile stress concentration zone in the middle-upper part of the slope and a shear stress concentration zone at the toe. This pattern manifests as “tension at the rear and shear at the front”， leading to a creep-tension deformation and failure mode of the landslide. The stability of the landslide is jointly controlled by gravitational stress， rainfall infiltration， freeze-thaw cycles， fault activity， and groundwater. It is prone to overall instability under extreme rainfall or seismic conditions. Once the landslide fails， it will further entrain slope deposits， mobilizing a volume far greater than that of the landslide itself. Upon entering the lake， it will generate surge waves. The surge waves may overtop the existing spillway and flow downstream， forming floods or even debris flows. Additionally， a large flood may remobilize existing loose deposits in the downstream channel， thereby forming larger debris flows， posing a severe threat to the downstream Qucai Village. This study employed a depth-integrated continuum mechanics model and the MacCormack-TVD finite difference algorithm to numerically simulate the disaster chain in segments. First， the Coulomb friction model was used to simulate landslide motion and entrainment. Subsequently， the model was switched to the Manning model to simulate the outburst flood propagation process. Key parameters were set based on experimental and back-analysis results from similar landslides in the region to ensure that their values conformed as closely as possible to the conditions of the actual disaster chain. The simulation results showed that the entire landslide movement process lasted about 200 s， reaching a maximum velocity of 39 m·s^-1. After 800 s， the flood reached the downstream Qucai Village， with a maximum flow velocity of 11 m·s^-1 and a maximum inundation depth of 4.2 m， ultimately inundating about 80% of the village area. By integrating geological mechanism analysis with dynamic process simulation， this study systematically reveals the disaster-causing mechanisms and spatiotemporal evolution patterns of the Tangzhen Co landslide-glacial lake disaster chain， providing quantitative predictions of motion parameters and risk zones. The proposed “sliding-entrainment-surge-dam breach” chain disaster process and parameterized simulation method offer essential references for early warning and engineering prevention and mitigation of the landslide on the north side of Tangzhen Co and other similar landslide-glacial lake disaster chains on the Qinghai-Xizang Plateau.

With global warming and increasing human activities， slope geohazards such as landslides， thaw collapse， and gelifluctions have become increasingly frequent on the Qinghai-Xizang Plateau. A large number of slope geohazards have developed along the Gongyu High-grade Highway on the eastern edge of the Qinghai-Xizang Plateau， posing serious threats to infrastructure construction and safe operation in the region. However， the distribution patterns and factors influencing unstable slopes along the entire highway remain unclear， which severely constrains the construction and development of highways. Using IPTA-InSAR technology， this study conducted surface deformation monitoring， early-stage identification of unstable slopes， and field validation along the Gongyu High-grade Highway. Then， the spatial distribution and types of unstable slopes were summarized， and their influencing factors of deformation were analyzed. The results showed that the surface deformation rate along the radar line-of-sight direction in the study area ranged from -249 to 335 mm·a^-¹. The overall area remained relatively stable， with significant surface uplift observed near the Duogerong Basin and Bayan Har Mountains. Based on the InSAR deformation monitoring results， a total of 974 unstable slopes were identified. Unstable slopes were widely distributed in the study area， mainly concentrated in the Heka Mountain to Wenquanxiang segment and Qingshuihe County to Yushu segment， with minor distributions near Maduo County and Yeniugou. Types of unstable slopes included creep， colluvial landslides， and bedrock landslides. Among them， one large-scale bedrock landslide was located along the Qinggenhe in Wenquanxiang. The development of this landslide was structurally controlled. Under the influence of rainfall and river erosion， the slope experienced sliding. In the future， it may block the Qinggen River， potentially triggering disaster chain effects such as barrier lake and dam break. One colluvial landslide was located in the mountainous area of Heka Mountain， mainly influenced by rainfall and temperature. The persistent and cyclic freeze-thaw actions caused extensive cracks on the slope surface. Precipitation infiltrated through cracks， altering the thermal state at the base of the colluvium and promoting ground-ice melting on the bedrock interface. This process formed a sliding surface. Meltwater and rainwater further eroded and transported the colluvium， triggering sliding. Additionally， the sliding of the lower gully channel at the slope’s toe exerted a retrogressive pull， causing the middle section of the slope to slide. There were 972 creep slopes distributed along the entire highway. One typical creep slope was located near the Heka Mountain. This slope was located in a small watershed catchment area with consistently high water or ice content. Under the influence of road construction and excavation disturbances， the underground ice layer was exposed and gradually melted. As the underlying materials became saturated with water， the soil structure was damaged， leading to an increase in pore water pressure and a reduction in shear strength. This resulted in the formation of a weak sliding surface， triggering slope failure. In the future， this landslide may gradually evolve into other types of geohazards. The deformation of unstable slopes in the study area is primarily controlled by the loose soil characteristics of grassland and desert areas. Increased rainfall and temperature variations serve as the direct driving factors for slope deformation. This study reveals the development characteristics and influencing factors of slope geohazards along the Gongyu High-grade Highway， providing scientific support for risk prevention， hazard control， and the safe operation of the Gongyu High-grade Highway.

The changes in Qilian Mountain glaciers are not only a direct reflection of climate change， but also has an important impact on fresh water resources. In recent years， the glaciers in Qilian Mountain have been in a state of retreat and instability， to deeply understand the response of the glaciers in the Qilian Mountain to climate change in the 21st century and the magnitude of the change， clarify the characteristics of the Qilian Mountain glaciers change， the process and the reasons， and further predict the trend and magnitude of glacier change， and analyze the Qilian Mountain glaciers destabilization and imbalance in the climate warming background of the mechanism， we selected the Qiyi Glacier， which is strongly influenced by the westerllies， and the Lenglongling No.2 Glacier， which is strongly influenced by the monsoon as examples by analyzing remote sensing images and numerical simulation， it is found that the two glaciers show obvious retreat at the end of the glacier in 1990 to 2022， with the area reduced by 10.28% and 19.04% respectively， and the retreat rate of Lenglongling No.2 Glacier is obviously larger than that of Qiyi Glacier， which is attributed to the fact that the eastern part of the Qilian Mountain is warmer than the western part， and the eastern part of the glacier has decreased in precipitation. To explore the future changes of the two glaciers， the OGGM model considering glacier flow was driven by CMIP6 model data.The relative errors were within 2.5% by comparing the glacier area of the two glaciers simulated by the model and the visually interpreted area from 2008 to 2022. The simulated values of the mass balance of the Qiyi Glacier were more in line with the observed values， with an average difference of 88 mm w.e. Between the two mass balances， the difference between the two is 88 mm w.e.， which indicates that the OGGM model can simulate the glacier changes better.The simulation results show that under the high-emission scenario （SSP5-8.5）， the area and volume of the Qiyi Glacier will decrease by 77% and 94%， and the area and volume of the Lenglongling No.2 Glacier will decrease by 93% and 99% by 2060， and by the 21st century 80s， the air temperature in the area of the Qiyi Glacier will increase by 4.1 ℃ compared with that of the period of 1990 to 2015， while the increase in precipitation will not be obvious， and the glacier will be completely extinguished by that time. In contrast， temperatures in the eastern section of the Qilian Mountain will rise even faster， and it is predicted that the Lenglongling No.2 Glacier will be completely extinct around 2075. Even under the most optimal carbon emission scenario （SSP1-2.6）， a temperature increase of 1.0 ℃ in the Qilian Mountain by 2050 compared with the 1990 to 2015 prognosis will lead to the retreat of both glaciers into ice buckets by 2080 when the area and volume of Qiyi Glacier will be only 22% and 8% of that of 2020， and the Lenglongling No.2 Glacier will be only 17% and 6%. The rapid ablation of glaciers will cause changes in glacier runoff， and the runoff of the Qiyi Glacier and the Lenglongling No.2 Glacier will peak in 2035 and 2024， respectively， and the runoff of the two glaciers will be reduced by 28.98% and 41.82% by the end of the century under the SSP5-8.5 scenario， respectively. Therefore， regardless of climate scenarios， Qilian Mountain glaciers will retreat significantly in the 21st century， or even disappear， due the temperature， precipitation atmospheric circulation， and other influences， which will also make the eastern Qilian Mountain glaciers than the western glaciers retreat faster. Clarifying the trend and magnitude of glacier changes， and understanding the process mechanism of Qilian Mountain glaciers imbalance and destabilization in the context of climate change will deepen our understanding of the current and future glacier changes in the Qilian Mountain region， so as to cope with the environmental and water resource problems caused by the glacier changes， which require us to plan in advance.

Frozen soil is widely distributed in Northwest China， and its physicochemical properties strongly affect the stability and durability of engineering structures in cold regions. It is highly sensitive to ambient temperature， often exhibiting significant differences before and after freezing. Therefore， evaluating the properties of frozen soil is a prerequisite for the remediation of frozen soil-related engineering hazards. Electrochemical impedance spectroscopy （EIS）， a non-destructive method， is widely used to evaluate the physical and chemical properties of porous geotechnical materials and to characterize the phase transition of pore water in frozen soil. As the pore water freezes or the ice melts with the fluctuations in the surrounding temperature， it is crucial to investigate the influence of pore water state on the EIS of soil. To investigate the variations in the electrochemical characteristics of soil during freezing， the EIS of silty clay in Qinghai-Xizang Plateau was measured under different moisture contents and temperatures. The field-collected soil samples were first leached to remove salts， preventing the initial salt content from affecting the test results. Subsequently， the salt-free soil was dried in the oven at 105 ℃ for 12 hours. The dried soil was crushed and sieved to obtain the test soil. Considering the test soil properties， four moisture contents were set as 10%， 15%， 20%， and 25%， respectively. The soil specimens were then compacted into cubes with a side length of 7 cm， and two copper sheets with a smooth surface were placed on its both sides as the two side electrodes for EIS measurement. The soil temperature was controlled by a cooling bath （TMS8035-R40） at a temperature range of 30 ~ -30 ℃， and a stepwise cooling method was adopted for temperature reduction. When the soil temperature was stable， the electrochemical test was conducted with CS353 AC impedance tester， and the frequency range was set from 10^-2 to 10⁵ Hz. The DC potential was 0.05 V， and the AC amplitude was 10 mV. Finally， ZSimpWin software was used to analyze the measured data. The results showed that the mobility of ions in pore water slowed down as the temperature decreased， leading to an increase in the soil impedance value and a gradual expansion of the capacitive reactance arc radius. The pore water was basically frozen at -30 ℃. However， due to the presence of residual ions in the pore water， ion migration occurred in the soil with 10% moisture content， resulting in a diffusion phenomenon， which was manifested as a diagonal line close to 45° in the low-frequency region in the Nyquist plot. The impedance modulus tended to stabilize at a frequency of 10⁵ Hz， and the impedance modulus at this frequency was selected for further analysis. Under the positive temperature condition， the impedance modulus increased linearly with decreasing temperature. The pore water froze at 0 ℃， and the formation of ice crystals was accompanied by volume expansion， causing changes in the internal structure of the soil and a significant increase in the impedance modulus values. In addition， an equivalent circuit model was established by analyzing the conductive pathways within the soil. The EIS data were fitted using ZSimpWin software with good fitting results， obtaining the changes in equivalent components of soil with different moisture contents during the cooling process. Taking the equivalent resistance element R₁ as an example， it can be concluded that the numerical change of the equivalent resistance element can effectively reflect the freezing process of pore water in the soil. This study transforms the analysis of the electrochemical characteristics of the soil from a qualitative approach to a quantitative one， which holds significant importance for understanding the electrochemical characteristics of soil during the freezing process.

Though rare， snowstorm in western South Xinjiang frequently cause severe damage to local economies and livelihoods， leading to substantial losses. To gain a more comprehensive understanding of its formation mechanisms and enhance operational capabilities， this study， using NCEP/NCAR reanalysis data， employed the Eulerian method and the HYSPLIT （hybrid single particle Lagrangian integrated trajectory） model to conduct a detailed analysis of the circulation background and the sources， transport， and contributions of water vapor from different regions during the 15 snowstorm days that occurred in western part of South Xinjiang from 1980 to 2022. The results showed that the primary influencing system on snowstorm days in this region was the Central Asian vortex （trough） type， with the water vapor mainly sourced from the Red Sea and its coastal areas—the Persian Gulf， Iran， Afghanistan， the key water vapor region—to the snowstorm areas. Before the snowstorm， the western boundary contributed the most to water vapor input， while during the event， the lower levels of the eastern boundary accounted for the majority of the water vapor input， which was closely related to the low-level easterly airflow and the complex topography of the South Xinjiang Basin. Analysis using the HYSPLIT model revealed that the primary water vapor sources affecting the snowstorm days in this region were Southwest Asia， the Mediterranean， Black Sea， and their vicinity， as well as Central Asia. Their contributions to the snowstorm areas were significantly higher in the lower troposphere than in the middle troposphere. The specific humidity and contribution rate of water vapor originating from Central Asia were significantly higher in mountainous areas than in plain areas. Below 700 hPa， water vapor from Southwest Asia and Central Asia was primarily transported to the snowstorm areas from heights below 1 500 meters. During the transport process， the specific humidity gradually decreased， which was inconsistent with the typical pattern of increasing specific humidity with decreasing height. After the water vapor reached the key region from its source via the westerly airflow， it was transported to the snowstorm areas at 500 hPa along a predominantly westerly path. At 700 hPa and below， the water vapor was primarily input into the snowstorm areas via both westerly and easterly paths， with the former being dominant. Water vapor transport at 850 hPa in the plain area was influenced by topography， resulting in relatively complex pathways， which should be given significant attention in practical operations. This study revealed that water vapor originating from the northeastern part of North America， the Norwegian Sea， the Arctic Ocean， and the southwestern part of Ili Prefecture could affect the western part of South Xinjiang under certain circulation conditions， with the moisture from the southwestern Ili Prefecture contributing significantly to the lower troposphere. Additionally， this study identified the key water vapor regions affecting snowstorm days in this area.

Intensive freeze-thaw cycles and rainfall processes in the seasonally frozen loess zone of China’s Loess Plateau progressively weaken soil structure and aggravate soil erosion， highlighting the need for green stabilization technologies that can maintain performance under cyclic freezing. Microbially induced carbonate precipitation （MICP） has attracted considerable attention as an environmentally friendly ground-improvement method. However， in cold regions， its application is constrained by the brittleness of the calcium carbonate cement， the non-uniform spatial distribution of precipitates， and the rapid degradation of erosion resistance under repeated freeze-thaw cycles. To address these limitations， a composite stabilization scheme combining MICP with polyacrylamide （PAM）， a water-retentive polymer， was proposed. Its effectiveness in improving the freeze-thaw durability of loess was systematically evaluated through coordinated macro- and micro-scale testing. Remolded loess was prepared as untreated specimens， MICP-treated specimens， and MICP+PAM specimens with different PAM dosages （by dry soil mass）. Cylindrical specimens were used for disintegration tests， and shallow plate specimens were employed for micro-penetration and rainfall-erosion tests. All specimens were subjected to 3， 5， and 10 freeze-thaw cycles between -20 °C and 20 °C， with 12 h of freezing and 12 h of thawing in each cycle， to simulate the seasonal temperature regime of the Loess Plateau. After the designated cycles， macroscopic indicators including disintegration rate， penetration resistance at a depth of 10 mm， and cumulative soil loss under artificial rainfall were measured. In addition， scanning electron microscopy （SEM） was used to obtain representative images of the microstructure of untreated， MICP， and MICP+PAM specimens before and after freeze–thaw cycling， thereby providing qualitative support for the macroscopic observations and revealing the main features of structural evolution. The test results showed that for all treatment types， disintegration resistance， penetration strength， and erosion resistance decreased with increasing number of freeze-thaw cycles and gradually tended towards a stable level， reflecting progressive microstructural deterioration followed by a new quasi-equilibrium state. Among the tested PAM contents， a dosage of 0.3% in the MICP+PAM group yielded the best overall freeze-thaw performance. After 10 freeze-thaw cycles， the MICP+0.3% PAM specimens exhibited pronounced improvements compared with both untreated and MICP-only loess. Specifically， the final disintegration rate decreased by 85.87% and 83.44%， respectively. The penetration resistance increased to 1.77 and 1.25 times that of the corresponding untreated and MICP-treated specimens. The cumulative soil loss decreased by 51.70% and 38.73%. These results indicated that adding an appropriate amount of PAM significantly enhanced the freeze-thaw durability of MICP-stabilized loess in terms of both hydraulic stability and near-surface mechanical strength. However， insufficient PAM produced weak bridging and water-retention effects， while excessive PAM tended to form locally dense films and reduced the efficiency of composite cementation. SEM observations provided a concise microstructural explanation for these macroscopic trends. Compared with untreated and MICP-only specimens， loess treated with MICP+0.3% PAM exhibited a denser and more continuous cemented fabric， in which biogenic CaCO₃ and PAM jointly bridged particles and refined pores. The presence of PAM promoted a more homogeneous distribution of CaCO₃ and introduced flexible polymer films around particles， which buffered ice-induced stresses and limited the development of microcracks during freeze-thaw cycling. Consequently， the integrity of the cemented skeleton was better preserved， and the loss of strength and erosion resistance was effectively mitigated over repeated cycles. Overall， this study demonstrates that the MICP+PAM composite technique can effectively alleviate freeze-thaw-induced deterioration of loess and significantly improve its resistance to disintegration， penetration， and rainfall erosion compared with both untreated and conventionally MICP-stabilized loess. By clarifying the synergistic action of microbial mineral precipitation and polymer film formation and by identifying an optimal PAM content of about 0.3%， this study provides a mechanistic basis and key parameter reference for the application of MICP+PAM in slope protection and soil and water conservation projects in seasonally frozen loess regions. The findings highlight the potential of this composite bio-polymer technology as a green and durable alternative to traditional cement-based stabilizers in cold-region loess engineering， and also underscore the necessity for further studies on long-term performance under combined freeze-thaw and wetting-drying cycles and variable rainfall conditions， so as to more effectively translate laboratory results into field applications.

The stability of permafrost subgrades is closely related to the surrounding hydrothermal environment. Under the influence of climate warming and humidification on the Qinghai-Xizang Plateau， the extensive scale and segmented design of permafrost subgrades in expressways lead to significant ponding issues. Based on long-term field investigations， unmanned aerial vehicle image modelling， and ground-penetrating radar surveys of Gonghe-Yushu Expressway （GYE）， combined with ground temperature and moisture monitoring in typical ponding sections， this study analyzed the spatial distribution of permafrost subgrade distress and their relationship with roadside ponding in the 340-km section from Elashan to Qingshuihe of the GYE. Additionally， it revealed how roadside ponding affected the development of large-scale permafrost subgrade distress. The results showed that subgrade distress in the permafrost sections of the GYE were primarily classified into three types： uneven settlement （77.3%）， transition settlement （13.5%）， and cracks （8.6%）. In the 340-km study section， more than 60% of the sections showed signs of subgrade distress， which were mainly distributed in the high plains and intermountain basins from Changshitoushan to Duoqinankelang and from Yeniugou to Qingshuihe. The distress rate reached up to 60% in permafrost regions with high temperatures and high ice content. The distress rate of the entire study section was 8.4%， while the distress rate in the permafrost section was significantly higher， reaching 18.05%. A strong correlation was observed between roadside ponding and the development of subgrade distress. Approximately 31% of the permafrost sections exhibited roadside ponding， which was mainly distributed in flat terrain. About 66.1% of subgrade distress was associated with roadside ponding. The severity of these distress increased with larger ponding area， greater ponding depth， and shorter distance to the subgrade slope foot. In the ponding sections， moisture content and temperature at the subgrade foot were significantly higher than those in the non-ponding sections， and the permafrost table beneath the subgrade was notably lower than that in the non-ponding sections. Roadside ponding served as a long-term heat source， continuously transferring heat to the permafrost layer beneath the subgrade. During the warm season， ponding absorbed external heat and stored it internally， while simultaneously transferring heat to the permafrost layer. In the early-to-mid cold season， the ponding surface froze， forming an ice layer that acted as a barrier to heat exchange with the external environment. However， due to the high salt content of ponding water， the freezing point was lowered， enabling continued heat transfer from the bottom of the ponding to the permafrost layer. This resulted in a shorter cold season for the soil beneath the ponding area， reducing the amount of cold absorbed by the soil. As a result， the ground temperature in the ponding area increased significantly， leading to asymmetric temperature distribution in the permafrost subgrade area. This process， coupled with increased moisture content， ultimately led to partial thawing of the permafrost layer beneath the subgrade， disrupting the subgrade and inducing various types of subgrade distress. The findings of this study provide valuable insights into the relationship between roadside ponding and subgrade distress in permafrost regions. The results provide a scientific basis for formulating preventive measures and maintenance strategies to mitigate ponding-induced distress in permafrost subgrades. Additionally， the study contributes to a deeper understanding of hydrothermal dynamics affecting subgrade stability and offers practical guidance for future construction and maintenance of subgrades in water-rich permafrost areas. Furthermore， the study highlights the critical role of roadside ponding in the management of permafrost subgrades.

Supraglacial debris comprising rock fragments， soil， and gravel， accumulates on glacier surfaces primarily due to processes such as mountain collapses， glacial erosion， and the exposure of englacial debris. It influences sub-debris ablation patterns through energy exchange and albedo variations， resulting in distinctive mass balance responses， hydrological effects， and hazard processes in debris-covered glaciers compared to debris-free glaciers. The impact of debris on underlying glacier melt is predominantly determined by the unique thermal processes driven by variations in debris thickness. When debris thickness is below a threshold （approximately 2~3 cm）， the absorbed heat is efficiently transferred to the sub-debris ice， accelerating glacial melt and generating basal meltwater that facilitates glacier movement. In contrast， thicker debris layers impede heat transfer， reducing melt rates， and potentially leading to over- or underestimation of glacier runoff. The spatially heterogeneous distribution of debris thickness results in differential melting across the glacier surface， giving rise to ice cliffs and supraglacial lakes. Ice cliffs absorb more solar radiation， intensifying localized melt and posing risks of collapse， while expanding supraglacial lakes can lead to glacial lake outburst floods （GLOFs）， threatening downstream populations and ecosystems. The thickness of supraglacial debris is thus a key factor in simulating glacier melt and accurately quantifying glacier runoff. It provides essential data support for research on dynamics， mass balance， and hydrological modeling of debris-covered glaciers， contributing to enhanced disaster risk management and water resource planning in downstream areas. The Langtang basin， located in the central Himalayas， is characterized by extensive supraglacial debris cover. Its unique geographical setting and glacial features make it a critical area for studying Himalayan glaciers and their responses to climate change. However， current research on supraglacial debris in this region remains predominantly focused on field measurements， with limited investigations into debris thickness and its spatial distribution at the watershed scale. It significantly impedes a comprehensive understanding of glaciers with debris coverage within this area. In response to the above questions， using Landsat 8 remote sensing imagery and high temporal resolution meteorological station data， this study calculated surface temperature， sensible heat flux， and net radiation flux in the Langtang region to solve the energy balance equation. Based on this approach， the spatial distribution of supraglacial debris thickness across glaciers in the Langtang region was retrieved. The study further analyzed the spatial distribution characteristics of debris thickness in representative glaciers and investigated the factors contributing to the heterogeneity in supraglacial debris thickness distribution. The study yielded the following findings： （1） The average thickness of supraglacial debris in the Langtang Basin glaciers was （0.25±0.02） m. Specifically， Lirung glacier had a thickness of （0.55±0.02） m， Shalbachum glacier （0.48±0.02） m， Langshisha glacier （0.31±0.02） m， and Langtang glacier （0.25±0.02） m. （2） Along the longitudinal profiles， the supraglacial debris thickness increased from the upper to the lower parts of the ablation zones. For Lirung， Shalbachum， and Langtang glaciers， the debris accumulation rates first decreased and then increased along the longitudinal profiles， whereas Langshisha glacier showed a continuous decrease. In terms of the transverse profiles， the thickness of the supraglacial debris was greater on the right side of the glacier movement direction for Lirung， Shalbachum， and Langtang glaciers， while Langshisha glacier exhibited thicker debris on both sides and thinner debris in the middle. （3） The variation in debris accumulation rates along the longitudinal profiles was primarily attributed to differences in surface flow velocities in the lower ablation zones. （4） The uneven distribution of debris thickness along the transverse profiles near the glacier termini was mainly due to differences in surface flow velocities， melt rates， and topographic features.

Compared with single snowstorm or gale disasters， wind-snow compound extreme weather events are more likely to cause extreme disaster losses. This study selected data including daily temperature， snowfall， minimum visibility， and maximum wind speed in Jilin Province from 1980 to 2023. The snowstorm weather index （SWI） was used to characterize the wind-snow compound extreme weather events. The random forest （RF） model was utilized to construct an extended sequence of the maximum wind speed. Furthermore， combined with the Mann-Kendall trend test and correlation analysis， the spatiotemporal variation characteristics of wind-snow compound extreme events in Jilin Province and their correlations with various meteorological and topographic factors were systematically analyzed. The results showed that the random forest model performed well in extending the maximum wind speed time series， with an R² exceeding 0.90. The number of snowstorm days （SD） at all intensity levels showed a fluctuating upward trend during the study period， and there was an abrupt change around 2020. Snowstorms could occur from October to March of the following year. The proportion of stations experiencing moderate or stronger snowstorms was highest in November and lowest in December and January. In terms of spatial distribution， snowstorms occurred mostly in the central and eastern parts of Jilin Province， especially in the eastern and southern regions where strong and extremely strong snowstorms were more likely to occur. The SWI in the eastern region showed a decreasing trend with high variability. Additionally， the SWI in the southern region demonstrated an increasing trend with low variability. The snowfall showed the strongest correlation with SWI， followed by minimum temperature and maximum wind speed. All three factors passed the significance tests with SWI， providing valuable indicators for evaluating future wind-snow compound extreme events.

Water resources， as a fundamental strategic resource， play a decisive role in the sustainable development of regional economy and society. In the context of intensifying global climate change， water scarcity has become a key bottleneck restricting the sustainable development of arid and semi-arid regions. Gansu Province， located in the arid and semi-arid areas of northwest China， has an inherent deficiency in water resources and an extremely uneven spatiotemporal distribution， with a water shortage rate reaching 14.1%. In recent years， with the rapid development of the regional economy and society， the contradiction between water supply and demand has become increasingly prominent. Especially under the dual-driven development model of industrialization and urbanization， the efficient allocation and sustainable utilization of water resources are not only crucial for the stable operation of the regional economy and society， but also hold significant strategic importance for maintaining the ecological security barrier in northwestern China and ensuring national ecological security. Against this backdrop， in-depth research on the characteristics and sustainability of water resource utilization in Gansu Province is of great theoretical and practical value for achieving coordinated development of high-quality regional growth and ecological civilization construction. Based on water footprint theory， this study systematically calculated the spatiotemporal evolution of the regional water footprint （WFP） in Gansu Province from 2011 to 2023， and developed a comprehensive evaluation indicator system for the sustainable utilization of water resources from four dimensions： water footprint structure， water footprint benefit， water resource ecological security， and water resource sustainability， aiming to reveal the characteristics and driving mechanisms of water resource utilization. Furthermore， the logarithmic mean Divisia index （LMDI） model was used to quantitatively decompose the contribution of population， economy， and technical efficiency to changes in WFP， thereby providing a scientific basis for the optimal management of water resources in Gansu Province. The results showed that： （1） the WFP in Gansu Province showed a significant upward trend， increasing from 310.57×10⁸ m3 in 2011 to 432.15×10⁸ m³ in 2023， with an average annual growth rate of 2.79%， demonstrating continuously increasing pressure on water demand. Significant spatial differences were observed， with Tianshui City having the highest average annual WFP （42.36×10⁸ m³）， while Jiayuguan City had the lowest （2.62×10⁸ m³）. The WFP structure was dominated by agricultural use， with agricultural water footprint （AWF） accounting for 92.15%， indicating that agricultural water conservation remained critical. （2） The water self-sufficiency rate （WSS） in Gansu Province remained above 96%， and the economic value of water footprint （EVWFP） showed a significant upward trend， increasing to 27.45 yuan·（m³）^-1 in 2023. However， the water scarcity index （WSI） and water pressure index （WPI） fluctuated within the range of 88.98% and 194.68%. During the “12th Five-Year Plan” period， the water resource status in Gansu Province was unsustainable. However， after implementing a series of water-saving measures， the water resource status improved to a sustainable state during the “13th Five-Year Plan” period. Nevertheless， in the early stage of the “14th Five-Year Plan”， the problem of water shortage became increasingly prominent， and the sustainable utilization of water resources faced significant challenges. （3） The LMDI model analysis indicated that economic and population effects positively drove changes in WFP， with the economic effect being the main driving factor （accounting for 58.47% of the contribution）， while the technical effect had a negative driving effect， accounting for 40.05% of the total effect. The inhibitory effect of technological progress on WFP was significant. The findings provide a theoretical basis and decision-making support for Gansu Province to formulate scientific and reasonable water resource management policies， optimize water resource allocation， and promote the sustainable utilization of water resources， thereby facilitating the coordinated development of water resources， economy， and ecology in Gansu Province.

Under global climate change and China’s Western Development Strategy， engineering construction in frozen soil regions faces significant challenges. In pile foundation in regions with frozen soil， long-term stability issue has become a critical technological challenge for infrastructure development in cold regions. Particularly， the shear creep effect at the frozen soil-structure interface significantly influences the performance of pile foundations and other engineering systems. Under dynamically varying loads and temperature fields， progressive damage induced by shear creep at the interface poses serious threats to structural performance. Due to the temperature sensitivity and seasonal deformation behavior of frozen soil， pile foundations are especially affected by freeze-thaw cycles. These foundations are subjected to the combined effects of temperature， stress， and displacement over long periods， which makes the creep behavior at the pile-soil interface increasingly complex. This often leads to a reduced pile bearing capacity and uneven settlement of structures， resulting in engineering deterioration. However， current research on the shear creep at the frozen soil-structure interface has several limitations. First， traditional experimental devices， such as direct shear and single-shear devices， can generally simulate two-dimensional shear conditions， making it difficult to replicate the three-dimensional complex stress state and true mechanical response of pile foundations and other engineering structures. Second， existing studies primarily focus on the creep characteristics of the frozen soil itself， with limited systematic understanding of the creep behavior and response mechanisms. Third， current constitutive models are insufficient in characterizing the nonlinear accelerated creep phase， making it difficult to accurately predict the creep behavior of interfaces under different engineering conditions. To address these limitations， this study investigated the mechanical behavior of the frozen soil-steel interface through triaxial shear creep tests on saturated frozen soil under different surface roughness， temperature， and confining pressure conditions. The experiments investigated the patterns of creep deformation， creep rate， and time variation， thereby revealing the shear creep mechanism at the interface under the influence of various factors. Additionally， a mathematical model based on the Burgers model was established to describe the accelerated creep phase of the saturated frozen soil-steel interface. The results indicated that in the triaxial shear creep tests with graded loading， the deformation process under different temperature conditions followed a four-stage evolution pattern： instantaneous creep， primary （transient） creep， steady-state creep， and accelerated creep. Shear stress， temperature， and roughness were identified as the dominant factors affecting shear creep. The deformation behavior was controlled by shear stress levels. Under low stress conditions， steady-state creep predominated， ensuring long-term interface stability. Under high stress conditions， creep intensified， leading to interface failure. Lower temperatures significantly reduced both deformation magnitude and rate at the interface. Smooth interfaces exhibited rapid increases in shear displacement during initial loading， exhibiting significantly higher creep rates and deformation compared to rough interfaces. Creep deformation showed a distinct nonmonotonic trend with increasing interface roughness—first decreasing， then increasing—indicating the existence of a critical roughness that optimized inte rface performance. The effect of confining pressure was relatively small. The proposed model effectively characterized the viscoelastic behavior during the non-accelerated phase and captured the displacement jump observed in the accelerated creep phase. The systematic analysis of the data revealed the effects of different conditions on the shear creep parameters of the frozen soil-steel interfaces. These patterns provide valuable insights for understanding and predicting interface behavior under similar conditions and offer references for engineering applications. In conclusion， this study provides important experimental evidence and theoretical support for establishing pile foundation design theories that consider interface creep effects and for developing methods for long-term stability evaluation of cold-region engineering.

Permafrost is widely distributed on the Qinghai Plateau， particularly in the South Qinghai Plateau and Qilian Mountains regions. The frozen soil environment is highly sensitive to climate change. However， current understanding of the spatiotemporal variation patterns and driving mechanisms of the freeze-thaw process on a large scale remains insufficient， especially regarding how geographic and climatic factors influence the freeze-thaw process. Based on data from 42 meteorological stations in Qinghai Province from 1961 to 2023， this study employed spatial variation analysis and climate abrupt-change testing to statistically analyze the spatiotemporal characteristics of the freeze-thaw index. The results indicated that： （1） the air freezing index （AFI） and number of freezing days （NF） in Qinghai Province exhibited a spatial distribution pattern increasing from north to south and from east to west， while the air thawing index （ATI） and number of thawing days （NT） showed the opposite trend. Over the past 63 years， AFI and NF decreased significantly ［with climatic tendency rates of -72.8 ℃·d·（10a）^-1 and -3.7 d·（10a）^-1， respectively］， while ATI and NT increased significantly ［69.5 ℃·d·（10a）^-1 and 3.7 d·（10a）^-1］. （2） The year 1997 marked an abrupt change in the freeze-thaw index. During the post-abrupt-change period （1997—2023）， AFI decreased by 236.5℃·d， and ATI increased by 205.9 ℃·d. Before the 1980s， the AFI anomaly showed a positive tendency， which shifted to negative after the 1990s， with the opposite pattern observed for ATI. Spatially， the high-altitude areas of the Three-River Source Region experienced the largest decrease in AFI， indicating greater sensitivity of freezing processes to climate warming at higher altitudes. The eastern agricultural area and the Qaidam Basin showed a significant increase in ATI， reflecting a more pronounced impact of warming on thawing processes at lower altitudes. （3） AFI showed a significant negative correlation with air temperature （T）， while ATI showed a highly significant positive correlation with T. Altitude was the primary geographic factor influencing the spatiotemporal variation of the freeze-thaw index. AFI exhibited a highly significant positive correlation with altitude and a significant negative correlation with longitude， with correlation coefficients of 0.792 and -0.437， respectively. ATI was significantly positively correlated with longitude and latitude， and highly significantly negatively correlated with altitude， with correlation coefficients of 0.332， 0.269， and -0.991， respectively. Altitude was the main geographic factor affecting the spatiotemporal variation of the freeze-thaw index. For every 100 m increase in altitude， AFI and NF increased by 92.7 ℃·d and 6.3 d， respectively， while ATI and NT decreased by 80.9 ℃·d and 6.3 d， respectively. Among atmospheric circulation influencing factors， the Tibet Circulation-2， East Asian trough intensity， northern boundary of the Northern Hemisphere subtropical high， and Northern Hemisphere polar vortex intensity showed significant relationships with changes in the freeze-thaw index in Qinghai Province. Accurately quantifying the spatiotemporal characteristics of the freeze-thaw index on the Qinghai Plateau provides solid data support and theoretical foundation for a holistic understanding of multi-sphere interactions on the plateau.

Since 2000， influenced by both climate change and human activities， the area of Chagan Nuur Lake has sharply shrunk from 105.3 km² to 30 km²， with a reduction rate of over 71%. Challenges such as water resource scarcity， soil salinization， and ecological degradation have become increasingly severe， posing a significant threat to regional ecological security. In this context， analyzing the hydrochemical characteristics of lakeshore zone water of Chagan Nuur Lake is of great significance for understanding the hydrogeochemical processes and their ecological effects in cold and arid regions. Ice-water samples were collected from 21 sites in the lakeshore zone water of Chagan Nuur Lake during the non-ice-covered period （June 27， 2023） and the ice-covered period （January 17， 2024）， including groundwater （well and spring water） and surface water （lake， river， and reservoir water）. By using methods such as Piper trilinear diagrams， Gibbs model， and ion ratios， the hydrochemical characteristics and their controlling factors were investigated. The results showed that： （1） significant seasonal variations were observed in ion concentrations in the lake’s ice-water system. The concentrations of non-ice-covered-period water were 1.32 times， 15.17 times， 17.01 times， and 12.51 times higher than those of ice-covered-period water， ice surface， ice interior， and ice-bottom water， respectively. （2） The cation distribution in the lakeshore zone of Chagan Nuur Lake followed the order： Na⁺ > Mg²⁺ > Ca²⁺ > K⁺， while the anions were mainly Cl⁻ > HCO₃⁻ > SO₄²⁻ > CO₃²⁻. （3） The hydrochemical characteristics of the ice-covered and non-ice-covered periods were generally similar， but significant differences were observed among water types： spring water was SO₄-Ca·Mg type， river and reservoir water were Cl·SO₄-Ca·Mg type， and lake water， lake ice， and well water were all Cl-Na type. （4） In terms of controlling mechanisms， lake water was primarily influenced by evaporation concentration （non-ice-covered period） and freezing concentration （ice-covered period）， showing a strong trend of sodium salt enrichment， indicating that the lake was evolving towards a salt lake. Spring water was mainly controlled by carbonate weathering and enriched in Mg²⁺. River and reservoir waters were influenced by both weathering and human activities. Well water was mainly controlled by the combined effects of carbonate weathering and albite dissolution. （5） Future research should focus on the combined effects of climate and human activities， establish a “water-salt-carbon-biology” coordinated management system， and use cross-scale models to achieve dynamic maintenance and early warning of lake water-salt balance. This will provide theoretical basis and practical references for ecological restoration of degraded lakes and sustainable regional water resource utilization.

In recent years， scholars have conducted extensive research in the field of frozen soil mechanics， establishing a systematic theoretical framework and laying a solid foundation for frozen soil engineering. However， current studies mainly focus on the mechanical properties of frozen soil within the temperature range of -30 ℃ and above. Faced with the increasingly frequent extremely low temperature environments in cold regions and the extreme temperature conditions encountered in deep space exploration， research on the mechanical behavior of frozen soil at ultra-low temperatures remains limited. With the advancement of national strategies such as “Xinjiang-Xizang Connectivity”， “Polar Security”， and “Deep Space Exploration”， research on ultra-low-temperature frozen soil mechanics has become a critical foundation for ensuring engineering safety in cold regions and exploring planetary cryospheres. This field represents a novel frontier in the study of frozen soil mechanics. Nevertheless， limited testing platforms and underdeveloped methodologies have confined this research to an exploratory stage， with systematic theoretical models yet to be defined and current capabilities being insufficient to meet the requirements of major engineering requirements. This study first systematically reviews the experimental devices and methods for studying the mechanical properties of frozen soil under ultra-low temperature environments. Existing studies have shown that the accuracy of mechanical testing equipment for frozen soil is significantly affected by ultra-low temperature environments， and that standardized testing protocols are still underdeveloped. Furthermore， at ultra-low temperatures， the temperature sensitivity of frozen soil compressive strength decreases， and failure modes generally exhibit more brittle characteristics. The stress-strain relationships demonstrate distinct stress-decay phases. Subsequently， the mechanisms underlying the change of mechanical properties of frozen soil at ultra-low temperatures are explored， followed by a review of the challenges faced in the research on ultra-low-temperature frozen soil mechanics and prediction methods applied in this field. Finally， future research directions for frozen soil mechanics at ultra-low temperatures are proposed， aiming to provide scientific references for theoretical advancement and critical engineering applications in extremely low temperature environments.