This paper systematically reviews the significant contributions of Professor Zhang Tingjun in his 40 years’ academic research career， with the principal framework of climate and frozen soil changes in cold regions and their impacts. The selected publications mainly represent innovative achievements in the following four themes. （i） He investigated permafrost and ground ice distribution in the Northern Hemisphere， including Alaskan Arctic and high mountains in west China. He led the papers （in 1999 and 2000） that estimated permafrost area and ground ice volume based on the map compiled by the International Permafrost Association， which are widely recognized and cited. （ii） He explored the physical basis of the changing frozen ground， snow cover， and other related climate elements in cold regions to reveal potential interactions of key elements of the land surface processes. He synthetically summarized the impacts and physics of seasonal snow cover on frozen ground， which provides an essential start for the research on this topic. （iii） He made efforts to develop， improve， and apply numerical models that capture critical physical processes in frozen soils. His pioneering works on permafrost numerical modeling were beneficial， effective， and operational experiences for large-scale simulations of permafrost dynamics， beyond empirical and semi-empirical methods prior. （iv） He also established cryospheric remote sensing methods for detecting freeze-thaw cycles and seasonal snow cover. Many of these achievements have greatly improved the understanding of the physical basis of the changing cold regions and thus became classic literatures in the research of frozen ground and climate change. His 40-year life of pursuing truth is also worth learning and memorizing.
The Intergovernmental Panel on Climate Change （IPCC） Working Group Ⅰ Sixth Assessment Report （AR6）——Climate Change 2021： The Physical Science Basis， systematically summarized the observed facts of global permafrost changes， assessment and future prediction in climate models， and impacts of permafrost changes. The report states that， increases in permafrost temperatures in the upper 30 m have been observed throughout the permafrost regions over the past three to four decades （high confidence）. Permafrost temperature has increased at （0.29±0.12）?°C near the depth of zero annual amplitude of ground temperature between 2007 and 2016 （medium confidence）. Stronger warming has been observed in the continuous permafrost zone ［（0.39±0.15）?°C］ compared to the discontinuous zone ［（0.20±0.10）?°C］. The active layer thickness has increased throughout the pan-Arctic （medium confidence）. Further global warming will lead to near-surface permafrost volume loss （high confidence）. The volume of permafrost in the top 3 m will decrease by about 25% per 1?°C of global surface air temperature change （medium confidence）. However， such decrease may be underestimated due to an incomplete representation of relevant physical processes in earth system models （ESMs）. As the climate warms， permafrost degradation has significant impacts on permafrost carbon feedback to climate， ecosystems and human infrastructure， which needs to be considered in the climate model and risk assessment.
The first working group report of the Sixth Assessment Report （AR6） of the Intergovernmental Panel on Climate Change （IPCC） summarized soil carbon storage， carbon source and sink effects， and greenhouse gas emissions under future climate scenarios in permafrost regions. The report identifies that 1 460~1 600 Pg of organic carbon is stored in surface soils and deep sediments in permafrost regions of the Northern Hemisphere （medium confidence）. As the climate continues to warm， permafrost degrades significantly， allowing soil organic matter to decompose rapidly and release into the atmosphere as carbon dioxide （CO2） or methane （CH4）， accelerating climate warming. Under future global warming scenarios， near-surface permafrost will decrease significantly， which will in turn release CO2 and CH4 into the atmosphere， resulting in a positive feedback effect of carbon-climate. The report also points out that by 2100， the CO2 and CH4 releases in permafrost regions are expected to be 18 （3.1~41） PgC and 2.8 （0.7~7.3） PgC for every 1 ℃ increase in temperature （low confidence）. However， due to the wide range of estimated data and limited agreement among models， as well as the incomplete understanding of driving factors and carbon models in the permafrost environment， there is low confidence in the timing and magnitude of permafrost climate feedback.
Cryospheric rivers play a critical role in linking the cryosphere region， middle & downstream regions of the river， and even the oceanic carbon pool. Studying of source， migration， and transformation of the riverine black carbon （BC） in the cryosphere region will improve understanding of the marine and terrestrial carbon cycle. Climate warming has caused rapid cryospheric shrinkage recently， resulting in stored BC released due to glacier melting and permafrost thawing， which profoundly affects BC concentrations in the river source regions. This paper reviews the research progress of riverine BC in typical cryosphere regions， including the Arctic， Tibetan Plateau， Alps， Rocky Mountains and Andes Mountains. The results show that the fluxes of riverine BC in the global cryosphere region are about 2.29 Tg·a-1， accounting for 5.33% of the global riverine BC flux. In addition to atmospheric dry & wet deposition and runoff erosion， glacier melting and permafrost thawing significantly influence the concentration and flux of riverine BC. The annual fluxes of BC released by glacier meltwater in the Tibetan Plateau and Alaska are about 10.00 Gg （7.74~12.30 Gg） and 0.60 Gg （0.47~0.73 Gg）， respectively. However， the effect of permafrost thawing on BC in cryospheric rivers remains unclear. The lack of research on BC in cryospheric rivers will seriously limit the systematic understanding of regional and even global carbon cycles. In the future， it is necessary to continue to strengthen the systematic monitoring and research on BC in cryospheric rivers， which provide scientific data for quantifying the change and impact of riverine BC in the cryosphere region under climate warming.
As an important climate variable with “memory”， soil moisture can affect the local and global atmosphere circulation by changing the energy and water exchange on the surface， so that it has attracted the attention of the Global Climate Observing System （GCOS）. In recent years， the construction of the Tibetan Plateau soil moisture observation network （station） developed rapidly， providing reliable data support for the study of local land-atmosphere interaction. This paper summarizes the research progress at home and abroad in recent years about the applicability of different soil moisture data on the Tibetan Plateau， the temporal and spatial characteristics of plateau soil moisture and its impact on climate. The study found that due to the high variability of soil moisture in time and space， reanalysis data， land surface data assimilation data and satellite remote sensing data are often used to replace observation data to study the interaction between soil moisture and climate on the Tibetan Plateau. However， due to the difference in model， algorithm and experiment schemes， the applicability of alternative data of soil moisture on the Tibetan Plateau is different， which makes the scholars draw different conclusions on how the soil moisture on the plateau affects the climate on China and the world， and relevant issues need to be further discussed. The key problems to be solved in the later study of soil moisture on the Tibetan Plateau are also proposed.
Mercury （Hg） pollution has been an environmental issue of global concern. In the context of global warming， permafrost degradation can significantly alter soil environment and hydrothermal processes， and thus has high potential to remobilize and release significant amount of Hg to the environment， posing a risk to the ecosystems and biological health. This study reviewed the levels of soil Hg concentrations and pools， the spatial distribution and influencing factors of soil Hg in the permafrost region， and the impacts of different permafrost degradation patterns （i.e.， vertical active layer thickening/gradual thawing and thermokarst/abrupt permafrost thaw） on soil Hg dynamics in the permafrost region and its associated environmental effects. Soils （0~3 m） in the Northern Hemisphere permafrost region stored a significant amount of Hg （597 Gg， interquartile range （IQR）： 384~750 Gg）， and atmospheric dry elemental Hg ［Hg（0）］ deposition driven by vegetation uptake provided an important source of soil Hg. The concentration level and spatial pattern of soil Hg were mainly influenced by atmospheric deposition， soil organic carbon content and post-depositional processes （e.g.， eluviation）. The extensive permafrost degradation could release large amounts of Hg previously stored in the soils to the atmosphere and aquatic ecosystems. Meanwhile， it could also increase the microbial Hg methylation and the formation of Methylmercury （MeHg）， which has induced great impacts on global Hg biogeochemical cycles and posed a risk to the environment. And such impacts were expected to be intensified in a foreseeable future. Multi-technical means （e.g.， Hg isotope method） should be combined in the future research to track the migration and transformation of Hg during the hydrological transport with permafrost melt water runoff； Strengthen the research on the influence of thermokarst （abrupt permafrost thaw） on soil Hg release in the permafrost regions； Combine the field in-situ simulation experiments and model simulations to evaluate the influence of permafrost degradation on the migration and transformation of soil Hg and its environmental effects.
As an important indicator of climate change， studying the freeze-thaw status of the seasonally frozen ground （SFG） can provide a basis for assessing regional climate change. Using the data of 20 stations located in the Three Rivers Source Region （TRSR） of Qinghai Province， this study analyzed the spatial distributions and temporal variations of SFG freeze-thaw status in the TRSR from 1961 to 2019 by calculating four indicators： maximum freeze depth （MFD）， the first date of soil freeze （FFD）， the last date of soil thaw （TLD） and freeze-thaw duration （FTD）. And by calculating the air freezing index （FI） and thawing index （TI） and their change trend， combined with geographical factors （altitude， longitude and latitude） and climatic factors （air temperature， precipitation and snow depth）， evaluated the influencing factors of MFD and the freeze-thaw states of SFG in the TRSR. Results indicate that MFD in the TRSR was 64.7~214.1 cm， FFD was from early September to the end of October， TLD was from late March to the end of June， and FDR was about 144.7~288.4 days. From 1961 to 2019， MFD was decreased ［2.5 cm·（10a）-1］， FFD was delayed ［2.9 d·（10a）-1］， TLD was advanced ［2.6 d·（10a）-1］， and FTD was shortened ［5.5 d·（10a）-1］. Temperature is a key factor of SFG freeze-thaw status in the TRSR， and the influence of cold season warming is most important. It has an impact on the FTD and the MFD， while geographical factors and precipitation have a small impact， and the influence of snow depth in the area is basically negligible.
The recent increasing atmospheric greenhouse gases （GHGs） concentrations play a significant role on rapid climate warming. Rivers are important links in the biogeochemical cycle among the land， atmosphere and ocean. The riverine GHGs emissions can significantly influence regional GHGs fluxes， which is considered an important source of global GHGs and paly a non-negligible role on the achievement of carbon neutrality. Based on the published literatures on the global river CO2 emissions during 2010—2020， this study systematically analyzed river CO2 concentrations and fluxes and their influencing factors at the global and regional scales. The results indicated that global headwaters， mainstreams and estuaries were all sources of atmospheric CO2， and the total global river CO2 fluxes were （2.16±0.38） Pg C·a-1， with CO2 fluxes of headwaters > mainstreams > estuaries. Headwaters， mainstreams and estuaries fluxes were （0.69±0.12） mol·m-2·d-1， （0.42±0.07） mol·m-2·d-1 and （0.17±0.03） mol·m-2·d-1， total fluxes were （0.84±0.15） Pg C·a-1， （0.56±0.10） Pg C·a-1 and （0.76±0.13） Pg C·a-1 respectively. The emissions of CO2 from the mainstreams and headwaters of natural rivers were greater than that from anthropogenically influenced rivers. The CO2 fluxes in monsoon regions were higher than that of non-monsoon regions， and also showed a pattern among different climate zones （tropics>temperate zone>frigid zone）. The emissions from the estuary of anthropogenically influenced rivers were greater than that from natural rivers. The release of river CO2 decreased at river order 1 to 6 with increasing river order （based on Strahler river classification）， while it increased at river order 6 to 8. Although river CO2 emission significantly affects the regional carbon cycle， the comparison of CO2 emissions from river headwaters and mainstreams and the influence of anthropogenic impacts on river emissions magnitude are still unclear. Quantitative and systematic studies are urgently needed to further reveal the mechanisms of river CO2 emissions and assess their impacts on climate and environment.
Rock glaciers are lobate or tongue-shaped bodies of frozen debris， which are important substitute index of permafrost boundary and widely distributed in high altitude and high latitude periglacial areas around the world. Active rock glaciers are typically moving downslope or downvalley slowly， but the climate change may cause significant increase of rock glaciers， resulting in serious geological disasters in cold regions. In addition， rock glaciers might consist of massive ice which could be important freshwater resource in arid and semi-arid areas. Based on the literature review， rock glaciers are divided into talus-derived， glacier-derived， ice cored， Kunlun Mountain type， and the formation mechanisms of different types of rock glaciers are discussed respectively. The movement of rock glacier may develop from permafrost creep， frost heave and thaw settlement， debris supply， and new ground ice formation， among others. The variation of rock glacier movement characteristics on the spatio-temporal scale is mainly affected by climatic conditions， topographic environment， internal structure， external forces， among others. The response of rock glaciers to global warming will lead to two opposite results： the acceleration or instability caused by the increase of ground temperature and the deactivation of rock glaciers caused by permafrost melting. In the context of continuous retreat of glaciers and the shortening of the duration of seasonal snow， the debris cover as a thermal insulation layer can protect the underlying ice bodies from melting， delay and reduce the response of rock glacier to climate change， thus rock glacier will become an increasingly important water resource in arid and semi-arid areas. At present， more than 13 511 rock glaciers with more than 945.51 km2 of total area have been inventoried in China， but there is still a significant lack of systematic rock glacier inventory and long-term on-site monitoring and model research.
The Tibetan Plateau （TP） is sensitive to climate change， where snow plays a vital role in the regional hydrological cycle and climate system. This paper uses the daily cloud-free snow cover remote sensing data from 1980 to 2020 to analyze the distribution characteristics and variation trends of the snow cover area （SCA） and snow cover days （SCD） for a hydrological year from August 1 to July 31 in the next year over the TP. The results show that： （1） The snow in the TP shows an apparent spatial differentiation and a vertical zonal distribution characteristic. It is widely distributed in the high-altitude mountainous areas in the Amu Darya， the Indus， the Tarim， the Ganges， the Salween， and the Brahmaputra basins. （2） During the hydrological year， the snow cover extent shows a unimodal variation in the plateau with the minimum in early August and the maximum in mid-late January， accounting for 5.2% and 38.6%， respectively. （3） In the past 40 years， the average SCA shows a significantly decreased trend with 3.9×104 km2·（10a）-1 （P<0.05）. （4） The SCD in the study area mainly shows a significant decrease with 0.47 d·a-1. Up to 71.4% of the plateau shows a decreasing trend， and 55.3% of the area is significantly decreased （P<0.05）. And in 17.1% of the region， the SCD shows a significant increasing trend， and it is mainly distributed in the high-altitude mountainous areas above 5 200 m. Moreover， in the areas from 5 200 m to 5 900 m， the increase in SCD increased with altitude.
Gound thermal regime plays a crucial role in the soil physical， chemical， and microbiological processes. The ground thermal regime is typically dominated by air temperature as well as local factors such as vegetation， snow cover， and soil properties. Snow cover controls the energy exchange between the atmosphere and the land and plays a decisive role in the ground thermal regime during the cold season. This study investigated the influences of seasonal snow cover on ground surface temperature in Xinjiang based on 77 stations from the China Meteorological Administration. Our results showed that the mean ground surface temperature was about -1.3 ℃ at stable seasonal snow-covered stations in the cold season during 2005—2020， while the mean snow depth was 5.9 cm， and the air temperature was -4.6 ℃. At unstable seasonal snow-covered stations， the mean air temperature was 1.4 ℃ and the ground surface temperature was 2.4 ℃. The numerical simulation results indicated that an increase in snow depth of 1 cm corresponds to an increase of 0.26 ℃ in surface offset. A change in air temperature of 1 ℃ corresponds to a difference of ground surface temperature by 0.57 ℃ in the shallow snow zone （<5 cm） and by 0.20 ℃ in the thicker snow zone （>30 cm）. Further simulation was carried out at a typical snow station during 2008-07-01—2009-06-30. The results indicated that the mean ground surface temperature below the snow increased by 2.2 ℃ as the snow density increased from 200 kg?m-3 to 400 kg?m-3. The mean ground surface temperature during the simulation period was -2.7 ℃， -5.5 ℃， and -3.6 ℃ for the three scenarios of the normal time series， while maintaining the same snow depth， early snowfall， and delayed snowfall. The results highlight the significant influences of snow duriation on the ground surface temperature.
The surface melting of the Greenland ice sheet affects the global sea level rise through mass balance， and is also a sensitive indicator of climate change. To explore the changes and interannual differences of the Greenland ice sheet surface melting in recent decades， enhanced-resolution passive microwave daily brightness temperature data were used. Based on the air temperature data of automatic weather stations， five ice sheet surface melting detection methods， Advanced Diurnal Amplitude Variations （ADAV）， M+30 K， ALA， MEMLS1 and MEMLS2 were evaluated. The overall accuracy and Kappa coefficient confirm the feasibility and reliability of the ADAV method in ice sheet surface melting detection， and based on its detection， the temporal and spatial changes of the Greenland ice sheet surface melting from 1996 to 2021 are analyzed. It is found that all regions of the Greenland ice sheet have experienced surface melting in 26 years， and the most severe melting region is located at the margins of the ice sheet. The southern margin has a larger melt extent and more melting days than the northern margin. Extreme melting events caused the melt extent of the ice sheet to fluctuate greatly， while the melt index shows an increasing trend， with the growth rate of 5.24×105 d·km2·a-1. Surface melting tends to expand to inland high-altitude areas. The average elevations of the areas with melting days of 11~30 d， 31~50 d and 51~70 d have increased significantly in the past 26 years， with the growth rates of 13.06 m·a-1， 9.30 m·a-1 and 11.20 m·a-1 respectively. Correlation analysis shows that the surface melting of the Greenland ice sheet is driven by large-scale circulation factors and directly influenced by air temperature. The effect of Greenland blocking is more significant than NAO （North Atlantic Oscillation）， and the abnormal melt indexes in 2012 and 2019 indicated abnormal Greenland blocking. The air temperature rise of 2 m above the ground in summer promotes surface melting in the southern dome.
There are many surge-type glaciers in Alaska， and Turner Glacier is of great interest because of its extremely short recurrence interval. Previous studies on this glacier have mostly used optical remote sensing images， which failed to obtain detailed information on the movement velocity and surface elevation changes， and its surging process and control mechanism still need to be studied in depth. In this paper， we use Sentinel-1A， TerraSAR-X/TanDEM-X， ICESat-2 and Landsat remote sensing data to obtain the surface flow velocity， surface elevation and changes of terminal position of Turner Glacier during its recent surge from 2019 to 2021. The results show that Turner Glacier underwent a mini-surge from December 2018 to July 2019， which triggered a rapid movement of the glacier in February 2020 with a peak flow velocity of （18.85±0.05） m·d-1； in August 2021， the glacier flow velocity dropped sharply and then calmed down. During this surge， the material in the accumulation zone of the glacier migrates downward with a maximum thinning of （105.18±4.18） m； the downstream receiving zone rises with a maximum thickening of （60.25±4.18） m， and the end advances （222±30） m. The higher peak flow velocity， shorter active period， and seasonal flow velocity variation demonstrate that Turner Glacier may be controlled by subglacial hydrological mechanisms. Combining the available data and literature， the interval between the last surge is found to be about 6 years. The notch-like bottom topography of the north flow line about 27 km （
The temperature in Alaska increased more rapidly than the average global warming， which results in drastic changes in cryosphere. Meanwhile， these changes in cryosphere would further provide feedback to the climate system， enhancing and amplifying the magnitude of global warming. As an important component of the cryosphere， river ice is a sensitive indicator of climate change. Changes in river ice affect hydrology， permafrost， ecosystem， etc. In addition， river ice thickness （RIT） directly influences traffic safety because the ice road is an important transportation in Alaska. Therefore， the study of river ice thickness changes in Alaska is of great significance. The scientific objectives of this study are to investigate variations in the maximum river ice thicknesses （MRIT） using ground-based measurements of 48 river gauge sites spanning 12 river basins across Alaska， to analysis the relationship between MRIT and temperature based on meteorological stations data， and to reconstruct and predict the MRIT from 1850 through 2100 using Stefan equation and daily air temperature outputs of the coupled model intercomparison project phase 6 （CMIP6） historical experiment and four shared socioeconomic pathways （SSP） future scenarios （SSP126， SSP245， SSP370 and SSP585）. The results show that： The average MRITs in 1850—1900， 1900—1950， 1950—2000， 2015—2015 and 2050—2100 showed a trend of gradually thickening from south to north. The average MRITs in 1850—1900， 1900—1950 and 1950—2000 thickened from east to west， but MRITs of 2015—2015 and 2050—2100 did not change significantly along the east-west direction. The overall MRIT across Alaska showed a significant decline from 1850 to 2100. The decline rate from 1850 to 2014 was （-0.72±0.25） cm·（10a）-1. The decline rate from 2015 to 2100 was （-1.39±0.76） cm·（10a）-1 under the SSP126 scenario， （-3.10±0.73） cm·（10a）-1 under SSP245， （-6.09±0.79） cm·（10a）-1 under SSP370 and （-7.45±0.63） cm·（10a）-1 under SSP585. The decline rates in 2015—2100 accelerate under the four future scenarios. The MRITs for all gauging sites showed a significant downward trend from 1850 to 2100. They decreased faster from east to west and from south to north， except for that in 1850—2014 and 2015—2100 under the SSP245 scenario with non-significant different along south-north direction.
Previous studies show that there are four surge-type glaciers on the Anyemaqen Mountain， and the surges of Xiaomagou Glacier on the west slope have caused four glacier collapse disasters. This work studied the distribution， dates of the occurrence， durations and repeating cycles of glacier surges on Anyemaqen Mountain by analyzing a vast amount of satellite images including Landsat images since 1977 and Sentinel-2 images since 2015， and taking the changes on the horizontal position of glacier termini and typical surface features as the reference. The results show that there are totally eleven surge-type glaciers （including the branches of several glaciers） on Anyemaqen Mountain， accounting for approximately 10% and 66% of the total number and area of glaciers in this region in 2022. The results also revealed that totally seventeen glacier surge events have occurred in this region since 1986， of which the surges of Qushen’an No. 22 and Qiemuqu No. 23 Glaciers are still in progress. The frequent surges of Xiaomagou Glacier since 2000 have caused five glacier collapse disasters with different extents， in which the 2021’s surge and collapse were not reported before. Furthermore， there are seven glaciers are suspected to have been also surged， which show slight advances in some periods after 1986， and accounting for about 6% and 13% of the total glaciers number and area of in 2022. The results also indicate that the glacier surges on Anyemaqen Mountain are primarily in slow manner， and mostly have >50 years surge cycles. The surge behaviors of Weigeledangxiong and Halong Glaciers have been changed under the climate change since 1990， represented by the decreased surge extents and speeds. Meanwhile， the large area of ice avalanches on the steep back wall of several glaciers caused by climate warming may increase the instability of those glaciers. The warming climate may also cause the declined stability of Maqen Kangri Icecap， which may in turn cause some changes in the surges and collapses of its outlet glaciers like Xiaomagou Glacier and Weigeledangxiong Glacier.
Based on the Global Open Glacier Model （OGGM） in conjunction with the Sixth Climate Model Comparison Program （CMIP6）， three climate scenarios （SSP1-2.6， SSP2-4.5， SSP5-8.5） simulated by five climate models （BCC-CSM2-MR， CESM2， CESM2-WACCM， FGOALS-f3-L， NorESM2-MM）， the changes in the area and reserves of Sawir Mountains glaciers during 2020—2100 were systematically analyzed. The results show that under the three climate scenarios， the area and volume of glaciers in Sawir Mountains show a shrinking trend， and the glacier area and volume loss under the SSP5-8.5 high forcing scenario is the largest， and the corresponding changes in area and volume are -0.154 km2·a-1 and -5.11×106 m3·a-1， followed by the SSP2-4.5 medium forcing scenario， the corresponding area and volume changes are -0.150 km2·a-1 and -5.05×106 m3·a-1， the SSP1-2.6 low forcing scenario has the smallest area and volume loss， and the area and volume change is -0.139 km2·a-1 and -4.93×106 m3·a-1. The changes in glacier area and volume in the Sawir Mountains in China are greater than those in Kazakhstan. The changes in glacier area basically conform to the process from relatively stable to rapid changes， and finally slow down， but the changes in volume are relatively flat. Between 2020 and 2050， glacier thinning will dominate in the Sawir Mountains； between 2050 and 2100， glacier retreat will dominate in the area of the region. By 2060， 47.8% of the glaciers on Sawir Mountains will have retreated， and this proportion will rise to 78.2% by 2080.
The freezing/thawing index is an important sensitive indicator of climate change， and also widely used in the permafrost change. The spatial-temporal variations of freezing/thawing index on a global scale can provide a basis for global permafrost environment assessment， engineering construction and climate change. Based on the daily observed temperature data from 1973 to 2021 covering the global land and more than 14 000 stations， we calculate the air freezing/thawing index and analyze its spatial and temporal variation characteristics， and discuss the relationship with geographical factors. Results show that the global average freezing index is 610.8 ℃·d in the past 49 years， the maximum value is 19 653.3 ℃·d， 667.9 ℃·d in the Northern Hemisphere larger than 152.4 ℃·d in the Southern Hemisphere. The global average thawing index is 4 709.6 ℃·d， the maximum value is 11 217.0 ℃·d， and 4 444.5 ℃·d in the Northern Hemisphere smaller than 6 927.3 ℃·d in the Southern Hemisphere. Spatially， the freezing index in low latitude regions such as near the equator is basically 0 ℃·d， and a thawing index of 0 ℃·d only appear in Antarctica and Greenland. The freezing/thawing index is affected by both latitude and altitude， and has obvious climatic distribution characteristics. The freezing index of global stations showed a downward trend at an average rate of 6.4 ℃·d·a-1， while the thawing index showed an upward trend at an average rate of 14.0 ℃·d·a-1. However， the changes of the freezing/thawing index tended to be flat at the beginning of the 21st century. From 1973 to 2021， the freezing index decreased by an average of 36% globally （net change of -160.2 ℃·d）， an average decrease of 68% in the high latitudes under the snow climate zone. The thawing index increased by an average of 11% globally （net change of 437.5 ℃·d）， while an average increase of 16% in the high latitudes under the snow climate zone. The trend of the freezing/thawing index reflects that the global climate is warming with different degrees， especially in high latitudes and high altitudes， and the warming trend in the cold season is greater than that in the warm season.
One of the common features between the Qinghai-Tibet Plateau and the circum-Arctic region is the presence of a large areal extent of frozen ground. Changes in frozen ground， including those of the past reconstructions and future predictions， are of great significance for the study of the triple pole linkage， one of the hot issues in the studies on global climate change. Based on the published scientific literature and assessment reports， we systematically summarize the study results of changes in frozen ground in the five aspects， including permafrost area， mean annual ground temperature （MAGT）， active layer thickness （ALT）， near-surface freezing-thawing status， and frost depth in these two regions. Results show that， based on different methods， permafrost area decreases， MAGT increases， ALT enlarges， duration of ground freezing decreases， and frost depth declines but with some differences with different methods. In future studies， it deems necessary to further strengthen the frozen ground monitoring， better understand physical processes and the interaction mechanisms between frozen ground and climate and environment， and take into account of the influences of lateral heat flow in predictive models. These progresses will provide data and methods for supporting for higher-precision model simulations and predictions in the future. The summary in this study can provide some ideas for the future study on frozen ground and provide references for engineering design and construction in cold regions. It also gives warning for people to recognize that changes in frozen ground environment may lead related hazards， and emphasize the importance of human protection on the frozen ground environment.
As one of the most widely distributed thermokarst landscape caused by permafrost degradation， thermokarst lake is an important source of methane （CH4） in the atmosphere. The evolution of thermokarst lakes and its influence on CH4 cycle are key issues in the study of climate change. In this paper， we reviewed the evolution， distribution and change of thermokarst lakes in permafrost regions of the Northern Hemisphere. CH4 production， oxidation， and emission from thermokarst lakes and their influencing factors were revealed. The results showed that the areas of thermokarst lakes around the Arctic was approximately 1.6×106 km2. Although thermokarst lakes may expand or form in some regions， the overall coverage of thermokarst lakes was decreasing. The area of thermokarst lakes on the Qinghai-Tibet Plateau was 2.83×103 km2， which showed that the number and area of thermokarst lakes increased significantly in the central region， but these showed a decreasing trend in the source region of the Yellow River. Influenced by the stability of organic matter and microbial community， the rich-organic matter layer in the near-surface and thawing permafrost layer under the thermokarst lake have the great potential for CH4 production， but the CH4 oxidation greatly limits the CH4 emissions. It has been estimated that the CH4 emissions from thermokarst lakes in the Arctic region ranged from 1.9 to 6.3 Tg CH4·a-1 with a large uncertainty. In the future， we need strengthen remote sensing technology to improve the recognition accuracy of thermokarst lake， and focus on the effect of lake drainage on the CH4 cycle. These will deepen the insight in the support for the prediction of the lake development trend and the assessment of permafrost carbon feedback. In addition， we should pay more attention for the long-term field monitoring in typical and sparse areas， and explore the mechanisms of CH4 production and oxidation in thermokarst lakes. Finally， the drivers of methane production and emission， such as substrate availability （the amount of organic matter that microorganisms can directly decompose and use）， microbial characteristics， hydro-thermal change and vegetation， should be incorporated into biogeochemical models in the future to better predict and evaluate the contribution of carbon emissions from thermokarst lakes to global warming.
Climate warming has caused the permafrost degradation， accelerating the formation and expansion of thermokarst lakes or ponds， and further increasing carbon release from permafrost regions. The physio-chemical variables of sediments play a critical role on methane （CH4） production in the thermokarst lakes and ponds. Quantifying CH4 production from the sediment of thermokarst lakes and ponds will improve our understanding on the response of methane emissions to climate change over the Qinghai-Tibet Plateau. In this study， based on sediment samples collected from eight thermokarst lakes or ponds in the central and eastern Qinghai-Tibet Plateau， we intended to investigate the relationship between the physio-chemical variables of sediment and CH4 production under laboratory incubation at 5 ℃， 10 ℃， and 15 ℃， respectively. The results showed that during the 50 days incubation period， the maximum CH4 production （167.63 μg·g-1 sediment） was found in MD-3 samples in an alpine wet meadow at the incubation temperature of 10 ℃. The minimum value （0.01 μg·g-1 sediment） was recorded in the AD-2 sample that under a wet meadow at the incubation temperature of 15 ℃. The CH4 production rates were higher in the sediments with relatively deeper thermokarst lakes and ponds， higher ammonia nitrogen content and lower the pH values. Meanwhile， the methane production of thermokarst lakes and ponds sediments was much higher than those in the Anduo area due to the high ammonia nitrogen and low pH values in the thermokarst lakes or ponds in the Maduo area. Temperature sensitivity of methane production （Q10） of 5~10 ℃， 10~15 ℃ and 5~15 ℃ revealed that the increasing temperature has a promoting effect on 61.11% of total CH4 production and an inhibiting effect on 18.06% of total CH4 production. The results indicated that temperature was an important factor determining CH4 production in the sediment of thermokarst lakes or ponds. This study investigated the effects of temperature on CH4 production and their possible influencing factors from the sediments of thermokarst lakes or ponds， and thus provide a scientific basis for evaluating the potential and modeling of greenhouse gas emissions from thermokarst lakes or ponds.
The active layer is the most thermodynamically active near-surface soil layer in the permafrost regions. It is vital to the permafrost eco-environments as it serves as the critical zone for the supply of water and nutrients for the growth of alpine/northern plants， as well as the habitats for most frequent microbial activities and critical biogeochemical cycles. It also plays an indispensable role in the exchanges of water and energy between the atmosphere and the near-surface ground. Recently， the active layer thickness （ALT） under natural and undisturbed conditions had been prevalently increased under the dual influences of climate warming and increasing anthropogenic activities， which poses significant adverse influences on the cold environment and frozen ground engineering. In this paper， we reviewed the influencing factors of the ALT under natural and undisturbed conditions in the aspects of macro-scale geology and geography and micro-scale local factors， the measurements and simulations of ALT， as well as the response characteristics of ALT to climate change. Moreover， we also discussed the impact of ALT change on the alpine ecological environment. The past modeling and observations demonstrated that the spatial heterogeneity of ALT was primarily attributed to the redistribution of solar radiation and its complex interactions with the underlying conditions. Presuming no differentiation in climate and local factors， the thicker ALT is always found in the vicinity of the lower limits of elevational permafrost or of the southern/northern limits of latitudinal permafrost. In the past three decades， ALT has increased sensitively to climate warming， which is characteristic of increasing with the rise of air temperature. The increase of ALT shows an obvious regional differentiation， among which the ALT at most of the mid-latitude alpine and mountainous permafrost regions， such as in the Tibetan Plateau and the Alps， has shown significant increasing trends， while the deepening of ALT to a certain extent was offset by the melting of ground ice and ensued thaw settlement or ground surface subsidence at high-latitude ice-rich permafrost areas. Therefore， not all sites at high latitudes have experienced significant increasing trends as revealed by the observations. However， when analyzing the sensitivity of ALT by the ratio of its changing rate to its average value， we have found that the sites in the Alps （1.46） and the Nordic regions （1.27） were the most sensitive， followed by the sites in Alaska （0.93） and on the Tibetan Plateau （0.91）， while those in Canada （0.25） had relatively low sensitivity. We conclude that the future research directions of ALT should focus on the precise simulation and mapping of ALT， the adaptive mechanisms of ALT to climate changes， the impact of changing ALT on the biogeochemical cycles， hydrological processes， and water resources and structures in cold regions， among many others.
Lead is a toxic heavy metal element that is extremely harmful to humans. Tracing the source of lead is a prerequisite for controlling lead pollution. Since the lead isotope is difficult to undergo obvious fractionation in the natural process and retains the characteristics of the pollution source area， it has become a powerful “fingerprint” tool for tracking lead pollution sources. As an important environmental medium of the cryosphere， snow and ice have the characteristics of preserving atmospheric lead. The study of snow and ice can reconstruct regional and global atmospheric pollution changes and explore the activity history of past lead emissions. This paper systematically summarizes the lead concentration and lead isotope records， spatial distribution in snow and ice in the three polar regions （Antarctic， Arctic， Qinghai-Tibet Plateau and surrounding areas）， and related studies on the identification and analysis of lead sources in snow and ice using lead isotopes. The results show that the lead concentration in snow and ice in the three poles presents the spatial distribution characteristics of the Third Pole >Arctic > Antarctic， and the lead concentration in snow and ice in the northern part of the Third Pole is higher than that in the south and has a lower 206Pb/207Pb ratio. Ice core environmental records show that lead pollution occurred significantly during the Ancient Roman period and the Middle Ages， the Industrial Revolution period （mining， smelting， coal burning）， and in the second half of the 20th century （before and after the use of leaded gasoline）， with the peak concentration of lead in snow and ice in the tripolar region mainly appeared around the 1970s. Lead isotope tracers show that mineral dust is the main natural source of lead in snow and ice in the tripolar region， and the snow and ice in different locations in Antarctic are also affected by volcanic activities to varying degrees. The lead pollution in Antarctic snow and ice mainly comes from Australia and South American countries， the lead pollution in Arctic snow and ice mainly comes from North America and Eurasia， and the sources of lead in different glaciers and snow ice of the Third Pole are different.
The migration and accumulation of water during soil freezing， a crucial step of frost damage， has always been a frontier and important topic in the research of frozen soil physics. Since the golden age of water migration research for frozen soil in the 1970s and 1980s， there has been no major breakthrough in classical theory and scientific cognition. Many mechanical problems and key bottleneck problems involved in water migration in freezing soil still cannot be answered accurately. The damage and engineering problems related to frost heave haven’t been completely solved until now. Because of its variability， microcosm and mutation， the phase transition zone （frozen fringe） is still a “black box” in the study of water-heat transport in freezing soils. This paper reviews development history， main research progress and current situation of the research on the driving force and process of moisture migration in freezing soil， and then analyzes the physical principle and basic law of pore-water pressure and soil-water potential related to moisture migration in freezing soil. We also summarize the latest progress and main scientific problems in the theoretical characterization and experimental tests of pore-water pressure and soil-water potential， and analyze three popular theories of the moisture migration， namely capillary-， film-， and vapor-water migration theory. This paper summarizes the key bottleneck problems that restrict the breakthrough of scientific cognition for moisture migration in freezing soil and looks forward to the focus and direction in future research. The research suggests that followings should be focused on in future： 1） Strengthening the invention of instruments and the application of new technologies. 2） Focusing on the physical essence and accurate dynamic monitoring of the driving force of moisture migration. 3） Further understanding of the micro process and mechanism of the moisture migration. 4） More attention should be paid to the process of heat-mass transport and ice-water phase change from the perspective of non-equilibrium thermodynamics. 5） The development of multi-physics coupling simulation and the construction of autonomous open-source computing platform. The purpose of this paper is to systematically review current situation and future research direction， and then promote the development and improvement of the basic theory in the heat-mass transport of freezing soil， and better serve the solution of freezing-thawing disasters and environmental problems in cold regions.
The thermal parameters of ice are the keys to reasonably simulate the ice phenology， distribution and thickness， but have always been a vulnerable group in the field of ice research. It may seem technically simple to precisely measure the ice thermal parameters， however the actual impact factors are numerous so a rigorous research process is required. In 1980s， after intensive exploring the thermal conductivity of ice in the field and laboratory， there has been no significant progress in China. Past efforts were mainly focused on the determination of “cold” ice thermal parameters and developed models based on simple porous medium theory. However， the constituents and pore structures of natural ice are very complicated； especially during the phase change of melting period， the interior structure is in quick response to changes in ice temperature， so how to accurately obtain the thermal parameters during this period is also important and challenges the experimental testing methods. In this century， after the introduction of mathematic methods， the inversion method has been usually used to identify the thermal diffusivity of ice in differing water bodies based on the analysis of the time-series data of vertical temperature profiles measured in situ. This method can obtain thermal diffusivity with differing ice temperature， especially at the phase transition of “warm” ice.The present paper briefly reviewed the physical background of thermal diffusivity of ice， summarized the typical parameterizing equations for thermal diffusivity of freshwater and sea ice， and presented the physical and mathematical background， identifying scheme， and numerical algorithm design for inverse identification using series of vertical profiles of ice temperature. Afterwards， we tidied up the field and in-door results on temperature profile series， interior structure， and thermal parameters of freshwater and sea ice in various water bodies， collected typical cases of using inverse identification to determine the thermal diffusivities of ice since 1980s in China. Aggregate analysis showed that the thermal diffusivity of natural freshwater ice and sea ice was different from that of pure ice selected in current simulation models due to the impurities in the unfrozen water between ice crystals， but the trends were consistent with a small number of laboratory tests of different types of salt water ice. It also indicates of the importance of accurate description of thermal diffusivity of ice within the phase transition zone under the background of global warming. At last， future efforts and considerations were proposed with respect to the accuracy and uncertainty of inverse identification， inevitability induced by global warming， differences in differing ice crystal structure， importance of in-ice gas pores， complex impacts of brine in sea ice， design on parameterizing scheme for natural freshwater ice， and future inverse identification of sea ice thermal diffusivity.
To continuously observe the physical process of lateral melting of natural ice， ice lateral melting tests simulating the hydrostatic and pure thermodynamic test conditions without radiation and wind in the low-temperature environment laboratory were conducted using natural ice samples collected in winter. After the ice samples were pre-treated， the melting tests of natural ice samples were carried out under different water temperatures with different specimen sizes， and the surface sizes， thickness， and mass of ice specimens were measured. The laboratory temperature and water temperature were recorded during tests. According to the experimental results， the latent heat flux and the lateral melting rate of the ice samples are calculated， and the process of the lateral boundary shape change and the factors affecting the lateral melting rate are analyzed. The test results show that the side， bottom， and surface of the ice melted simultaneously during the melting process， the lower part of the side of the ice became concave， the ice thickness became thinner， and the ice-water interface changed in an arc shape. The lateral melt of ice samples increased with the increase of depth， and the amount of lateral melt in the ice bottom was the largest， which was the result of the combined effect of the ice-water lateral heat flux and the upward heat flux of the ice bottom. The lateral melting rate of ice samples showed a trend of decreasing first and then increase with time， while the ice thickness kept a relatively stable trend of decreasing continuously， and the variation range of mass gradually decreased with time. The larger the thickness and diameter of the ice， the smaller the latent heat flux and the lower the lateral melting rate of the ice sample. This is because the increase of the thickness and diameter of the ice requires the ice sample to absorb more heat transmitted by the water body， which leads to the decrease of the water temperature and thus slows down the ice melting. The influence of ice sample thickness and diameter on the lateral melting rate is comprehensively considered by using the ice-water lateral contact area. The larger the ice-water lateral contact area is， the lower the average lateral melting rate is. The smaller the ice density is， the larger the latent heat flux and the higher the lateral melting rate of ice samples. The reason is that the ice melting is accelerated by water penetrating the unenclosed pores or brine channels in the ice. By fitting the relationship between the latent heat flux of lateral melting and the area of lateral contact with water and porosity of ice samples， the 3-d surface reflects the characteristic that the smaller the area of lateral contact with water and porosity of ice samples， the larger the latent heat flux of lateral melting and the larger the lateral melting rate of ice samples. The change of ice shape characteristics will also change its melting response to the thermal action. The melting test results of four different ice shapes show that the lateral melting rate is in the order of right triangle， circle， trapezoid， and square from large to small. At the same time， the shape of the ice sample became closer and closer to the circle during the melting process. The results of the present study are beneficial supplements to the thermal melting characteristics of natural ice， and help to guide the development direction of the parameterization scheme of the lateral melting model of the in-situ sea ice. In the future， the results of laboratory tests will be calibrated based on collecting and accumulating more in-situ observation data， to promote the leap of field application of laboratory research results.
In permafrost and ice engineering， the study on mechanical properties of ice is of great importance， wherein changes in content of unfrozen water inside ice can lead to overall changes of properties of ice. At present， the study at microscopic molecular scale on control factors of unfrozen water content in ice presents non-adequate. As a relatively special solid material， ice can be classified into single-crystal and polycrystalline ones in terms of their morphological difference， the latter of which usually exists in a polycrystalline structural state， in nature and artificial laboratory environments. And the structural difference between polycrystals and single crystals is mainly that the former have grain boundaries， that is， some relatively disordered or less neat regions in the middle of two regularly arranged lattices. In contrast， the entire structure of a single crystal is a long-range ordered crystal structure. Namely， polycrystals can be viewed as being composed of many single crystal particles. The exploration on mechanical properties of polycrystalline ice basing single-crystal ice can help us to further understand its microstructural changes. In this study， after building single-crystal and polycrystalline ice models， setting up molecular force fields， and initializing the system for models， uniaxial tensile and compressive tests were implemented on single-crystal and polycrystalline ices by molecular dynamics simulations， under five temperatures， three strain rates， four grain sizes and three heating rates. The mechanical properties of ice crystals and internal microstructural changes inside them were then investigated， under different influencing factors， aiming to reveal the underlying influencing mechanism of unfrozen water proportion on mechanical properties of ice crystals. In terms of simulation results， for single-crystal ice， mechanical properties of them appear weak relation to unfrozen water proportion， but significant dependence on breakage degree of six-membered ring structures， as well as they are influenced by temperature， strain and strain rate. And during both tension and compression tests， single crystal ice shows significant brittle damage， i.e. the strength increasing along with temperature decreasing or strain rate increasing. In addition， the compressive strength of it tends to be higher than tensile one， but shows an obvious nonlinear mechanical response after the yield point. By contrast， due to the existence of grain boundaries， mechanical properties of polycrystalline ice are more sensitive to the unfrozen water ratio and the ice crystal strength increases along with decrease of unfrozen water ratio， which is mainly controlled by temperature， grain size and its interfacial state. The smaller the crystal size is and the larger the grain boundary area is， the easier the unfrozen water forms at the grain boundary. And the mechanical properties of polycrystalline ice change drastically under the influence of unfrozen water ratio. Moreover， due to strain-induced amorphization and collective sliding between grain boundaries， polycrystalline ice with nanograins is unstable and the sensitivity of it to temperature and strain rate is more pronounced in comparison of that of single-crystal ice， indicating structural changes between grain boundaries play a non-negligible role in mechanical properties of polycrystalline ice. In addition to elastic deformation， the combination of grain boundary slippage， grain rotation， amorphization and recrystallization dominate the plastic deformation of polycrystalline ice. Under influence of external forces and warming， the crystal structure in polycrystalline ice crystals gradually changes， from stable hexagonal ring to unstable quadruple ring， five element ring， seven element ring and disordered water molecule structures， etc. Along with the increase of proportion of unfrozen water， the ultimate strength of ice crystals decreases. In this study， relying on molecular dynamics simulation， the authors focus on generation and changes of unfrozen water at the molecular scale during process of ice crystal deformation and also the intrinsic mechanism of influencing effect of unfrozen water on mechanical properties of ice crystals， with an ultimate hope of explaining the acting mechanism behind some macroscopic phenomena， from the perspective of molecular structure changes.
By using Climatic Research Unit （CRU） data， the study has evaluated the simulation ability of 17 global climate models of Coupled Model Intercomparison Project Phase 6 （CMIP6） and their ensemble average annual precipitation in the Northern Hemisphere and permafrost regions in the historical period （1985—2014）. The spatial and temporal changes of future annual precipitation in the Northern Hemisphere and permafrost regions under different future scenarios （SSP1-2.6， SSP2-4.5， SSP3-7.0 and SSP5-8.5） have also been analyzed. The results show that CMIP6 models have reasonable ability for simulating the spatial distribution of annual precipitation， but generally overestimate about 11% and 42% respectively compared with the observation data in the Northern Hemisphere and permafrost regions. The trends of annual precipitation in the Northern Hemisphere and permafrost regions will accelerate with the increase of radiation forcing level in the future. Under SSP5-8.5 scenario， the annual precipitation’s increase rate is the fastest， about 13 mm·（10a）-1 for the Northern Hemisphere and 20 mm·（10a）-1 for the permafrost regions， which are 134 mm and 178 mm higher than that in the last year of the historical period （2014）. The average annual precipitation in the Northern Hemisphere is always higher than that in the permafrost regions， but the increase rate in the permafrost regions is higher than that in the Northern Hemisphere. Under SSP3-7.0 scenario and SSP5-8.5 scenario， the increase rates of annual precipitation in the permafrost regions are more than 1.5 times that in the Northern Hemisphere.
This study investigates the association of spring （April-May） Arctic sea ice melt area with simultaneous precipitation over East Asia from 1979 to 2020 by using observational datasets and the Community Atmospheric Model version 5.4. The Arctic sea ice melt area exhibits interannual variability during the study period， and simulation experiments forced by Arctic sea ice anomalies reproduce the major characteristics of observational associations. Observed evidence and numerical experiment results show that spring precipitation anomalies associated with Arctic sea ice melt display an increasing trend over East Asia. Correlation analysis reveals that the Arctic sea ice melt area could be explaining approximately 26% precipitation variation over East Asia. The 500 hPa geopotential height anomalies exhibit a wave train structure， and a dominant positive center is located over the Ural Mountains with two negative centers over East Asia and West Europe. This atmospheric circulation anomaly is different from the traditional Eurasian pattern and North Atlantic-Eurasian teleconnection pattern owing to their different spatial modes. High Arctic sea ice melt years are often associated with strengthens polar zonal westerly winds and weakens zonal winds in mid-latitude. It may contribute to the Arctic anticyclonic anomalies enhancement， reduce cloud cover and enhance the Arctic incoming shortwave radiation in the spring， leading to an increase in surface air temperature in the Arctic， which in turn favor sea ice melt， a positive ice-albedo feedback process emerges. Then， the enhanced anticyclonic circulation anomaly induces a wave train southeastward propagating into the mid-low latitudes， causing the decrease of geopotential over the Lake Baikal， resulting in a cyclonic circulation anomaly， affected by the anomalous northerly flow， it is conducive to southward and accumulation of cold air. At the same time， Japan and its surrounding seas geopotential increasing， leading to an anticyclonic circulation anomaly， affected by the anomalous southeasterly flow， warm and humid air from the northwest Pacific Ocean can be brought to East Asia. For the high melt years， such a wave train pattern is usually associated with an anomalous low-pressure center over the Mongolian Plateau， which accelerates the East Asian subtropical westerly jet. The intensi?ed subtropical westerly jet， concurrent with lower-level convergence and upper-level divergence， enhances the local convection and consequently favors rich spring precipitation over East Asia.
Snow cover is an important freshwater resource， which is one of the key physical parameters for understanding climate change， ecosystems and human economic and social development. The Third Pole and the Arctic are the main snow cover areas over the Northern Hemisphere. In this study， we first evaluated the accuracy of five snow water equivalent products （GlobSnow V2.1， GlobSnow V3.0， CanSISE， GLDAS-2.0， GLDAS-2.2） and analyzed the spatiotemporal characteristics of snow cover resources in these regions. The results showed that the long-term （1981—2010） mean maximum snow water accumulation was （46.07±7.44） km3 for the Third Pole and （1 255.73±81.35） km3 for the Arctic， respectively. At the Third Pole， the most abundant snow cover resource appeared in the Himalayas， Karakoram， and Nyainqêntanglha Mountains. The largest snow cover resources in the Arctic were mainly distributed in the eastern part of the Russian Far East， West Siberia， the Mackenzie Mountains and eastern Baffin Island in Canada. There were decreasing trends in maximum snow water accumulation in both regions， but with different inter-annual variations.
Accuracy areal precipitation is one of the key factors influencing the ecological barrier construction and water balance estimation in the Qilian Mountains， but the large difference was found among the current results due to rare measurements especially in the alpine regions. Therefore， total 23 T-200BM3 and 6 TRwS204 weighted rain and snow gauges were installed in the middle-high regions in the Qilian Mountains， together with two precipitation intercomparson fields with elevation of 2 980 and 4 651 m. Based on these measurements and precipitation data measured by the Chinese Standard Precipitation Gauge （CSPG） at 27 stations in or near the Qilian Mountains， the areal precipitation in 2019 and 2020 was estimated after observation error correction and different gauge types unification. The results show that： 1） the averaged observation error is close to 10% at all stations， especially in the alpine regions where the snowfall and precipitation error is about 30% and 20%， respectively； 2） the TRwS204 gauge could observe the “true” precipitation while the T-200BM3 gauge measurement is 12% higher even after disturbance elimination； 3） the annual precipitation increase with elevation in the Qilian Mountains but there may exist one or more indistinctive maximum precipitation elevation in some profiles； 4） the estimated areal precipitation is about 565.2 mm and 446.3 mm in 2019 and 2020， both which is 70% higher than the result that only uses the CSPG data at national meteorological stations， that is， the latter is about 59% of the former which is also close to the downscaling GPM data by using the alpine measurements. Although the result has high accuracy at present， it still has some uncertainties due to rare observation in the southwest high regions in the Qilian Mountains， as well as the low measurement accuracy of T-200BM3 gauges in these regions.
Frozen ground is an important component of the cryosphere and one of the most sensitive regions to climate change. The difference of water and heat conditions caused by the change of frozen ground environment is an important factor that triggers the energy exchange， water cycle and carbon cycle of the vegetation ecosystem. Water use efficiency （WUE） is the key to link the relationship between carbon cycle and water cycle in ecosystems， reflecting the adjustment and adaptation strategies of vegetation ecosystems to frozen ground degradation. Based on the gross primary productivity （GPP） and evapotranspiration （ET） products of MODIS， this study estimated and analyzed the spatial variation characteristics of vegetation GPP/ET/WUE in the permafrost and seasonally frozen ground area of the Qilian Mountains from 2000 to 2020， and combined with the self-calibrating Palmer drought severity index （scPDSI） to study the response of WUE in permafrost and seasonally frozen ground area to drought. The results showed that： （1） The average values of WUE， GPP and ET of vegetation in the Qilian Mountains from 2000 to 2020 were 0.56 gC·m-2·mm-1， 307.79 gC·m-2 and 443.02 mm， respectively. The spatial distribution characteristics of the three were high in the southeast and low in the northwest； the value of WUE higher than 0.8 gC·m-2·mm-1 is mainly distributed in seasonally frozen ground area， while lower than 0.4 gC·m-2·mm-1 is mainly distributed in permafrost. （2） In the past 20 years， the vegetation WUE in the Qilian Mountains has generally shown a slow decrease trend， but the monthly change trend is “increase-decrease-increase”. The increase trend of vegetation WUE is the most significant in November and the decrease trend is the most significant in June； WUE change in the study area is mainly determined by seasonally frozen ground area. （3） The spatial distribution characteristics of the correlation between WUE and scPDSI in the Qilian Mountains are affected by the distribution of seasonally frozen ground area and permafrost. The negative correlation between WUE and scPDSI is the most significant in June， the positive correlation between WUE and scPDSI in seasonally frozen ground area is the most significant in March， and the positive correlation between WUE and scPDSI in permafrost is the most significant in November. The difference of temporal and spatial distribution between permafrost and seasonally frozen ground is the direct reason that affects the spatial heterogeneity of WUE. The heterogeneity of temporal and spatial change of WUE is the direct reflection of the water consumption of vegetation change and the adaptation of productivity to ecosystem changes.
The cryosphere is one of the five spheres in the climate systems. Effects of changing cryosphere on the various ecosystems both within the cryosphere and the ones closely associated with cryosphere has become a hot research topic of global change， and the concept of “cryospheric ecosystems” have been widely used these years. Based on the existing results on the interaction between the cryosphere and the biosphere， this review systematically described the composition of the cryosphere ecosystem and the relationship between the elements of the cryosphere and the biomes. The composition of the cryospheric ecosystem frame was also outlined. From the perspective of ecosystem， the main research topics of the cryospheric ecosystem mainly include the characteristics of cryospheric habitats and the impacts of habitat changes. In the cryospheric habitats， ecosystems include endophytic ecosystems of cryosphere and cryosphere-related ecosystems. According to whether the biological communities inhabit the cryosphere elements directly， ecosystems can be classified as endophytic ecosystems of the cryosphere， which are primarily microorganism-dominated， and cryosphere-related ecosystems. The endophytic cryospheric ecosystems mainly include glaciers （ice sheets）， snow， frozen ground， sea ice， and other endophytic ecosystems， with microbial communities mainly inhabiting the ice and snow. The endophytic ecosystems of glaciers can be further divided into three types， i.e.， the supraglacial， englacial， and subglacial ecosystem. Similarly， the endophytic ecosystems of snow and sea ice consist of surface， interior， and bottom ecosystems， while the biological communities of the endophytic ecosystems of permafrost mainly refer the microbes within the active layer and the permafrost. In addition to the endophytic ecosystems of the cryosphere with the biological communities mainly are dominated by the microorganisms， the biological communities of the cryosphere-related ecosystems are diverse. Frozen soil is closely related to vegetation. The hydrothermal process and carbon-nitrogen cycle in the active layer of permafrost directly affect vegetation. Freeze-thaw cycles can directly cause root cell damage and reduce nutrient absorption due to root damage. Freeze-thaw process can affect nutrient retention and increase soil moisture during the growing season. Vegetation changes will affect herbivores. In addition， many organisms such as insects， mammals and birds also live in permafrost regions， and these species may also be affected by the permafrost degradation. Sea ice is the bridge connecting land and marine ecosystems. In addition to endogenous ecosystems in sea ice， sea ice affects both terrestrial and marine ecosystems. The impact of sea ice on terrestrial biological communities is mainly through sea ice benthic organisms and sea ice bridges， providing food， predation sites， and habitats for terrestrial animals. Sea ice benthic rich plankton is the food source for penguins， sea ice provides seabirds with access to food resources. Globally， cryosphere elements provide a habitat for various cold-adapted， cold-tolerant， or psychrophilic organisms on a wide range of scales. Many plants and animals use glaciers， ice caps， snow， frozen ground， sea ice， and other cryosphere elements as their habitats for living and survival. Changes in the elements of the cryosphere will affect the physical environment， energy and nutrient transport， and biogeochemical processes of the biome， thereby affecting the stability of the ecosystem. Overall， the cryosphere ecosystems are special ecosystems. Under the impacts of persistent global warming， the cryosphere is undergoing rapid changes， and resulting in changes of cryosphere habitats， which further affect the biological communities within the cryosphere and closely related to the cryosphere at different temporal and spatial scales. Therefore， understanding the cryosphere ecosystem systematically and clarifying research concepts are the scientific basis for properly study the impacts of changes in the cryosphere on biology. Current researches on cryosphere ecosystems mainly focus on the responses of different species， such as terrestrial plants and animals， freshwater organisms， marine organisms， coastal animals， and avian animals， to cryosphere changes. In the future， the impact of cryosphere changes on biomes should be quantitatively analyzed， and the time and spatial scale of the impacts of changes in various elements of the cryosphere on biomes should be considered.
As the headwaters of the Yangtze， Yellow and Lancang River， the Three Rivers Source Region （TRSR） is an important water conservation region and ecological barrier in China. In the context of climate change， the widespread frozen ground in the TRSR degrades significantly which exerts profound effects on vegetation changes and ecological environment. However， the vegetation variation characteristics and their response to climate and frozen ground change in the recent 20 years remain largely unknown. Based on the datasets for vegetation， climate and soil freeze-thaw during 2001—2020， the variation characteristics of vegetation phenology and their response to climatic and soil freeze-thaw factors over the past two decades on the TRSR were analyzed. The results show that the normalized difference vegetation index （NDVI） showed a spatial pattern of high values in the southeast and low values in the northwest on the TRSR. During 2001—2020， the vegetation on the TRSR showed an overall greening trend， with the growing season NDVI increasing by 0.017 per decade. The vegetation phenology also changed significantly， and the lengthening of the duration of the growing season ［6.3 d·（10a）-1］ was mainly contributed by the advances of the start of the growing season （SOS） ［4.9 d·（10a）-1］. Based on the results from statistical analysis， air temperature and precipitation were the most important dominant factors for growing season NDVI， and the sensitivities of NDVI to precipitation were larger in the relatively warm-dry region with relatively higher air temperature and lower precipitation； precipitation before the growing season was the most important dominant factor for SOS. The impact of soil freeze-thaw changes on vegetation growth showed spatial heterogeneity. The prolonged soil thaw duration could suppress vegetation growth in the relatively warm-dry region. In general， the rates of vegetation greening and growing season lengthening were higher in the seasonally frozen ground region than those in the permafrost region across the TRSR. However， permafrost degradation could suppress vegetation growth by reducing surface soil moisture content， which was more evident in the regions where permafrost was unstable or degraded into seasonally frozen ground. The results of this study could provide scientific reference for understanding vegetation and phenology changes in alpine cold frozen ground regions represented by the TRSR.
Under climate change， snow depths， snow density and snow water equivalent on global scale are undergoing numbers of changes， which further affect the hydrothermal status， biogeochemical cycle process， and structure and function of terrestrial ecosystems. The present paper reviews the current results of snow cover changes including snow depths， periods of snow cover in the Northern Hemisphere， ecologically experimental methods to simulate snow change and their pros and cons， the interaction between snow and ecosystem， and the effects of snow changes on soil nutrient turnover， soil fauna and microorganisms， root biomass and functional traits in three major terrestrial ecosystem types （grassland， shrub and forest） and their mechanisms. Thus， the main conclusions we have are： snow has the most significant warming effects in the depth of between 40 and 70 cm. The higher albedo under no snow or depth less than 40 cm may significantly reduce heat reaching the ground surface， thus soil temperature obviously decreased， and even the frequency and depth of soil freezing and thawing in winter are increased so to compensate for the huge heat loss. When the snow depth is greater than 70 cm， the latent heat of snow melting causes surface heat loss， which reduces soil temperature and weakens the heat preservation effect of snow. Increased snow cover leads carbon and nitrogen loss by accelerating soil carbon and nitrogen cycling， especially in humid habitats， but this is not the case in arid habitats. Snow reduction mainly affects root mortality， microbial activity， soil organic matter accumulation， nitrification and denitrification processes by acting on soil freeze-thaw cycles， thereby affecting soil nutrient turnover. In the grassland ecosystem， the response of soil available phosphorus content to the increase of snow cover is regulated by the ecosystem water conditions， that is， the available phosphorus increases in the humid habitat and decreases in the arid habitats. The increase of snow cover promotes the growth of plant roots in grassland ecosystem by improving the nutrient availability and water availability of shallow soil and changing the root morphological characteristics. The impact of reduced snow cover on plant root growth in ecosystems depends on the dynamic balance between negative effects （root damage and death） and positive effects （increased soil nutrient availability）. Compared with grassland and forest ecosystem， the response of plant root growth to snow cover change in shrub ecosystem was more stable. We also summarize the shortcomings and future trends in the study of effects of snow cover on belowground ecosystems. First of all， the response of phosphorus to snow cover increase is different from that of carbon and nitrogen， and it affects the coupling mechanism of soil carbon， nitrogen and phosphorus through influencing the decomposition rate of organic matter， which is very important for underground ecosystems. In the future， more attention should be paid to the response of soil phosphorus cycle and carbon， nitrogen and phosphorus coupling under snow cover change in different ecosystem environments. Secondly， at present， in the northern hemisphere， the effects of snow cover on root morphology and stoichiometric characteristics of plants in different ecosystems are not comprehensive. Further research on the response of root biomass， morphological characteristics and stoichiometric characteristics to snow cover changes in different ecosystems is still needed. Finally， to reveal the response mechanism of belowground processes in different ecosystems to snow cover changes， it is necessary to carry out more extensive and longer-term continuous field positioning observations and multi-factor comprehensive control experiments.
Dissolved organic carbon （DOC） plays a pivotal role in the global carbon cycle. At present， there are many studies about DOC in permafrost catchments of the Arctic regions， yet some studies about DOC in permafrost catchments of the Qinghai-Tibet Plateau （QTP）. To explore the spatiotemporal variability， sources， the responses to climate warming and permafrost degradation and influence factors of DOC in QTP river， here we conducted field investigations of DOC in 8 catchments （ZMD， TTH， YSP， FHS1~5） in the Yangtze River source region （YRSR）. The seasonal variations and source characteristics of DOC were deciphered using stream sampling， laboratory analyses， flux calculation， stable carbon isotopic technique （
As an important part of the natural ecosystem， grassland ecosystem provides important grazing resources for the development of livestock economy and plays a very important role in regulating climate change and maintains the balance of the ecosystem. Aboveground biomass （AGB） is an important indicator of the physiological state of grassland vegetation. Its size reflects the level of primary productivity of grassland， and is an important indicator of the energy cycle and material flow in grassland ecosystem， and plays an important role in the carbon cycle of terrestrial ecosystem. In recent decades， along with the rapid development of the livestock economy and global warming， the stability of grassland ecosystem has been reduced， the ecological environment has been degraded， and the aboveground biomass and carbon sequestration capacity of grassland are bound to be affected. Large-scale， dynamic and high-precision monitoring of aboveground biomass in grassland is of great importance to the accounting of grassland carbon stocks and the sustainable development of animal husbandry， and remote sensing technology， with its high spatial and temporal detection capability， provides the solution. Machine learning algorithms have been widely used in various research fields due to their superiority， efficiency， robustness and accuracy， and the use of machine learning algorithms for rapid， accurate and large-scale monitoring of grassland aboveground biomass is currently a hot research topic. As a result， the construction of an accurate aboveground biomass estimation model， the accurate estimation of aboveground biomass and the analysis of its spatial distribution characteristics can effectively measure the stability of grassland ecosystem and maintain the sustainable development and use of grassland ecological resources， providing a basis for the sustainable use and scientific management of grassland resources in the region， which is of great significance to the ecological security protection and sustainable development of animal husbandry in the region. In this study， the above ground biomass data of grassland in Xinghai County， Qinghai Province was used as the study area， and the random forest （RF） and extreme gradient boosting （XGBoost） methods were used to combine remote sensing data， climate data， topographic data and soil data with high spatial resolution. The aboveground biomass estimation models of grassland in Xinghai County were constructed， and the accuracy of the two models was evaluated using two accuracy verification indicators， namely R2 and root mean square error （RMSE）， to achieve high accuracy simulation and mapping of aboveground biomass of grassland and analyze its spatial distribution pattern. The results show that： （1） the accuracy of aboveground biomass estimation in grassland based on the XGBoost model （R2=0.75， RMSE=44.64） was higher than the simulation accuracy of RF （R2=0.72， RMSE=46.36）， and the aboveground biomass estimated by the XGBoost model was closer to the measured above-ground biomass values in grassland； （2） the spatial distribution of grassland aboveground biomass data estimated by the two machine learning models was similar to that of the measured grassland aboveground biomass， with high values of grassland aboveground biomass in the eastern part of the study area and the lowest values of grassland aboveground biomass in the western part， but the model simulations could better reveal the spatial heterogeneity of grassland aboveground biomass distribution； （3） in terms of spatial distribution characteristics， the XGBoost model estimated more detailed spatial variation in aboveground biomass of grassland， especially in the eastern part of the study area. Based on two machine learning algorithms， this study achieved high-precision （30 m spatial resolution） estimation and digital mapping of aboveground biomass in grasslands and analysed their spatial distribution patterns， which can provide scientific basis for monitoring grassland ecosystem and sustainable use of grassland resources， and is of great theoretical and practical significance for maintaining ecosystem balance and predicting the impact of future climate change on grassland ecosystem.
Forest soil in permafrost regions is an important carbon pool， and it is of great significance for the global balance of CO2， N2O and CH4. The freeze-thaw cycles （FTCs） is an important feature in permafrost regions and the effects of FTCs on the decomposition of organic matter and greenhouse gas emissions from the different forest types of humus soil need further research. In this study， a 42-day indoor cultivation experiment was carried out on the humus soils of Xing’an Larix gmelinii forest （coniferous forest）， Betula platyphylla forest （broad-leaved forest）， and mixed forest by Xing’an Larix gmelinii forest and Betula platyphylla （mixed forest）. This study explored the effects of FTCs on the soil physicochemical properties， the organic carbon and organic nitrogen mineralization rates， and the greenhouse gas emissions. The FTCs led to an increase in soil dissolved organic carbon （DOC） in the forest humus layer， the Larix gmelinii forest， the Betula platyphylla forest and mixed forest increased by 314.8%， 91.4% and 43.2%， respectively. The temperature sensitivity （Q10） of soil organic carbon decomposition decreased significantly after FTCs in the short term， and CO2 emissions decreased significantly during the experimental period， which was decreased by 24.7%， 36.4% and 29.5% at 25 ℃， respectively. At the same time， freeze-thaw effect has a certain inhibitory effect on the absorption of CH4， but has no obvious effect on N2O. The FTCs action reduced the global warming potential （GWP） of greenhouse gases in forest soils in which CO2 inhibition plays an absolutely dominant role， and the influence of FTCs action on the humus layer of Betula platyphylla forest was more obvious than that of the other two forest types.
Rock glaciers are a common periglacial landform in alpine regions， and their distribution and movement patterns provide crucial insights into the state of mountain permafrost. Daxue Shan is situated at the transition zone between the Tibetan Plateau and Sichuan Basin， where the cold and rainy climate is ideal for the formation of maritime glaciers and periglacial landforms. However， the distribution of rock glaciers in this region has not been fully explored. In this study， we utilized 90 Sentinel-1A ascending SAR images acquired between June 2019 and June 2022 to apply the time-series interferometric synthetic aperture radar （InSAR） technique for deriving the mean annual surface displacement velocity over the southern Daxue Shan. We also comprehensively considered the kinematic characteristics from InSAR measurement and geomorphological characteristics from optical image interpretation to compile a rock glacier inventory. We classified rock glaciers into four types according to the geomorphological units at the upslope region directly connected to them， namely， talus-connected， debris-mantle-slope-connected， glacier-connected， and glacier-forefield-connected types. A total of 860 rock glaciers were compiled， with 67% being talus-connected and only 4% being glacier-connected. The rock glaciers are mainly concentrated in the northern part of the study area， indicating a more favorable periglacial environment for rock glacier development. Talus-connected rock glaciers are widespread in the study area with uniform distribution. Glacier-connected and glacier-forefield-connected rock glaciers are primarily distributed in the northeast region with strong glaciation， while debris-mantled-slope-connected rock glaciers are mainly found in the southwest area. Geomorphic and kinematic parameters were calculated to analyze the development and motion pattern of these rock glaciers. The area， length， and slope angle of rock glaciers are concentrated between 0.04~0.12 km2， 250~700 m， and 12°~24°， respectively. Compared to the other two types， glacier-connected and glacier-forefield-connected rock glaciers are relatively larger， longer， and located in gentler slopes. The distribution altitudes of the local rock glaciers are between 3 638~5 107 m， with no rock glacier development identified below an altitude of 3 915 m in this research area. The majority （74%） of rock glaciers face west， northwest， north， and northeast， with similar aspect distribution patterns shown by the four types of rock glaciers. Notably， no glacier-connected rock glacier faces southeast， south， and southwest. Overall， rock glacier activities in the study area are low， with a maximum and mean downslope velocity of 250.25 mm·a-1 and 21.97 mm·a-1， respectively， and most of them creep below 100 mm·a-1. Rock glacier activities in the northern part of the study area are relatively more evident， indicating that abundant ice within rock glaciers interacts strongly with hydrothermal conditions in this region. Correlation coefficients between the geomorphic and kinematic parameters of rock glaciers were calculated. The correlation between the geomorphic parameters of rock glaciers reflects their development characteristics. However， no evident linear relationship was found between the geomorphic and kinematic parameters， indicating a complex mechanism of rock glacier dynamics. Based on the empirical model in the previous study， the water storage of rock glaciers in this area was primarily assessed at 0.963~1.445 km3. This study presents the first rock glacier inventory in the southern Daxue Shan， revealing the distribution and motion pattern of rock glaciers in this region. The results provide essential and reliable data for further studies on rock glaciers' hydrological contributions and mountain permafrost evolution in the southeast Tibetan Plateau. The method used in this paper for rock glacier inventory can also provide a feasible technical route for compiling large-scale rock glacier inventories in western China.
Under the influence of climate warming， permafrost on the Qinghai-Tibet Plateau is undergoing extensive degradation， mainly manifested by the frequent occurrence of solifluction events， which have profound impacts on ecosystems and local infrastructure. Accurate identification of solifluction can help to understand the occurrence and evolution mechanism of solifluction. Despite the progress of deep learning-based solifluction identification in recent years， the identification capability of machine learning algorithms in this field still needs to be explored. In this study， an optimized object-based solifluction identification algorithm based on GF-2 satellite remote sensing data is constructed， introducing spatial information such as texture and geometry to assist the identification of solifluction， and improving the misclassification problem of the identification model based on object-based technology. In addition， ensemble learning integrates the advantages of different machine learning models to obtain identification accuracy no less than that of commonly used deep learning models. The results show that the recursive feature elimination （RFE）-based feature selection algorithm eliminates redundant features in the multidimensional feature dataset and demonstrates that texture and geometry information are effective data complement for solifluction identification. Among the optimized object-based machine learning models， random forest （RF） has the highest recognition accuracy with an overall accuracy of 87.43%. McNemar’s test showed that the ensemble machine learning model significantly improved the identification accuracy of solifluction compared to the single model， with an overall accuracy of 93.14%. The statistical analysis of the topographic characteristics of solifluction in the study area revealed that the solifluction mainly occurred between 3200~3500 m a.s.l.， with slopes ranging from 5°~25°， and the major slope orientations were northeast， north and northwest. The ensemble machine learning model proposed in this study provides an effective identification method for the increasingly frequent solifluction in the permafrost region of the Qinghai-Tibet Plateau， and can produce label data for the deep learning identification model based on satellite remote sensing images.
The maximum frost depth （MFD） is an important indicator for seasonally frozen ground. The forecasting of future changes in the MFD is important for understanding environmental changes， ecological protection， agricultural and livestock production， and engineering construction in the Third Pole. However， future changes in the MFD of seasonally frozen ground in the Third Pole have not been report yet. In this study， an ensemble simulation strategy （i.e.， the model is run 200 times， and the arithmetic mean of 200 times is used as the final simulation result） is used to simulate the MFD for the 2050s and 2090s under four SSP scenarios， i.e.， green scenario （SSP126）， historical scenario （SSP245）， weak mitigation scenario （SSP370） and rapid and unconstrained growth scenario （SSP585） using a well-trained support vector regression model for the baseline period （2001—2010）. The spatial and temporal characteristics of the MFD in the seasonally frozen ground area from the baseline period to 2050s and 2090s are analyzed. In addition， the characteristics of the MFD change in different elevations and biomes （mountain grasslands and shrubs， temperate grasslands and shrubs， desert and dry shrubs， and temperate coniferous forests） are analyzed. The results indicated that the future MFD in the Third Pole would continue to decrease by 10.41 cm， 24.00 cm， 37.71 cm， and 47.71 cm， respectively， from 2001 to 2100 under the SSP126， SSP24， SSP370， and SSP585 scenarios， excluding the degradation of permafrost to seasonally frozen ground. The change of MFD was elevation-dependent and related to biomes. The decrease rate of MFD increased with the increase of elevation， but the decrease rate of MFD gradually decreased after the elevation exceeded 5 000 m， which is consistent with the elevation dependence of warming. Under the four SSP scenarios， the decrease rate of MFD is highest in the mountain grassland and shrub area， with an average decrease of 1.80 cm， 3.77 cm， 5.77 cm， and 7.24 cm per decade， respectively， by the end of the century. Among the 21 basins related to the Third Pole， the largest MFD reduction occurred in the Qinghai Lake basin. The future MFD dataset in 2050s and 2090s can be free downloaded through the National Tibetan Plateau Data Center （DOI：10.11888/Cryos.tpdc.273002）. The results provide basic data and information for understanding the future seasonally frozen ground change in the Third Pole and its ecological and hydrological effects in the context of global warming.
In order to delicately characterize the surface freeze-thaw state and its annual cycle， the trend of long time series changes and its interaction with terrestrial ecosystem， a high-resolution （1 km） surface freeze-thaw state identification is required. The low spatial resolution of passive microwave cannot finely describe the spatial variation of surface freeze-thaw state； active microwave （synthetic aperture radar） can identify the surface freeze-thaw state with high resolution， but its low temporal resolution cannot accurately capture the surface freeze-thaw transition in spring and autumn. The fusion of active and passive microwave remote sensing information and the development of high-resolution surface freeze-thaw state recognition algorithms are of great significance for the detailed characterization of surface freeze-thaw state and the production of high-resolution surface freeze-thaw remote sensing products. In this paper， the soil temperature and moisture observation network in the central Tibetan Plateau is used as the research area， and the soil temperature data in the central Tibetan Plateau is divided into two categories with 0 ℃ as the threshold， and the “true value” of surface freezing and thawing state is determined （indicating the surface freezing state and surface melting state， respectively）， and the fusion of active and passive microwave remote sensing information and auxiliary spatial and temporal information based on random forest model is used to build a fusion algorithm. Based on the random forest model， a high-resolution （1 km） surface freezing and thawing state recognition algorithm is constructed by fusing multiple sources of data such as active and passive microwave remote sensing information and auxiliary spatial and temporal information. In order to analyze the influence of active microwave information on the recognition accuracy of the algorithm， four input feature combinations corresponding to four random forests are designed for the lifting track， and the overall accuracy and confusion matrix of the random forest training stage are used as evaluation indexes to determine the best input feature combinations and to analyze the importance of the input feature combinations. Finally， the random forest-based surface freezing and thawing state recognition model is obtained and input to the corresponding 1 km data set to obtain the high-resolution surface freezing/thawing state and its probability， which is verified using the “true value” of the surface freezing/thawing state independent of the training data. The results show that the recognition accuracy is 100.0%/97.8% for the surface freezing period （January to March each year）， 99.1%/99.4% for the surface thawing period （June to October each year）， 82.0%/74.1% for the surface freezing/thawing transition period （April to May each year）， and 95.0%/74.1% for the surface thawing transition period （November to December each year）， with/without active microwave information， respectively. The accuracy of identification during the period of surface freezing and thawing transition （April to May every year） is 95.1%/90.0%， and the accuracy has decreased. The reasons for this were mainly due to the influence of elevation， slope direction and soil texture. By ranking the importance of the input variables and analyzing their contributions in the random forest， DOY （day of year），
Soil organic carbon （SOC） is an important part of the carbon cycle of the terrestrial ecosystem， and also an important indicator to evaluate regional soil quality， land degradation and crop yield. The estimation of soil organic carbon content in alpine ecosystem is of great significance for soil carbon pool accounting and soil quality evaluation in alpine regions. Visible and near infrared reflectance spectroscopy （Vis-NIRS） has been proven to be an efficient method for predicting soil properties， and the combination of spectral transformation and models can improve the simulation accuracy of SOC， but the best combination of mathematical transformation methods and inversion model are unknown for alpine ecosystem soil. In this study， Three Rivers Source Region （TRSR） of Qinghai-Tibet Plateau as the study area and 272 soil field samples were collected. The collected samples air-dried， sieved through 0.25 mm mesh， handpicked to remove roots. Then the well-done soil samples were used to indoor chemical analysis and soil spectrum test. SOM was determined by potassium dichromate sulfuric acid external heating in laboratory. The soil spectral curves were tested using US ASD FieldSpec 4 Standard Res in a dark room. The original spectral curve undergoes a series of mathematical transformations， such as first-order differential （FD）， second-order differential （SD）， reciprocal logarithm （RL）， Continuum removal （CR）， multivariate scattering correction （MSC）， etc. The characteristic band were selected based on correlation analysis between SOC content and spectral reflectance. The hyperspectral inversion models of SOC content were built using four methods， namely， multiple linear regression （MLR）， partial least squares regression （PLSR）， support vector machine （SVM） and random forest （RF）. Meanwhile， the best spectral transformation and model combination mode for SOC content inversion were determined by the comparison of all the models based on the validation sets’ simulation accuracy and stability evaluation. Results showed that： （1） SOC content in the Three Rivers Source Region was high and the average value was 34.68 g·kg-1， which was significantly higher than the first grade of national standard for nutrient grade I （23.2 g·kg-1g） of China. Meanwhile， SOC content in different vegetation types or different soil types was various different， and the order of soil organic carbon content from large to small in different vegetation types is Alpine meadow， forest， Alpine meadow grassland， Alpine grassland， temperate grassland， farmland and desert grassland. SOC content of Alpine meadow （average value was 48.59 g·kg-1） was significantly larger than desert grassland （15.38 g·kg-1）， statistical analysis in this study also found the coverage and biomass of Alpine meadow （90%， 130 g·kg-1） significantly larger than desert grassland （46%， 84 g·kg-1）. SOC content of Meadow swamp soil was highest with 59.12 g·kg-1， grass felt soil also was high with 50.83 g·kg-1， the content of SOC in chestnut soil and calcic soil is lowest with 21.48 g·kg-1， and 20.77 g·kg-1， respectively. （2） In general， the inversion model constructed by first-order differential （FD） transformation obtained the highest simulation accuracy among the single mathematical transformation form （especially the model combination of FD-RF， with R2=0.86， RMSE=8.40， RPD=2.64； in its validation set）. Because FD transform can significantly improve the characteristic spectral information of soil organic carbon， reduce noise， and increase the correlation between SOC and reflectance. Compared with a single mathematical transformation， the simulation accuracy of multiple mathematical transformation combinations is further improved （such as CRFD-RF combination， R2=0.87， RMSE=8.03， RPD=2.76； MSCFD-RF combination， R2=0.87， RMSE=8.14， RPD=2.72）. （3） Among the 4 simulation models， the overall simulation accuracy of random forest （RF） is the highest， among， CRFD-RF combination is the highest， and MSCFD-RF is the second. This study can provide a theoretical basis for SOC inversion and dynamic monitoring in Qinghai-Tibetan region using hyperspectral remote sensing. In the future， the formulation of different valid prediction models （covering several vegetation or soil types） is critical， as spectral characteristic band may vary from one type to another.
