As a significant ecological barrier in northwest China， the Qilian Mountains， it is crucial to clarify the spatial-temporal variation of its ecosystem services （ESs） and evolution mechanism for regional ecological protection and restoration and sustainable development. Using the InVEST （Integrated Valuation of Ecosystem Services and Trade-offs） and CASA （Carnegie-Ames-Stanford Approach） models， we have quantified six critical ESs between 2000 and 2020 including water yield， carbon sequestration， soil retention， nutrient retention， carbon storage， and habitat quality in Qilian Mountains and then analyzed their spatial-temporal variability and spatial distribution characteristics at various scales， and the hotspots of ESs were identified. According to the results， since 2000， the six ESs in Qilian Mountains have been increasing in the region as a whole as well as in different land types， with the mountainous areas being the high value areas and the hotspot areas of category III and IV in the hotspot analysis showing an increasing trend； The six ESs in Qilian Mountains are spatially characterized as being “high in the east and low in the west”， with high values of supply per unit area for forest and grassland， and large differences in the supply capacity of ecosystem services in different topographic gradients， mostly peaking above 3 100 m above sea level， with category V hotspots in the hotspot analysis mainly located in the eastern mountainous areas. The results of the study identify the spatial-temporal variability of six ESs in the Qilian Mountains and explore the natural and anthropogenic factors affecting their spatial and temporal variability， which can provide the basis and suggestions for the formulation of targeted ecological conservation strategies and high-quality sustainable development policies in the Qilian Mountains.

Vegetation phenology is an important indicator of vegetation response to changes of the natural environment. As an important component of the cryosphere， permafrost is extremely sensitive to climate change. Accurate monitoring of vegetation phenology in permafrost areas is important for studying the response of cold-zone ecosystems to climate change. The Greater Khingan Mountains permafrost zone is the only high-latitude perennial permafrost in China， and studying the vegetation phenology in this zone can improve our understanding the response of cold-zone ecosystems to global climate changes. In order to investigate the impact of climate change on vegetation in the permafrost zone of the Greater Khingan Mountains， this paper assesses the spatial and temporal characteristics of vegetation phenology in the study area over the past 20 years and its response to climate change based on MODIS EVI time series data， with a view to enriching the study of vegetation phenology in cold-zone ecosystems. This study firstly compares the differences and adaptability of Normalized Difference Vegetation Index （NDVI）， Enhanced Vegetation Index （EVI） and Solar-induced Chlorophyll Fluorescence （SIF） in phenology studies of the permafrost zone， and the results show that EVI is more effective than the other two indices. Secondly， combines MODIS EVI time series data and meteorological data from 2000 to 2019， key phenological parameters， the beginning （SOS）， end （EOS） and length （LOS） of vegetation growing season， were extracted using Savitzky-Golay （S-G） filter and dynamic threshold approaches to analyze the spatial and temporal variation of vegetation phenology in the study region and its response to climate changes. In this case， the NDVI and EVI have a spatial resolution of 250 m and a temporal resolution of 16 d. Data processes such as mosaicking， masking and projection transformations were performed on each period of data to obtain time series data of NDVI and EVI for the study area from 2000 to 2019. The 16 d of SIF data were combined into one period according to MODIS data synthesis rules， with 23 periods of data per year. And the mean value of SIF for the study area was calculated to obtain time series data for SIF from 2015 to 2019. Monthly mean temperature and monthly total precipitation in growing seasons were selected and used the method of partial correlations analysis to analyze how climate factors affect vegetation phenology. The results show that： （1） the time series of NDVI， EVI and SIF can reflect the seasonal changes of vegetation growth in the study area， and the change curves of EVI and SIF are more consistent than those of NDVI； （2） the variation of SOS in the study area from 2000 to 2019 ranged from 96 to 144 d in chronological days， with a mean value of 129.46 d； the variation of EOS ranged from 272 to 320 d， with a mean value of 295 d and the LOS was concentrated between 128 and 224 d， with a mean value of 165.65 d. Due to the existence of the inversion phenomenon and the differences of vegetation types， the LOS in the continuous multi-year permafrost area was greater than that in the island thaw zone； （3） the mean values of SOS and EOS trends in the study area were -1.23 d·（20a）^-1 and -0.46 d·（20a）^-1， respectively， both showing early trends and the mean value of LOS trend was 2.39 d·（20a）^-1， showing an extended trend. Vegetation SOS in the study area was negatively correlated with the mean temperature of March to May （P<0.05）， and EOS was positively correlated with the mean temperature and precipitation August to October （P<0.05）. The extent and trend of significant changes in vegetation phenology were greater in the continuous permafrost zone than that in the island thaw zone permafrost zone， indicating that continuous permafrost zone is more sensitive to climate change. This research improve our understanding of vegetation phenology characteristics of permafrost region under the background of climate change.

Glacial lakes are important indicators of climate change. Against the backdrop of global warming， the continuous monitoring of glacial lakes is of significant importance for regional water resource management and disaster prevention and mitigation in alpine regions. Affected by glacier melt and climate influences， the boundaries of glacial lakes undergo seasonal and inter-annual variations. Existing glacial lake mapping researches often first identify the location of each lake on remote sensing images， and then capture its detailed boundary. In recent years， glacial lake inventories of different regions have been growing， offering a wealth of historical boundaries for glacial lakes. Monitoring these glacial lakes with known locations only requires the extraction of their current boundaries， and the historical boundaries can serve as the starting point for glacial lake boundary iteration， accelerating the update of glacial lake inventories. This study takes 488 glacial lakes with favorable imaging conditions， and another 80 glacial lakes affected by snow， ice， clouds， and mountain shadows in the Himalayas as the research subjects. The former is categorized into three classes based on their area size. Using historical boundary information provided by the 1990 glacial lake inventory， we compared the glacial lake extraction results on post-2014 Landsat-8 OLI images using manual thresholding， OTSU thresholding， U-NET， bimodal iterative method， OTSU iterative method， and C-V iterative method. The results showed： OTSU iterative method and C-V iterative method can effectively utilize the statistical information within the glacial lake buffer zones， achieving F1 scores as high as 88.89% and 89.30% respectively， which are significantly better than manual thresholding， OTSU thresholding， and bimodal iterative method. They can also extract frozen and cloud-covered glacial lakes more completely. For glacial lakes covered by snow， the C-V iterative method achieved the highest extraction accuracy. The U-NET model achieved an F1 score of 89.80%， accurately extracting glacial lakes connected to mountain shadows. This research offers methodological support for the long-term monitoring of glacial lakes with known historical boundaries.

Glacier is an important part of the cryosphere and is known as the "indicator" and "early warning device" of climate change. Under the influence of its own gravity， the glacier will move from the high-altitude area to the low-altitude direction to form the glacier ice flow. The maximum flow trajectory of the glacier ice is called the glacier mainstream line. The centerline of the glacier is the center line of the main stream line of the glacier and one of the key parameters to reflect the geometric shape of the glacier. It has important significance in measuring the change of the glacier length with time， analyzing the change characteristics of the glacier movement speed， estimating the glacier volume and building a one-dimensional glacier model. Based on the Tyson Polygon method， combined with the shortest distance formula and Dijkstra algorithm， according to the topographic characteristics of the glacier surface， this paper proposes a new method for automatic and rapid extraction of glacier centerlines in mountain glaciers from the perspective of morphology. This method encrypts the glacier vector boundary line according to the area of different glacier’s using empirical parameters， and combines the digital elevation model data to extract the glacier center line using the Tyson Polygon method with the lowest point and the local highest point on the glacier vector boundary line as the end point and the starting point. At the same time， the extracted local highest point is screened using the shortest distance formula， and only the point with the highest elevation is reserved as the starting point of the glacier centerline， In order to avoid the problem of multiple starting points of the glacier centerlines， the Dijkstra algorithm is used to filter the extracted glacier centerlines with the distance between the starting point and the end point as the weight， and only the shortest distance between the starting point and the end point is retained as the final center glacier centerline extraction result. This method was applied to 1 014 mountain glaciers in the Qinghai Tibet Plateau， and 2 114 corresponding glacier centerlines in glaciers were automatically generated， with an average length of 3.13 km and an extraction success rate of 100%. By comparing the proposed method with the methods of Kienholz et al.， Zhang et al.， it is found that the extraction results of the proposed method and the method of Kienholz et al.， Zhang et al.， have little difference for the simple structure of the single valley glacier. For compound valley glaciers and slope glaciers with more complex structures， the number of glacier centerlines extracted by the proposed method is more than that by Kienholz et al.， but less than that by Zhang et al.， and for ice cap glaciers， the number of glacier lines extracted by the proposed method is more than that by Kienholz et al.， Zhang et al. The results show that the glacier centerline extracted by this method has a more complete coverage， and the average length ratio of the glacier centerline extracted by this method and Kienholz's method is 0.98， which has a high consistency. This paper proposes a method of extracting the glacier centerline based on Tyson Polygon method， which takes the data of the digital elevation model and the boundary line of the glacier vector in the glacier catalog data as the input， can realize the automatic and rapid extraction of the glacier centerline without manual participation， and can also accurately extract the centerline in the tributaries of the glacier. This method solves the problem that only a single center line can be extracted based on the glacier axis method， and improves the extraction efficiency of glacier centerline.

With the general improvement of the national living standard， the ice and snow industry are developing rapidly， which brings many benefits to the social and economic development. Understanding the correlation between ice and snow industry and social economic development is the basis for rational development and management of ice and snow industry. In this study， based on the main components of the ice and snow industry that ski industry in Xinjiang， the research establishes the relationship between the snow and ice industry and the main economic industries， combining input-output relationship between regions， evaluating the economic and social effects of the snow and ice industry with the input-output analysis method. The results show that the production capacity of ice and snow industry in Xinjiang is about 34.5 billion and 181.9 billion RMB in 2012 and 2017， and capacity of other economic industries driven by the total output value is can reach 80.9 billion and 415.1 billion RMB. And the ice and snow industry influence coefficient are 2.28 and 2.33 in 2012 and 2017， which is larger than the that of most major economic industries； The largest four industries driven by the ski industry include the transportation， warehousing and postal services； Petroleum， coking products and nuclear fuel processing products； Wholesale and Retail； Accommodation and catering industry. Among them， the relative impact of wholesale and retail， accommodation and catering increased significantly， and the impact of petroleum and coking product departments decreased relatively. Through decomposition of economic structure found， the economic pull effect of ski industry is further strengthened. In particular， with the rise of ice and snow industry the accommodation and catering output value increased 259%. At the same time， the ski industry can provide more local employment opportunities. In a word， Xinjiang， relying on the advantages of the natural endowment of the ski industry， can bring opportunities for social and economic development. In order to maximize the economic leverage of the snow and ice industry， it is not only necessary to have a reasonable economic structure as the basis， but also to develop other closely related industries as support. In addition， in order to provide enough space for the development of the ski industry， the four major industries that are most driven by the ski industry should also be the key industries to promote their development， and fully absorb the economic dividends of the development of the ice and snow industry， so as to obtain more economic and social benefits.

The glacial lake outburst floods （GLOFs） in the Yarkand River Basin （YRB） may cause severe disasters. In the context of global warming， establishing a GLOF database and understanding the variation characteristics of GLOFs are the basis for glacier lake research and the flood risk assessment and management. Based on the hydrological runoff data， scientific expedition data and flood disaster data， a database of Kyagar GLOFs in the YRB from 1961 to 2021 is established. Then， the frequency and peak discharge variations of GLOFs are analyzed by using statistical methods such as linear trend analysis， rescaled range analysis， accumulated anomaly analysis， Mann-Kendall test and Morlet wavelet analysis. In addition， the peak discharges under different return period scenarios are estimated by using the Multi-Distribution Fitting Tool software， and the corresponding risk assessment is conducted. The results show that the frequency and peak discharge of Kyagar GLOFs in the YRB both exhibit a unimodal pattern on the monthly timescale. The risk of GLOFs is the highest in August and September. The frequency of Kyagar GLOFs from 1961 to 2021 does not show an obvious linear trend but exhibits a decadal oscillation. The peak discharge， however， shows a significant linear decreasing trend with a rate of 15.5 m³·s^-1·a^-1， indicating that the benefits outweigh the drawbacks when utilizing floodwater resources. On the quasi-19-year period， the frequency of Kyagar GLOFs is at a low level in conjunction with a relatively low risk in the short term. It will continue to decline until around 2027， after which it will enter the next rising phase. Although there was a short-lived high-frequency period of Kyagar GLOFs due to global climate change in the early 21st century， this cannot alter the long-term oscillatory decrease trend. The Kyagar GLOFs pose a certain level of disaster risk under the 5-year return period scenario， while the disaster risk is extremely high under the 10-year or above return period scenarios. Therefore， we recommend strengthening the dynamic monitoring and risk assessment of the Kyagar Glacier Lake in the Yarkand River， as well as the research on the mechanisms of GLOFs. Additionally， we need to accelerate the construction of water conservancy projects and enhance the flood control and disaster reduction capabilities.

Land-bridge of 1401 milestone in the Qinghai-Tibet Railway is located in the permafrost area of Tanggula Mountains. After the railway official operation， serious settlement disease occurs in the bridge and serious endangers the line. Settlement data from 2009 to 2018 show that the pile foundation had stabilized after a series of treatment in 2009， 2011—2012， and 2017. The further analysis shows that the rise of permafrost temperature and base resulting from warm season， climate warming and leakage of deep artesian sub-permafrost groundwater were the main reasons of settlement problems. Complicated and long treatment process implied that thermal disturbance from countermeasures must be considered before it was put into practice. It is suggested that the land-bridge settlement should be monitored periodically. Earlier detection， earlier prevention and earlier treatment should be more welcome.

Glacial lake outburst floods are characterized by wide range， long duration， high hazard and often accompanied by debris flow. At present， there is a lack of quantitative studies on the dynamic evolution characteristic of glacial lake outburst floods. To this end， the evolutionary characteristics of Cirenmaco glacial lake outburst flood disaster are studied based on field survey and multi-period remote sensing images， and the sediment transport model and hydrodynamic model coupling method are used to reveal the evolutionary characteristics of glacial lake outburst flood erosion. The model is based on the digital elevation model （DEM） topographic data with an accuracy of 12.5 m， simulating the inversion of the 1981 Cirenmaco glacial lake outburst flood dynamics evolution process， comparing with the actual measurement results， verifying the applicability and feasibility of the model， and conducting prediction analysis of the glacial lake outburst again， to quantitatively evaluate the characteristics of flow depth， flow velocity， erosion and deposition of glacial lake outburst flood in the evolution process. The outburst flood scours erosion on the moraine deposit of the Zhangzangbu branch ditch and the loose colluvium of the downstream ditch bank during the evolution， and the high sediment concentration flood gradually evolves into turbulent debris flow. At the 707 landslide， the flow depth is 8~10 m， the maximum flow velocity is 13.7 m·s^-1， and the erosion depth is 8~9 m. The turbulent debris flow formed a barrier dam at the main ditch deposition， with a dam height of 9~11 m， which briefly blocked the Boqu River. turbulent debris flow to the downstream landslide group of Zhangmu port for scouring side erosion， erosion depth of about 10~13 m， easy to trigger large-scale secondary disasters， turbulent debris flow reaches the hydropower station， siltation buried hydropower station intake， resulting in the failure of the hydropower station. On the whole， outburst flooding in the evolution process， the flood water to the upstream ditch bed and ditch bank for strong erosion entrainment， flood peak flow enhanced. In the midstream， turbulent debris flow laterally erodes the gully bank， and the flow velocity increases in the narrow part of the gully， and undercut erosion is enhanced. In the wide part of the channel， low velocity decreases， solid materials deposition， overall reach the balance of flushing and siltation， flood peak flow gradually decay with distance， to the downstream， channel topography open， flow velocity slows down， turbulent debris flow gradually deposition， while the lateral erosion on both sides of the channel， overall for deposition. The model can well reveal the evolutionary dynamics of glacial lake outburst flood disaster erosion characteristics.

The formation and growth of ice interlayers in rock fractures are significant characteristics and causes of rock frost weathering. Fractured rock masses consist of fractures and rock matrix， which differ greatly in properties and can be considered as a porous medium. Water is typically the primary storage and transport channel in the rock mass， while unsaturated fractures often involve both gas-phase and liquid-phase migration， making it difficult to directly explain the formation and growth process of ice layers in the fractures using existing in-situ frost heave and frost segregation mechanisms. To investigate the occurrence and distribution of ice layers in unsaturated rock mass fractures， the authors conducted a unidirectional freezing test on two cement test blocks with a single vertical fracture under warm-end water replenishment conditions. After the test， three horizontal fractures and one vertical fracture appeared in the test block， with thin ice layers forming in the fractures. The negative temperature zone of the sample showed significant frosting， and the total amount of water migration during the entire process reached 221 mL， primarily in the gas phase. Based on the fundamental principles of heat transfer， the authors established a frost model for a single fracture wall surface under natural convection conditions. The frost surfaces were divided into classes I~Ⅲ based on the characteristics of different cold surfaces of the test samples， and the thickness， density， and unit mass of frost layer on the three types of cold surfaces were calculated over time. The experimental results validated the calculation results， demonstrating the reliability of the frost model on a single wall surface. Based on the frost model and relevant literature analysis， the direct factors affecting the amount of frost on the negative temperature zone wall surface of rock fractures include the convective heat transfer conditions in the fracture， the relative humidity of the air， and the size of the negative temperature zone wall surface area. The greater the values of these three factors， the more frost will form on the fracture wall surface within a certain time， and the more significant the ice formation in the fractures. The temperature gradient along the fracture is the essential reason， as the greater the temperature gradient， the stronger the convective heat transfer in the rock fracture， and the larger the negative temperature zone wall surface area， the more frost will form within a certain time.

Canals are damaged by frost heaving， thawing， and soil erosion in seasonal frozen soil areas. In this study， a canal lining cushion was proposed， composed of polystyrene foam board， ceramic， and epoxy resin， to alleviate the damage to the canal. In this study， the test method used was the alternating dry-wet and freeze-thaw test to investigate the degradation law of thermal insulation and bond strength of the composite liner. Also， this study investigated the protective effect of the multi-layered combined bedding on the concrete slab. Finally， this study determined the optimum admixture of ceramic sand by combining statistical methods. The test results show that ceramic can reduce the thermal conductivity of multilayer cushions. The thermal insulation of the cushion deteriorates， caused by the polystyrene foam board and concrete， and decreases linearly with the increase of test times. The ceramic content did not affect the bond strength of the polystyrene foam board interface. However， it greatly influenced the strength of the concrete interface， and the content was inversely proportional to the strength. Multilayer cushions had a better protection effect on concrete. Compared with ordinary concrete， EP-concrete reduced the loss of quality and compressive strength by 1.04% and 8.58%， respectively. The thermal insulation of the cushion and the bonding strength of the contact surface of the polystyrene foam board were taken as the control objectives， and the calculation method of the optimal amount of ceramic sand was established. The research results of this experiment can provide a certain reference for the design of the canal cushion.

In Northeast China， the annual temperature difference is relatively large， and the subgrade soil presents freeze-thaw cycle with the change of temperature， belonging to seasonal frozen soil， which is easy to cause serious deformation of subgrade in Northeast China during service， the occurrence of subgrade diseases such as frost heave， frost heave and mud heave leads to the change of soil structure， the reduction of subgrade stiffness， and the deterioration of bearing capacity. The stress-strain relationship can effectively represent the law of stress and deformation of soil， and the establishment of a reasonable freeze-thaw damage model has guiding significance for the structural design of seasonally frozen soil subgrade. Consequence， in order to explore the stress strain relationship and damage mechanism of subgrade soil in seasonal frozen area under different water content under different freeze-thaw cycles and confining pressure，choosing a highway section in Fuxin City， Liaoning Province as the test section， using the ring knife method to select the subgrade soil sample of the test section and freeze-thaw cycle and triaxial compression test， the experiment obtained the relation of the stress-strain curves of subgrade soil under different freeze-thaw cycles and confining pressures. According to the damage mechanics and statistics principles， the Weibull distribution is combined with Lemaitre effective stress principle to establish a damage constitutive model of subgrade soil in the seasonal freezing zone. The results show that the optimum moisture content of subgrade soil in seasonally frozen area is the limit moisture content of its stress-strain curve from strain softening to strain hardening. When the moisture content is less than the optimum moisture content， the peak stress of the curve increases with the decrease of the moisture content， and decreases with the increase of the number of freeze-thaw cycles. When the moisture content of the test piece is greater than the optimum moisture content， the curve has no peak stress with the increase of the moisture content， showing obvious strain hardening characteristics. When the number of freeze-thaw cycles is small and the confining pressure is low， the stress-strain curve of the specimen shows a peak stress， and the curve shows a strain softening feature with the increase of confining pressure， the strength of the specimen increases. Strain hardening is easy to occur when there are many freeze-thaw cycles and large confining pressure. Through comparative analysis， the established damage constitutive model is in good agreement with the test stress-strain curve. It can reflect that the stress-strain curve of subgrade soil in the seasonally frozen area shows a change rule of first increasing and then tending to be stable. And the parameters required by the model can be obtained through triaxial tests， which shows that the model can better describe the stress-strain relationship of subgrade soil in seasonal frozen area， and is practical. In addition， it can be seen from the test results that in order to reduce the impact of freezing and thawing cycles on the subgrade strength in the seasonal freezing area in Northeast China， it is very important to do a good job in the waterproof and drainage of the subgrade in advance for the prevention and treatment of subgrade freezing and thawing diseases.

In recent years， the retreat of global glaciers has accelerated in the context of climate warming. However， there is a positive glacier mass balance anomaly in the West Kunlun Mountains of the northwestern Qinghai-Xizang Plateau. Previous studies found that the mass gain in this region started as early as 1970s， and several glaciers have recently surged. Many large-scale glaciers are concentrated in the main peak area of West Kunlun Mountains， and the mass balance anomaly of glaciers in the West Kunlun Mountains has attracted extensive attention. However， there many states of the glacier tongue simultaneously， such as advance （normal）， shrinkage， stable and surge. The advance and retreat of glaciers are not only related to the mass balance influenced by climate change， but also closely related to the glacier velocity. Previous studies mostly focused on the former， but less on the latter. Therefore， this study uses the ITS_LIVE v01 velocity product， combined with the glacier surface elevation changes and thickness to analyze the characteristics and possible causes of glacier surface velocity change with different states of terminus dynamic in the West Kunlun Mountains from 2000 to 2018， to further understand the glacier motion and velocity changes under the same climate change scenario. The results showed that during the study period， the mean velocity of glaciers in the West Kunlun Mountains was 6.35 m⋅a^-1， and showed a fluctuating upward trend， which was caused by the mass gain caused by the thickening of glacier ［（0.15±0.02） m⋅a^-1］. The mean velocity of advanced glacier was about 4.07 m⋅a^-1， and showed an increasing trend during the study period， which was resulted by the slight mass gain， the average velocity of retreat glacier was about 4.86 m⋅a^-1， and showed an decreasing trend， this was caused by the mass loss during the period， the mean velocity of stable glacier was 3.04 m⋅a^-1， and the mass balance of this type glacier was basically stable from 2000 to 2018. In the study area， the mean velocity of advanced glacier is smaller than the retreat glacier， because the value of retreat glacier is larger than the advanced glacier. From 2000 to 2018， seven glaciers surged， and the surge period was concentrated in 2010—2015. During the study period， the thickness of the surge-type glaciers increased significantly ［（0.35±0.02） m⋅a^-1］， resulting in an acceleration in velocity. The retreat glacier with a large length in the West Kunlun Mountains experienced a surge before 2000 and in a quiet period now， mass cannot be transported from the upstream to the downstream， leading to the thinning and retreat of the glacier tongue.

Small glaciers are very sensitive to climate change， and monitoring and quantitative assessment of this type of glacier changes can help to understand the magnitude and mechanism of glacier response to climate change. In this study， we combined multi-source remote sensing data （satellite and UAV data） and meteorological data to analyze the change in area of Kuoqionggangri No. 1 Glacier in the Nyainqêntanglha Mountains， Qinghai-Tibet （Xizang） Plateau over the past 50 years， and quantitatively evaluated the magnitude of recent elevation changes and the spatial distribution of Kuoqionggangri No. 1 Glacier. The study shows that the area of Kuoqionggangri small cirque glacier shrank from （1.444±0.013） km² to （0.712±0.001） km² during 1968—2021， with a shrinkage of 50.7%， and the average rate of glacier terminus retreat was （6.23±0.71） m⋅a^-1. Based on the high-precision UAV survey data from 2020 to 2021， we found that the average elevation difference of the ice surface of Kuoqionggangri No. 1 Glacier reached （-2.41±0.69） m⋅a^-1. The average ice surface elevation change at the glacier terminus was greater than 3 m， and a decrease in the central part of the glacier in the range of 1.5 to 3 m. The study also found that glacier supraglacial streams play an important role in the spatial variation of elevation. There are 13 supraglacial streams developed on the surface of this glacier， and the average offset of the streams to the northwest from 2020 to 2021 is about 2 m. Downward and lateral erosion of the supraglacial streams resulted in causing significant spatial differences ice elevation changes at the glacier terminate.

In the context of climate warming， fencing is a common management method for grassland ecosystems. Understanding the effects of warming and fencing on soil organic carbon accumulation is very important to evaluate the carbon source and sink roles of grassland ecosystems. In this study， using the amino sugars as biomarkers， we investigated the effects of long-term warming and fencing on necromass accumulation in an alpine grassland. We collected soil samples from plots after 15 years treatment of warming and fencing. The results showed that： （1） Compared to the control， there was no significant difference in the carbon source derived from bacteria and fungi between the warming + enclosure and enclosure treatments， but bacterial residual carbon contributed more to soil organic carbon than fungal residual carbon. （2） Compared to the control， there was no significant effect on the total amount of amino sugars between the warming + enclosure and enclosure treatments， but significantly reduced the long-term turnover of microbial carbon in the soil， and the short-term turnover was not significant. （3） Total amino sugars， fungal necromass carbon and bacterial necromass carbon were significantly positively correlated with conductivity， moisture content， soil organic carbon （SOC）， and total nitrogen （TN）. In summary， bacterial necromass carbon contributed higher proportions to soil organic carbon than fungal necromass carbon， while 15 years warming and fencing had no significant effects on microbial necromass carbon in soils.

Through the collection of soil samples， this study analyzed the characteristics of stable isotope variations and their spatial distribution in soil water at different depths in the source region of Yangtze River. The sources of soil water were determined by the study using an end-element mixed runoff partition model. The findings revealed that within the range of 0~100 cm， soil temperature exhibited a decreasing trend， while the region of 0~80 cm and the range of 80~100 cm showed a rising tendency in soil moisture content. Within the depth range of 0~100 cm， the δ²H and δ¹⁸O values of soil water varied， and their trends were similar. The range of variation for the soil water’s δ²H and δ¹⁸O values was， respectively， -151.273‰ to -63.605‰ and -21.052‰ to -8.676‰. Spatially， soil water’s δ²H and δ¹⁸O content was often enhanced in the western and depleted in the southeast. The evaporative line （EL） was represented by δ²H = 7.23δ¹⁸O-5.27 （R²=0.88）. The altitude gradients and soil moisture content had an impact on the stable isotope composition of soil water. The contribution of precipitation to soil moisture increased with increasing altitude gradient， and it was greater than the input of meltwater from ice melt water. Precipitation accounted for 70% to 94% of the soil’s water content between 4 150~4 777 m above sea level. Between 4 150~4 555 m above sea level， ice melt water contributed 8% to 30% of soil water； between 4 613~4 777 m above sea level， it reduced to 6% to 8%. In an effort to improve our understanding of soil water circulation and stable isotope research in the source region of Yangtze River， this study provides a theoretical basis.

To investigate the spatial characteristics of total organic carbon （TOC） content in sediment in upper and tributary streams of the Yarlung Zangbo River （YZR） and its influencing factors， responses of TOC content in sediment on particle size were studied and effects of land use in the watershed on spatial trends of TOC were analyzed based on the random forest analysis （RFA）. The results indicate that the mean TOC ［（6.56±6.37） g·kg^-1］ in upper reaches of the YZR is lower than that of other large- and medium-sized rivers， which ranges from 5.95 to 49.06 g·kg^-1 with a mean value of （19.77±14.05） g·kg^-1. With the lowest TOC in sediment ［（2.57±0.97） g·kg^-1］ in headwater， TOC is highly positively correlated （r=0.64， P<0.01） with clay content in sediment in the YZR. Based on literatures and results of this study， a highly significant exponential correlation between TOC in sediment and the content of particles smaller than 63 μm （r=0.77， P<0.01） was found across rivers， and thus， the content of particles smaller than 63 μm could be an indicator of TOC in sediment in rivers. Furtherly， RFA results indicate that in upper reaches of the YZR， TOC in sediment is majorly terrestrially sourced from grassland （48.12%） and increased spatially along the direction of water flow due to transitioning of land use from low-coverage grassland and bare land in headwater to medium， high-coverage grassland， as well as to agricultural， pastoral， and industrial （Class M1） land in upper-middle and upper-end sections of the YZR. We， therefore， concluded that the lower TOC in upper reaches of the YZR can be generally attributed to less anthropogenic activities and predominant sandy texture.

The difference of slope aspect in high altitude permafrost region may cause the asymmetry of temperature field on the two slopes， and then cause uneven settlement and longitudinal cracks of infrastructure. At present， the research on the influence of slopes mainly focuses on the monitoring and simulation of the east and west slopes of the Qinghai-Xizang Railway， seldom on the other slopes. But the trend of the linear project in plateau may involve different directions， we cannot promise the railway always in one orientation in the linear engineering and the research on the status of the water and heat difference on other slope aspects is insufficient. In this study， to found a relationship of soil temperature and moisture content in Qinghai-Xizang Plateau， and study the influence of the different slopes. A monitoring entity with eight directions （known as an octagonal platform） was built in the Huashixia permafrost station， the base of the observation of the frozen soil in Qinghai-Xizang Plateau. Soil temperature and moisture content sensors were installed on the meddle of eight slopes in （10 cm， 20 cm， 30 cm） 3 depths near the surface and the top surface. To monitor and study the impact of slope aspect differences on the state of water and heat on the slope near surface. The results shew that the difference of the near surface temperature on the east to west slope was the smallest， the monthly average temperature difference from 0.1 ℃ to 2.3 ℃， and in this difference， the maximum temperature difference occurs in May； The temperature difference near the surface of the south to north slope was the largest， with the monthly average temperature difference of 1.3 ℃ to 7.7 ℃， and the maximum temperature difference occurs in February in this temperature difference. The remained near surface temperature difference of the other two relative slopes was between the east-west slope and south-north slope， and the temperature difference of the northeast-southwest slope was smaller than that of the northwest-southeast slope. From the perspective of temperature difference near the surface of slope， the north-south， whose thermal stability of linear engineering in high altitude permafrost region is better， followed by the northwest-southeast direction， the influence of slope was not significant and the temperature field was symmetrical. Similarly， the overall difference of near surface soil moisture content of the eight slopes， whose data came from 3 depths like the near surfaced soil temperature， and the device of the sensors was the same to the soil temperature， which was the smallest in the orientation of northeast-southwest， and the maximum monthly average moisture content difference was 0.06 m³·m^-3 during the melting period； The east-west slope surface difference was the largest， and the maximum monthly average water content difference in the same period was 0.11 m³·m^-3 This difference seemed on the opposite of the soil temperature. The difference of temperature and moisture content also causes a significant difference in the number of freeze-thaw cycles on different slopes， which has an important impact on the freeze-thaw damage of slope protection materials with rubble. In this research， south slope， whose freeze-thaw cycles times was the top one and higher than the west slope to 88 times， other slopes like east and northwest were close to 60 times， north slope and northeast slope were higher than west slope and northeast was higher than north. In this time， freeze-thaw cycles time in the top surface was the third high in all the slopes. The research results have a certain guiding significance for the future plateau linear engineering planning and the treatment of the different diseases of the solar and lunar slopes of the existing projects.

Against the backdrop of global climate warming， the activity of rock glaciers in the Qinghai-Xizang Plateau has been enhanced， leading to an increasing trend of disasters. By using methods such as remote sensing image analysis， field investigation， meteorological and runoff monitoring， characteristic data of the Wulian rock glacier in southeastern Qinghai-Xizang Plateau were obtained， and the triggering factors and processes of the formation of debris flows were analyzed. The findings indicate that the Wulian rock glacier’s mean velocity ranged from 1~3 m·a^-1 between 2010 and 2017， with the frontal velocity being substantially higher than that of the middle and rear. This advancement and associated instability of the rock glacier front can be attributed to the channel transportation of loose material sources. Notably， the rock glacier transported approximately 5 600~9 000 m³ of loose material， which triggered debris flow events in 2013 and 2014. The generation of such flows requires not only abundant loose sources but also sufficient water to initiate. In this study， the maximum runoff occurred during May and June， primarily driven by the melting of ice and snow resulting from elevated temperatures. By examining the two debris flow events， it was discovered that high temperature was the primary contributing factor， and the disaster mechanism involved the rapid melting of surface ice and snow due to elevated temperatures， causing the accumulation of meltwater in depressions on the rock glacier surface， resulting in a pond that， when discharged， caused loose material in front of the rock Glacier to form debris flow. As climate warming exacerbates in the future， rock glacier activity is anticipated to increase. Additionally， under the influence of rapid warming or heavy rainfall， the scale and frequency of debris flows may increase. Therefore， it is imperative to mitigate the risk of rock glacier disaster formation through debris flow prevention measures.

The content variation of trace elements in mountain glaciers can be used as a good index to evaluate the impact of human activities on atmospheric environment. In order to illuminate the spatial distribution and the main sources of trace elements in snow ice of Qilian Mountains， on September 15， 2020， surface snow samples were collected from seven glaciers in the Qilian Mountains， including Aerjin Glacier， Zhazigou Glacier， Laohugou Glacier， Qiyi Glacier， Glacier No.4， Bayi Glacier， and Ningchanhe Glacier No.3. The samples were acidified to analyze the spatial distribution of trace elements and their primary sources. Using ICP-MS， the concentrations of 15 metallic elements （Al， Fe， As， Ba， Co， Cr， Cu， Li， Mn， Mo， Pb， Sr， Tl， Zn， and Cd） were measured， and the average trace element concentrations were compared using the Jonckheere-Terpstra nonparametric test. Higher trace element concentrations were found in Zhazigou Glacier and Laohugou Glacier in the western part of the Qilian Mountains， while lower concentrations were found in Ningchanhe Glacier No.3 in the eastern part. The overall spatial distribution trend of trace element concentrations was west>middle>east. Enrichment factors （EF） for total trace element concentrations were calculated， showing that elements Co， Cr， Cu， Tl， Fe， Li， Mn， Mo， and Sr were mostly influenced by natural dust input （EF<10）， whereas elements Pb， Cd， and Zn were more influenced by anthropogenic sources （EF>10）， such as industrial emissions， coal mining， and fossil fuel combustion. The trace element concentration data were subjected to Principal Component Analysis （PCA）， revealing that elements with high positive loadings on the total concentration in the first principal component mainly originated from natural dust sources， especially in the Xinjiang region. The second principal component represented varying degrees of anthropogenic source effects. Additionally， clustering analysis combined with a backward trajectory demonstrated that the study area is primarily influenced by the westerly circulation， and the natural input sources of trace elements mainly come from deserts and basins in central and northern Xinjiang. Anthropogenic sources include human activities in some cities along the Hexi Corridor， transportation emissions， mineral extraction， and metal smelting. This study provides data support for the impact of glacial human activities on the environment in various segments of the Qilian Mountains. These findings provide important scientific and technological support for the construction of ecological civilization and sustainable social and economic development in the Qilian Mountains.

The “lake-glacier” interaction of glacial lake contacted glaciers is significant， and they act together on the water cycle process in alpine areas， playing an important role in water conservation， runoff regulation， and ecological diversity maintenance in mountainous areas. Based on Landsat remote sensing images and ERA5-Land reanalysis data， this paper discusses the evolution of glacial lakes and the mass balance of glaciers in the glacier area of Ulugh Muztagh in eastern Kunlun Mountains. The results show that the glacial lakes in the glacier area of Ulugh Muztagh are mainly distributed at an altitude of 5 275~5 400 m a.s.l.， and they are mainly glacial blocking lakes. From 1990 to 2020， there were 16 glacial lakes. Among them， the two glacial lakes formed by the blocked lakes of glaciers 5Y624E0022 and 5Y624F0020， which had three and two events of the glacial lake outburst， respectively. Both lakes are periodically collapsing glacial lakes. The former had large-scale glacial lake bursts in 1999 and 2001， respectively. The former had large-scale glacial lake bursts in 1999 and 2001， with an area of （0.250±0.044） km² and （0.500±0.097） km² before glacier lake outburst flood （GLOF）， corresponding to a volume of （0.014±0.003） km³ and （0.026±0.006） km³， respectively； the latter is a glacier-blocked lake ［（0.110±0.030） km²］ formed in 2000， which burst in that year with a water volume of （0.006±0.002） km³. During this period， the number of glacial lakes has increased， while the area and reserves have decreased. However， the accumulative mass balance of glaciers showed an increasing trend from 2000 to 2010， and a decreasing trend from 2010 to 2020， and the average annual mass balance of glacial lake contacted glaciers showed a significant decreasing trend （-0.024 m w.e.·a^-1） slightly greater than that non-glacial lake contacted glaciers （-0.022 m w.e.·a^-1）. In addition， due to glacial thermal erosion and glacier avalanche， the locations where Binglinchuan and Muztag glacial lakes are connected to the glacier have accelerated their retreat by 0.65 km and 0.28 km， respectively. In short， the number of glacial lakes in this region is increasing and the space of glacial lakes is expanding， which accelerates the material loss of parent glaciers.

Hailuogou Glacier， located on the southeast edge of Tibetan Plateau， is a typical monsoonal maritime glacier， which is extremely sensitive to climate change. It is of great practical significance to recognize the present situation and changing trend of glaciers. In this paper， Landsat series remote sensing images are used as data sources， based on remote sensing and geographic information technology， band ratio threshold method and visual interpretation method are used to extract the glacier boundaries of five periods in 1974， 1991， 2000， 2009 and 2020. Combined with digital elevation data， the variation characteristics of glacier length， total area，coverage of debris and glacier cover units in different directions in Hailuogou in recent 46 years are analyzed. The results showed that： （1） From 1974 to 2020， the glacier length shortened by 1 087.10 m， and the average annual retreat rate was 0.15%， among which the retreat rate at the end of the glacier accelerated continuously after 2000. （2） From 1974 to 2020， the area of glaciers in the study area decreased by 0.55 km²， with the area change rate of -0.52% and the annual average retreat rate of 0.03%， among which from 2000 to 2009， the largest decrease was about 0.23 km²， the area change rate of -0.93% and the annual average retreat rate of 0.10%. （3） Glaciers in different directions are also shrinking in different degrees， and the area of glaciers in the east direction is shrinking fastest. The distribution of glaciers is characterized by more in the east and less in the south and less in the north， and there is a positive correlation between the area distribution of glaciers in each orientation and the total shrinkage of corresponding orientation areas from 1974 to 2020. （4） Through comparative analysis， it is found that the debris in the northwest， northeast and southwest of Hailuogou Glacier increases after 2000 and further increases after 2009. It is predicted that the debris cover of Hailuogou Glacier will continue to expand in the future.

The water temperature of glacial lakes is the basis for studying the physical， chemical， biological and hydrodynamic processes of glacial lakes， as well as the material and energy exchange among glaciers， glacial lakes and moraine dams. In addition to the influence of solar radiation on the water temperature of the glacial lake， the melt water impounded by the glacial lake aggravates the complexity of the temporal and spatial variation of the water temperature of the glacial lake. Moreover， the change of the lake water temperature affects the temperature field and water field of the moraine dam through the heat exchange interaction between the lake water and the dam body， and therefor affects the stability of the dam body. Thus， it is of great significance to study the ablation of lake-terminal glaciers and the stability of moraine dams by establishing long-term field monitoring of lake water temperature and deeply analyzing the characteristics of water temperature changes in glacial lakes. Based on the data of preglacial lake temperature， solar radiation and water temperature at 10~200 cm depth obtained by the automatic observation station of Longba Saba Lake （27°57′17″ N， 88°04′55″ E， 5 499 m a.s.l.） from 2012 to 2021， this paper discussed the characteristics and influencing factors of water temperature change in preglacial lake. The results show that the changes of water temperature and annual freezing period of Longbasaba Lake is the result of many factors， such as air temperature， solar radiation intensity and glacier meltwater， among which the inflow of glacier meltwater has the most significant influence on the variation of water temperature of glacial lakes. In summer， due to the influence of a large amount of glacier meltwater， the water temperatures at different depths are not much different， and the average temperature is about 4 ℃. The water temperature of the observed 10~200 cm water depth varied little （the temperature difference is less than 0.2 ℃）， and there is no obvious stratification in the mixed state. The maximum water temperature appeared in August or September， and the temperature peak had a lag of 1-2 months to the peak of air temperature. In general， the solar radiation is reduced and the water temperature decreases at night. However， due to the influence of hydrodynamic mixing caused by glacier meltwater or floating ice on lake， it is found that at about 12 p.m.， the observed temperature rises at 10~200 cm depth （about 1~2 ℃） or hinders the cooling process of the lake water， forming the phenomenon of inverse stratification at night. Moreover， the frequency of this reverse stratification phenomenon at night also increases with the increase of water temperature. When the water temperature is > 4 ℃ throughout the day， the water temperature of the glacial lake is mainly affected by the solar radiation intensity and weather conditions of daytime， so that the surface water is stratified in days （the temperature difference between the lake water at a depth of 10 cm and 100 cm is greater than 1 ℃）， generally lasting for 2~6 h. In winter， the temperature of lake water at different depths varies greatly， and the temperature difference between 10 cm and 100 cm depths is about 1~7 ℃， and the lake water temperature shows obvious stratification feature. The annual freezing period of Longbasaba Lake is mainly affected by summer water temperature and glacier meltwater recharge. It is about 200 days from late October to late May of the next year. The thickness of lake ice changes significantly， and the thickness of lake ice is the thickest in February， generally between 100 cm and 200 cm. The inflow of glacier meltwater in summer inhibits the rise of lake water temperature to a certain extent， reducing the daily temperature difference of lake water temperature， and affecting the internal temperature field of dam body through the heat exchange between lake water and dam body， consequently affecting the melting process of buried ice and frozen soil in dam body.

Glaciers， as an integral part of the cryosphere， are highly susceptible to both local and global climate change. Ice crevasses， which are prominent features on the surface of glaciers and the important channels for glacier meltwater， play a crucial role in understanding the condition， stability， internal stress and mass balance of glaciers. Mountain glaciers are subject to cloud cover and area limitation， and the spatial resolution of traditional satellite remote sensing data is low， which is difficult to be used for extracting ice crevasses， so there are fewer studies related to ice crevasses on mountain glaciers. In this study， the objective was to address the challenge of identifying and extracting glacier crevasses quickly and accurately. This research takes the mountain glacier： Baishui River Glacier No.1 in Yulong Snow Mountain in Lijiang， Yunnan Province as the research object， and takes the cloud-free orthophotos of the glacier surface with a resolution of 0.12 m in 2021 and 2022 acquired by aerial photography of the DJI M300RTK drone as the data source， and applies the U-Net Deep Learning Network to carry out the extraction of ice crevasses of the Baishui River Glacier No.1.

The results demonstrate that the U-Net network outperforms traditional methods such as the Canny operator and SVM algorithm in terms of crevasse extraction accuracy. The overall accuracy can be as high as 93%. Furthermore， the U-Net network exhibits strong generalization capabilities， which can be used to automatically extract unmanned aerial imagery from different time periods. From the perspective of spatial distribution of ice crevasses， the crevasses observed on BRG1 predominantly consist of transverse crevasses， splaying crevasses， and En échelon crevasses， which show the typical characteristics of mountain glacier ice crevasses in the low advection lifecycle. As the altitude decreases， there is a gradual transition from transverse crevasses to splaying crevasses. From the perspective of temporal change of ice crevasses， comparing the extraction results from different time periods reveals an increase in the number and average length of crevasses. This proves that the ablation of BRG1 is intensifying， and the glacier mass is gradually losing. The orientation of the ice crevasses was almost unchanged， indicating that the stress inside the glacier didn’t change dramatically. In summary， the study of intelligent extraction of ice crevasses based on UAV images and deep learning methods creates new possibilities for extracting ice crevasses from mountain glaciers， and can provide technical support for monitoring glacier changes and their relationship with climate change.

In recent years， with the advancement of the western development strategy， China has undertaken extensive rock engineering construction in alpine and cold mountainous areas. During the construction process， inevitable engineering disasters such as slope instability and tunnel collapse have significantly compromised engineering safety. These disasters are usually caused by the gradual accumulation of damage and failure of rock under external load. In addition， low-temperature environmental factors will lead to the change of mechanical properties of rock， and then affect the load-failure process. Damage localization is the necessary stage and precursor of rock failure under load. In this paper， the damage localization of frozen rock under load is studied， and the influence of initial saturation degree is discussed. Uniaxial compression tests of frozen sandstone specimens with different initial saturation degrees were carried out at -20 ℃. The digital image correlation （DIC） and acoustic emission （AE） systems were used to collect surface deformation and AE signals during the test. Based on the mechanical test results， the mechanical and deformation characteristics of frozen sandstone are analyzed， and the influence of initial saturation degree on the mechanical properties of rock is demonstrated combined with the change of AE count distribution characteristics. Through the analysis of surface deformation， the surface strain localization mode of the specimen is divided into mode I （tensile strain localization）， mode II （shear strain localization） and mode III （mixed localization）. The initial saturation degree of 40% is the turning point from mode I to mode II， and mode III appears after the initial saturation degree exceeds 90%. All the modes can correspond to the final failure mode of the specimens. Based on the RA-AF analysis， the types of microcrack propagation in the rock during the strain localization process are shown： type I （tensile crack）， type II （shear crack） and type III （mixed crack）， which are basically consistent with the surface strain localization mode. Finally， combined with the changes of phase composition in the pores of frozen rocks at different initial saturation degrees， the influence of phase composition on the damage localization mode of frozen rocks is analyzed. That is， at low initial saturation degree， the ice content in the pores of frozen sandstone is small， and the external load is mainly borne by the rock skeleton. Because the tensile strength of the rock is usually lower than the shear strength， the rock will be tensile failure， and the damage localization mode is also tensile. At medium saturation degree， the ice content in the pores increases， and cements with the rock skeleton. The tensile strength of the ice helps the frozen sandstone resist tensile failure， and the damage localization mode becomes shear. At high initial saturation degree， the pore ice completely fills the microcracks， forcing them to expand， and a mixed damage localization mode is presented. This study helps to deepen the understanding of the mechanical properties of frozen rocks， and can provide important reference for the prediction of failure and instability of rock engineering in cold regions.

Global changes have led to increased temperatures and accelerated glacier retreat， and a large number of radioresistant-antioxidant microbial resources have evolved at the glacier foreland. As one of the important taxa influencing the successional process of glacier frontiers， the study of radiation-antioxidant bacteria in the newly melted moraine habitats in the ice tongue area of glacier frontiers is relatively rare. Based on the phylogeny of 16S rRNA gene sequences， this study not only investigated the diversity of culturable bacteria in moraine habitats in the glacier tongue at the frontier of the Lahugou No.12 Glacier， but also screened and evaluated the strains for their radiation-resistant and antioxidant capacity. The study showed that 259 bacterial strains isolated in the study area belonged to Actinobacteria， Proteobacteria， Bacteroidetes， Firmicutes and Deinococcus-Thermus， among which the highest number of strains was found in Actinobacteria， followed by Proteobacteria>Bacteroidetes>Firmicutes>Deinococcus-Thermus. In terms of species diversity， Actinobacteria and Proteobacteria had the highest species richness. TN， TOC， WC and pH were the main factors affecting the structure of culturable bacterial communities. The strains with D₁₀ （lethality of 10%） dose of UVC irradiation intensity higher than 100 J·m^-2 accounted for 94.9% of the total culturable bacteria， and the strains with D₁₀ dose of H₂O₂ tolerance concentration higher than 10 mmol·L^-1accounted for 100% of the total culturable bacteria； among them， there were 20 strains. And the survival rate after oxidative stress of the radiation-resistant strains was above 90%. In addition， the strains with higher survival rate than Deinococcus-radiodurans R1 after oxidative stress were all radiation-resistant strains with higher than 50% survival rate after 100 J·m^-2 UVC irradiation. This study can not only provide a theoretical basis for the diversity and ecological adaptation of bacteria in the glacier foreland environment， but also provide a rich resource of radiation-resistant and antioxidant glacier bacteria for the subsequent research on the protective mechanism of irradiation and oxidative damage.

As one of the major alpine ecosystems， the carbon-water balance status of alpine meadows during the non-growing season is of great significance to the whole ecosystem. However， little is known about the environmental drivers of carbon and water fluxes during this season. Therefore， this study investigates the carbon and water fluxes and their influencing factors during the non-growing season （from Nov. 2022 to Apr. 2023） in alpine meadows in the Qilian Mountains based on eddy observations and meteorological gradient towers. It was found that in the alpine meadow ecosystem of the Qilian Mountains， during the entire non-growing season， the net ecosystem carbon exchange （NEE）， gross primary productivity （GPP）， and ecosystem respiration （Reco） were -18.0574 mg CO₂⋅m^-2， 27.3565 mg CO₂⋅m^-2， and 9.2991 mg CO₂⋅m^-2， respectively. The total evapotranspiration （ET） was 74.8762 mm， which was 15.0762 mm lower than the total precipitation， and the water cycle of this ecosystem was relatively balanced. Normalised vegetation index， photosynthetically active radiation and soil temperature were the main factors affecting net ecosystem carbon exchange during the non-growing season. Whereas relative humidity， precipitation and net radiation were the main factors affecting evapotranspiration in the non-growing season. The results of this study revealed the characteristics of carbon dioxide fluxes and water fluxes during the non-growing season in alpine meadows in the Qilian Mountains and their influencing factors， which provide an important reference for understanding the carbon balance process and water balance in the region and its response to climate change.

Glaciers are very sensitive to climatic and environmental changes， and are closely linked to changes in water resources and geological disasters. Under the influence of global warming， the melting of glaciers on the Qinghai-Xizang Plateau and its surrounding areas is on the increase. This paper addresses the problem that the large difference in the quality of ICESat-2 laser photon data under different terrain conditions leads to a large error in the glacier elevation change volume estimation when monitoring glacier elevation change using ICESat-2 satellite photon point cloud data， using Qiaqing Glacier， China’s largest oceanic glacier on the Qinghai-Xizang Plateau， as an example. Using ICESat-2 laser altimetry data and 30 m resolution SRTM DEM products， we propose a multifunction fitting glacier elevation change estimation error correction model with a robust estimation criterion that reduces the glacier elevation change estimation error caused by terrain slope； then， we estimate the elevation change and mass change of Qiaqing Glacier between 2000 and 2021. The results show that the multi-function correction models proposed in this paper can effectively improve the accuracy of the elevation change monitoring results. Meanwhile， the correction effect of the function model incorporating robust estimation is better compared with the corresponding traditional least squares estimation results. Subsequently， the effectiveness and feasibility of the error correction model can be demonstrated by cross-validating the correction results with multiple data. The model can effectively improve the accuracy of extracting glacier elevation and mass change information， and it was concluded that the glacier elevation change rate of the study area during the 21-year period was approximately （-0.52 ± 0.56） m⋅a^-1， and the glacier mass change was about -1 277.38 million tonnes. In addition， combined with the meteorological data， the annual and monthly average temperature and precipitation data are analyzed， and the overall temperature rise and precipitation fluctuation decline were the main reasons for glacier melting in 21 years. Combined with the data of glacier elevation changes in different seasons， it was also analyzed that spring and winter were the accumulation periods of glaciers， while summer and autumn were the melting periods of glaciers. At the same time， it can be predicted that the glacier will continue to melt in the short term in the future. The situation of continuous degradation and melting of the glaciers is not optimistic.

Two-phase closed thermosyphons （TPCTs） play a significant role in permafrost regions due to their thermal semiconductor effect. Many TPCTs have been used to collect cold energy from the ambient air in permafrost regions to cool the underlying stratum. In this study， an ammonia-steel TPCT test bench suitable for permafrost regions was built to optimize the working conditions of the TPCT. The effects of filling rate and inclination angle on the heat transfer performance of low-temperature TPCTs were analyzed. Experiments were carried out at different filling rates （20%， 30%， 40%） and inclination angles （10°， 30°， 50°， 70°， 90°） under negative temperature conditions. To simulate low-temperature conditions， the ambient temperature difference between the evaporator section and the condenser section was set to 13 ℃. The heat transfer performance of the TPCT was evaluated based on the temperature distribution inside and outside the TPCT and the variation in the heat flux density of the outer wall. The heat transfer performance of the TPCT was evaluated using indexes such as isothermal characteristics， thermal resistance， and transfer efficiency. The experimental results show that， under three experimental filling rates， all TPCTs with an inclination angle of 10° have the best axial internal isothermal performance. The minimum thermal resistance occurs at an inclination angle of 50° when the TPCT filling rates were 30% and 40%， corresponding to the highest heat transfer efficiency. For the TPCT with a filling rate of 20%， the minimum thermal resistance and the highest heat transfer efficiency both occur at an inclination angle of 30°. Overall， the TPCT with a filling rate of 30% has the optimal heat transfer performance. It is recommended that the inclination angle of TPCTs is between 10° and 50°. This study provides meaningful reference for the best structural parameters in the design of TPCTs for cold regions engineering.

The climate change in the source region of the Yellow River Basin was projected from 2121 to 2060 using eight GCMS of the Coupled model inter-comparision project Phase 6 （CMIP6） and two shared socio-economic low-carbon paths （SSP1-2.6 and SSP2-4.5）. The inter-decadal variation of discharge in the source region of the Yellow River from 2021 to 2060 was predicted by using eight GCMs climate models to drive HBV and SWAT hydrological models. The results show that： （1） In comparison with the baseline period （1995—2014）， in terms of the ensemble mean， annual mean air temperature， annual precipitation will increase by 1.3 ℃ and 1.6 ℃， by 11.6% and 11.5% under SSP1-2.6 and SSP2-4.5 scenarios， respectively. And the warming and wetting trends become obvious from 2021—2060 under the SSP1-2.6 and SSP2-4.5 scenarios in the source region of the Yellow River. （2） The multi-models ensemble mean （MEM） multi-year mean annual discharge for 2021—2060 was expected to increase by 8.6% and 8.5% under the SSP1-2.6 scenario and the SSP2-4.5 scenario， respectively. It was projected to increase in each decade from the 2020s to the 2050s under the scenario of SSP1-2.6 and SSP2-4.5. The increase degree in 2040s and 2050s is higher than 2020s and 2030s. （3） The proportion of discharge was projected to decrease by 0.1%~1.0% from May to August during 2021—2060 under both scenarios but was projected to increase by 0.1%~2.1% from April to May and September to December for most GCM_S under the SSP1-2.6 scenario and the SSP2-4.5 scenario， respectively. （4） Extremely high monthly discharge was expected to increase by 2.5%~2.7% in the water storage season under two scenarios during 2021—2060 in the source region of the Yellow River whereas the extreme high discharge in the flood season increased by 0.1% under the SSP1-2.6 scenarios but decreased by 1.3% under the SSP2-4.5 scenarios， and decreased by 0.7%~1.0% in dry season under two scenarios in the next 40 years. Extremely low discharge was projected to increase by 0.8%~1.9% in the flood season and the dry season but to decrease by 1.8%~2.3% in the storage season under two scenarios in the study region.

The effect of pipeline insulation layer in permafrost area is closely related to the randomness of evaluation index changes， and the previous evaluation model of pipeline insulation layer effect in permafrost area has certain limitations and shortcomings. In order to improve the application effectiveness of insulation measures for pipelines in permafrost regions， based on the comprehensive evaluation method and combining the Driving force-State-Response （DSR） model theory， coefficient of variation weight theory and cloud model theory， in this paper a suitable comprehensive cloud model has been built for the evaluation of the thermal effect of insulated pipelines in permafrost regions. The DSR framework is used in the developed model to ensure the reliability of the index system. The weight of the index calculated by the coefficient of variation method ensures objectivity； the cloud model solves the randomness problem that cannot reflect the standard classification and index data acquisition of the environmental evaluation process of pipelines in permafrost regions. In order to validate the evaluation model of insulating effects for buried oil pipelines in permafrost regions， three representative monitoring sites at Kilometerage Posts MDX007， MDX113， MDX304 and MDX364 along the China-Russia Crude Oil Pipelines （Mo’he-Daqing Line or MDX） are taken as examples. The comprehensive cloud model is used for evaluating the thermal effect of pipeline insulation measures. The results show that the comprehensive model-based analysis can effectively guide the implementation plan of insulation measures for Russian-Chinese Crude Oil Pipelines， and give corresponding measure suggestions based on the corresponding insulation effect failure types. Based on the evaluation results， the implementation plan of insulation measures for the China-Russia Crude Oil Pipelines is analyzed and advised for optimizing the thermal effect of insulative measures. The comprehensive cloud model can combine the advantages of different theoretical models to take into full account of the effectiveness and soundness of the comprehensive analysis of cold regions engineering environment evaluation， with a promising application potential. The research results can provide strategic scientific support for the safe and stable operation and sustainable development of China-Russia Crude Oil Pipelines.

In high-altitude and high-altitude areas， the stability of steep mining slopes containing years of frozen layers is controlled by the mechanical properties of frozen rock layers. Excavation of the strata exposes the frozen rock layers to the air， and coupled with blasting vibrations or mechanical disturbances during excavation， the frozen rock layers gradually soften and heat melt， leading to a decrease in slope stability. With the continued global climate warming， the thermal thawing softening of frozen rock layers accelerates， further exacerbating the risk of instability in mining slopes. The strength of frozen rocks will soften during the hot melt process， which is the most susceptible stage to failure. Studying the thermal thawing softening law of frozen rocks is crucial for evaluating the stability and safety of frozen strata during the thawing process. This article conducted uniaxial compression tests on frozen rocks at different melting temperatures. Based on the restoration of rock pore structure and precise calibration of microscopic parameters， the particle flow software （PFC^2D） was used to simulate the compression failure process of frozen rocks. Based on the analysis of the initiation law and propagation law of microcracks， this paper explores the control effect of pore ice on the thermal thawing softening law of frozen sandstone. The research results indicate that： （1） the strength， elastic modulus and other parameters of frozen rock show a two-stage trend with the increase of temperature. There is a certain temperature between -4 ℃ and -2 ℃， which causes a sudden decrease in the strength and deformation parameters of the sample. （2） As the temperature increases， the failure of frozen rocks at peak stress gradually shifts from being dominated by mineral particle frameworks to being dominated by ice. When the temperature is less than -2 ℃， the degree of damage to the frozen rock skeleton is higher； When the temperature is greater than -2 ℃， the damage to porous ice is more significant. When the temperature is below -15 ℃， the initiation and propagation of microcracks are mainly controlled by the contact strength between mineral particles； When the temperature is between -2 ℃ and -15 ℃， it is mainly controlled by the contact strength between ice particles and ice minerals； When the temperature is greater than -2 ℃， it is mainly controlled by the contact strength between ice particles. （3） By analyzing the supporting and bonding effects of pore ice during the load failure process of frozen rocks， it was found that between -6 ℃ and -4 ℃， the bonding strength between ice particles and between ice minerals rapidly decayed， leading to a weakening of the supporting and bonding effects of ice， which is the essential reason for the rapid weakening of mechanical properties in this temperature range. For high-altitude and high-altitude areas， the thermal thawing softening of frozen rock layers is a core process related to their stability and safety. Therefore， studying the thermal thawing softening law and load failure process of frozen rocks is of great engineering significance.

Frost heave of the unsaturated coarse-grained soil is the main disaster cause for the subgrade engineering in seasonally frozen soil regions， which is affected by the complex coupling of water， temperature and stress field. In order to investigate the frost heave deformation properties of unsaturated coarse-grained soil， a numerical model is proposed to describe the water vapor migration in unsaturated coarse-grained soil under temperature change based on Darcy’s law and Fourier’s law， which is solved by the coefficient partial differential equation of the finite element software COMSOL Multiphysics. The model comprehensively considers the influence of ice-water phase change and water vapor migration on the deformation of frozen soil， describing the hydrothermal vapor and mechanical effect of unsaturated frozen soil. The model introduces the dynamic relationship among pore ice content， soil negative temperature and water migration， and considers the influence of negative temperature on the soil water migration. The rationality and feasibility of the model is verified through the comparisons of calculated temperature， water and displacement filed with the indoor experimental results. The effect of vapor on water migration in the freezing process is comprehensively analyzed by analyzing gaseous and liquid water fluxes changes. The variation characteristics of water vapor migration and frost heave deformation of coarse-grained soil under different temperature gradients are further investigated by numerical experiments. The results found that under the action of uni-directional freezing， the temperature change process is divided into three stages： rapid cooling stage， slow cooling stage and stable stage. The water content reaches the extreme value at the freezing front. In the early stage of freezing， the frost heave rate is large， and the soil has a large deformation. As time increases， the deformation basically remains unchanged. During the freezing process of coarse-grained soil， in the unfrozen area （height 0~9 cm）， the temperature fluctuation is large， and the gaseous water flux is much smaller than the liquid water flux， indicating that in the unfrozen area， liquid water migration is dominant. In the range of freezing depth， the gaseous water has an obvious upward migration trend. The reason is the ice formed by the migration of liquid water here has little blocking effect on the migration of gaseous water. The gaseous water can continue to migrate upward through the ice layer， which has an important contribution to the frost heave deformation of the soil. Therefore， in the freezing process of unsaturated coarse-grained soil， there is an obvious gaseous water migration， which cannot be ignored. Through numerical experiments， it is found that different temperatures at the cold end have significant effects on soil sample temperature， water content and frost heave deformation. Under the action of temperature gradient， gaseous water moves upward， resulting in the migration of water within the soil towards the frozen area， resulting in frost heave deformation. The larger negative temperature will accelerate the movement speed of gaseous water flux and the freezing front significantly. Therefore， the frost heave deformation of the soil gradually increases with the decrease of cold end temperature. Under different cold end temperatures， the sample with a height of 20 cm reaches the maximum water content at a height of about 3 cm， and the maximum frost heave deformation on the surface can reach about 4 cm. The trend of the calculated value and the measured value is almost the same and the curve is in good agreement， which shows that the proposed model is suitable to simulate the phenomenon of soil frost heave. This model is beneficial for both the theoretical research and numerical implementation of the subgrade engineering in seasonally frozen soil regions.

Taking the unified strength theory as the yield criterion of frozen surrounding rock of tunnels in cold regions under the unloading state， taking into account the influence of the non-homogeneous of surrounding rock and the intermediate principal stress effect on the strength of frozen surrounding rock， an elastic-plastic mechanical model of stress displacement of tunnels in cold regions is established， and combining with the boundary conditions of each region， the elastic solution， the plastic unified solution and the implicit equation of the radius of the plastic zone under the homogeneous and non-homogeneous state of frozen surrounding rock are calculated， discussion and analysis of stress and displacement fields respectively. The research shows that considering the heterogeneity of frozen surrounding rock， the peak value of hoop stress in plastic zone increases by 40%， the range of plastic zone decreases by 40.4%， the displacement of inner wall decreases by 9.3%， the elastic ultimate bearing capacity increases by 41%， and the plastic ultimate bearing capacity increases by 14%， which has a significant impact. The intermediate principal stress effect can give full play to the bearing potential of frozen surrounding rock， the calculated bearing capacity is significantly increased， and the plastic radius is significantly reduced. The results can provide theoretical guidance for tunnel excavation and support design and numerical simulation in cold regions.

To explore the influence of freeze-thaw cycles on the strength of moraine soil， this study established a prediction model of the strength deterioration of moraine soil under different numbers of freeze-thaw cycles. Taking the moraine soil in the western high-cold and high-altitude area as the research object， a total of more than 180 triaxial samples with a diameter of 101 mm were made by combining the indoor high-frequency freeze-thaw cycle test with the unconsolidated undrained triaxial test. The strength test was carried out after the high-frequency freeze-thaw cycle test with different frequencies. The test variables were the number of high-frequency freeze-thaw cycles， the initial water content of the sample， and the confining pressure of the triaxial test. The influence of the number of high-frequency freeze-thaw cycles， the initial water content， and the confining pressure on the failure strength of the moraine soil sample was mainly explored. A BP neural network prediction model with three layers of input and one layer of output was established by MATLAB software. The data obtained from the experiment were imported into the BP neural network model for learning and training. The strength degradation prediction model of moraine soil under high frequency freeze-thaw cycle was established. At the same time， the SSA algorithm was introduced to improve the model. The number of neurons， the number of neural network layers and the number of iterations in the BP neural network were optimized to improve the prediction accuracy of the model. The experimental results show that freeze-thaw cycles have a great influence on the strength of moraine soil. Under different testing conditions， the strength of moraine soil degrades in the range from 30% to 40%， and tends to stabilize after 15 to 20 freeze-thaw cycles. Based on the SSA-BP neural network model， the complex nonlinear relationship between the failure strength of moraine soil and the number of freeze-thaw cycles， initial water content and confining pressure can be reflected. The prediction error is significantly smaller than that of the BP neural network model， and the prediction results are more accurate. Comparing the SSA-BP neural network model with the Logistic model used in linear fitting， it is found that the two are in good agreement， which further illustrates the accuracy of the SSA-BP model. Through later verification， it is found that the output error of the model is small and has certain application prospects. This research can provide important parameters for engineering stability evaluation in western cold and high-altitude areas.

The freeze-thaw front within active layer is the interface between the frozen and the unfrozen soil layers during the freeze-thaw process， and the hydrothermal parameters of the soil layers on both sides of freeze-thaw front are significantly different. Therefore， the accurate simulation of the freeze-thaw front movement in the land surface model is essential to improve models both in simulating the hydrothermal characteristics of permafrost and simulating the energy-water balance of the land surface. In this study， the simulation depth of the Noah-MP land surface model was extended to 20 m， and the 4 soil layers of the Noah-MP land surface model was increased to 19 soil layers， and the organic matter scheme and vegetation root scheme were introduced. After these modifications， in order to strengthen the ability of the Noah-MP land surface model on simulating freeze-thaw front， the Stefan method was coupled. Then， the simulation effect of the augmented Noah-MP land surface model on the hydrothermal process of the Xidatan permafrost site was evaluated. Two experiments， CTL experiment （coupled Stefan method） and STE experiment （not coupled Stefan method）， were conducted to simulate the soil temperature and soil liquid water content of 0~20 m in 2012， and the simulation results were verified by the observed daily soil temperature and soil liquid water of 0~3.2 m and the observed yearly ground temperature of 3 m， 6 m and 10 m. The results showed that the freeze-thaw front （0 °C isotherm） obtained by interpolation of soil temperature simulation values had obvious step-like characteristics， and its maximum freeze-thaw depth was larger than the measured. Coupling Stefan method enhanced the ability of Noah-MP model to simulate the freeze-thaw front， so that the model was able to better simulate the change trend and maximum depth of the freeze-thaw front. At the same time， coupling Stefan method also improved the simulation of soil temperature. The mean RMSE and the mean MBE of the soil temperature in the soil layers of 0~3.2 m decreased to 0.89 ℃ （decreased by 44%） and -0.13 ℃ （decreased by 86%） respectively， and yearly ground temperature of 3~20 m was closer to the measured. And it also improved the simulation of the soil liquid water content. The mean RMSE and the mean MBE of the soil liquid water content in the soil layers of 0~3.2 m decreased to 0.06 m³·m^-3 （decreased by 33%） and -0.01 m³·m^-3 （decreased by 67%） respectively， and the soil water melting time of 20 cm， 40 cm， 80 cm and 120 cm in the active layer was closer to the observed. It can be seen that coupling the Stefan method that can better model the movement process of freeze-thaw front in the land surface model can greatly improve the simulation ability of the model， which is one of the effective ways to improve the land surface process model. The results of this study can provide a reference for improving the simulation of the land surface model in the permafrost area. This study will provide a reference for improving the ability of land surface model to simulate hydrothermal processes of permafrost.

This paper investigates the impact of floating beads on temperature dissipation during the construction stage and their influence on the effective compaction time of asphalt pavements. As a high thermal resistance material， floating beads have been used in asphalt pavements in permafrost region to reduce heat absorption and mitigate thawing induced settlement during the operational stage. However， limited research has focused on the influence of floating beads on temperature dissipation during the construction stage and their impact on the effective compaction time of asphalt pavement. The addition of floating beads alters the thermo-physical properties of the asphalt mixture， thereby affecting pavement compaction and heat transfer to the frozen soil subgrade. During the compaction process of asphalt mixtures， the material undergoes a transition from a loose state to a dense state， resulting in decreased porosity and increased strength. Consequently， the thermo-physical parameters of the asphalt mixture dynamically change throughout the construction stage of asphalt pavement. This study considers the asphalt mixture modified with floating beads as a composite material comprising coarse aggregates， asphalt matrix， and air. Laboratory tests are carried out to measure the thermo-physical parameters of the three components in the modified asphalt mixture. The Williamson formula is utilized to calculate the thermo-physical parameters of the modified asphalt mixture in both dense and loose states. To analyze the temperature field in the asphalt pavement and the heat flux entering the soil subgrade， a finite element model is developed. The model takes into account different floating bead contents and paving thicknesses. Based on the numerical results， a multivariate regression analysis is performed to identify the key factors influencing temperature fields during construction stage of asphalt pavement. This analysis leads to the derivation of an empirical equation for estimating the temperature of the asphalt mixture. The research findings reveal that the inclusion of 15% volume of floating beads reduces the thermal conductivity of the asphalt mixture by 78%， while exhibiting negligible changes in specific heat capacity （i.e.， less than 1%）. With an increase in air temperature 0 °C，10 °C，20 °C，30 °C， the addition of 15% floating beads enhances the effective compaction time by 21.1%， 20.2%， 17.5%， and 16.8% in the dense state， and by 21.6%， 21.2%， 20.3%， and 18.7% in the loose state. Compared to the dense state， the effective compaction time in the loose scenario is approximately 1.7 to 2.1 times longer. Hence， in low-temperature conditions， re-compaction and final compaction are crucial to be promptly and continuously completed to achieve the specified compaction level within the designated timeframe. Furthermore， the inclusion of 5%，10%，15% volume of floating leads to a reduction of 1.9%， 3.1%， and 6.4% in heat flow to the subgrade in the dense state， and a reduction of 3.5%， 7.3%， and 12.8% in the loose state. These results demonstrate the benefits of adding floating beads in mitigating heat disturbance to the soil subgrade during the construction stage. By minimizing heat flow， the potential adverse effects on the frozen soil subgrade are significantly reduced， contributing to the overall durability of the asphalt pavement. In conclusion， the addition of floating beads alters the thermo-physical properties of asphalt mixtures， thereby influencing the compaction process and heat transfer dynamics. The findings in this study provide valuable insights for optimizing construction practices of asphalt pavement， including adjusting compaction procedures and ensuring appropriate timeframes for effective compaction. Moreover， the benefits of floating beads in reducing heat flow to the subgrade indicate their potential for enhancing the long-term performance of asphalt pavements in challenging environmental conditions. Overall， this research contributes to the understanding of the thermo-physical behavior of asphalt mixtures containing floating beads and provides guidance for the design and construction of resilient asphalt pavement in permafrost regions.

Snow cover is an essential component of the cryosphere， and it has low thermal conductivity， high reflectivity， and large latent heat of melting， playing an important role in the energy balance process of the Earth. There are significant spatial differences in the response of snow cover to climate change. Snow cover is widely distributed in mid to high latitudes mountainous areas in the Northern Hemisphere， and spatiotemporal variations in mountainous snow cover have important implications for human life， the economy， and ecosystems. The Altai， the Alps and the Cascade Mountains are respectively one of the most important mountains in Asia， Europe and North America. The three mountains are located in the mid to high latitudes of the Northern Hemisphere， with rich snow resources. The three mountains are the only three typical mountain ranges in the mid to high latitudes of the Northern Hemisphere that can be horizontally compared. Based on daily snow cover products from MODIS， four snow cover parameters of the Altai， the Alps and the Cascade Mountains after cloud removal processing were obtained， including snow cover fraction （SCF）， snow cover duration （SCD）， snow onset day （SOD）， and snow end day （SED）. A comparative analysis was conducted on the spatiotemporal distribution and trend of snow cover parameters in three mountain ranges from 2002 to 2021， and the impact of climate factors on snow cover parameters was studied using satellite and reanalysis data. The results indicate that： （1） The spatial distribution differences in snow parameters among the three mountain ranges are evident and closely related to elevation. The Altai Mountains exhibit the maximum snow cover fraction， the longest snow cover duration， the earliest snow onset day and the latest snow end day， which are 38.00%， 141 days， 66 days， and 207 days respectively. The Alps has intermediate values for snow cover parameters. The snow cover fraction， snow cover duration， snow onset day and snow end day in the Alps are 21.68%， 79 days， 97 days， and 194 days respectively. The Cascade Mountains show the minimum snow cover fraction， the shortest snow cover duration， the latest snow onset day and the earliest snow end day， which are 15.18%， 56 days， 103 days， and 183 days respectively. （2） In terms of trends， snow cover fraction increases in the Altai Mountains， with earlier snow onset and later snowmelt leading to an increase in snow cover duration. Conversely， the Alps experience a reduction in snow cover fraction due to earlier snowmelt， resulting in a decrease in snow cover duration. The Cascade Mountains exhibit an increase in snow cover fraction， with later snowmelt contributing to an increase in snow cover duration. （3） The land surface temperature and precipitation in the three mountain ranges are increasing. The land surface temperature and precipitation have a greater impact on snow cover fraction and snow cover duration of the three mountains. The correlation coefficient between land surface temperature and snow onset day， and the correlation coefficient between precipitation and snow onset day and snow end day are relatively smaller. Temperature has a greater impact on snow cover parameters in the three mountain ranges compared to precipitation.

The existing research on the macroscopic strength of frozen soil under the influence of macroscopic control factors mainly relies on experimental methods， and has achieved good results based on actual conditions. Generally speaking， both indoor and outdoor tests have shortcomings such as long cycle time and high cost. With the emergence of new technological means， exploring simpler methods and building predictive models has been a long-term endeavor of scientific researchers. At the same time， the influence of macroscopic control factors on the macroscopic strength of frozen soil is exerted through the medium of the internal characteristics of the soil. Since ultrasonic waves are a good carrier of relevant information such as the physical and mechanical properties of rock and soil media， ultrasonic testing can reflect the internal characteristics of the soil due to its non-destructive， fast and simple characteristics. Therefore， this paper designs a strength prediction model containing different types of parameters based on different ideas. Idea 1-macro-controlling factors to macro-strength characteristics， idea 2-macro-controlling factors to internal soil characteristics reflected by ultrasonic wave velocity and then to macro-strength properties. Through experiments， the ultrasonic wave velocity and uniaxial compressive strength of soils with different salt contents undergoing different freeze-thaw cycles were obtained as basic data. The experimental control variables are used as idea 1 parameters， the ultrasonic characteristic parameter group constructed with compressional and shear wave velocities is used as idea 2 parameters， and the combined two ideas parameters are used as model input. A BP neural network prediction model for uniaxial compressive strength was established， and the prediction model was evaluated using the default factor test method. Tests show that as the number of freezing and thawing times and the salt content increase， the uniaxial compressive strength decreases overall. The wave velocity fluctuates significantly in the early stages of freezing and thawing， slows down in the middle stage， and returns to near the initial value in the later stage. Under the action of controlling factors， the uniaxial compressive strength decreases step by step as the wave velocity increases. The idea 2 parameters after gray correlation and rough set optimization are used to establish a BP neural network model for the optimal subsequence responding to the internal characteristics of the soil. The average absolute error of the model is less than 0.05 kPa， the coefficient of determination is greater than 0.96， and the average sensitivity index of each parameter is 1.4251. Sensitivity analysis successfully verified the assumed status of controlling factors and optimal subsequences in the model building process. A single controllable parameter has a greater impact on uniaxial compressive strength than a single ultrasonic characteristic parameter. The 29 parameters can be divided into four levels according to their contribution weight to the model. In the subsequent dimensionality reduction and feature selection of the number of parameters， the fourth level parameters should be discarded first， and the third level parameters should be optimized through parameter construction innovation and data sample expansion. This can reduce the number of overall parameters and increase the contribution weight， thereby better optimizing the model. The BP neural network model of uniaxial compressive strength established based on the different ideas of ultrasonic testing has strong predictive ability and good interpretability of model parameters. The ultrasonic characteristic parameter group under the influence of control factors plays an important role in the construction of the strength model. It also verifies the reliability and effectiveness of the BP neural network model in predicting the uniaxial compressive strength of saline soil. The model has high accuracy and strong practicability， and can provide a reference for strength prediction and parameter selection of frozen soil models.