Spatiotemporal variability of permafrost degradation on the Qinghai-Tibet Plateau

HuiJun Jin, DongLiang Luo, ShaoLing Wang, LanZhi Lü, JiChun Wu

State Key Laboratory of Frozen Soils Engineering, Cold and Arid Regions Environmental and Engineering Research Institute,

Chinese Academy of Sciences, Lanzhou, Gansu 730000, China

Spatiotemporal variability of permafrost degradation on the Qinghai-Tibet Plateau

HuiJun Jin*, DongLiang Luo, ShaoLing Wang, LanZhi Lü, JiChun Wu

State Key Laboratory of Frozen Soils Engineering, Cold and Arid Regions Environmental and Engineering Research Institute,

Chinese Academy of Sciences, Lanzhou, Gansu 730000, China

Based on data from six meteorological stations in the permafrost regions, 60 boreholes for long-term monitoring of permafrost temperatures, and 710 hand-dug pits and shallow boreholes on the Qinghai-Tibet Plateau (QTP), the spatiotemporal variability of permafrost degradation was closely examined in relation to the rates of changes in air, surface, and ground temperatures. The decadal averages and increases in the mean annual air temperatures (MAATs) from 1961-2010 were the largest and most persistent during the last century. MAATs rose by 1.3 °C, with an average increase rate of 0.03 °C/yr. The average of mean annual ground surface temperatures (MAGSTs) increased by 1.3 °C at an average rate of 0.03 °C/yr. The rates of changes in ground temperatures were -0.01 to 0.07 °C/yr. The rates of changes in the depths of the permafrost table were -1 to +10 cm/yr. The areal extent of permafrost on the QTP shrank from about 1.50×106km2in 1975 to about 1.26×106km2in 2006. About 60% of the shrinkage in area of permafrost occurred during the period from 1996 to 2006. Due to increasing air temperature since the late 1980s, warm(＞-1 °C) permafrost has started to degrade, and the degradation has gradually expanded to the zones of transitory (-1 to -2 °C)and cold (＜-2 °C) permafrost. Permafrost on the southern and southeastern plateau degrades more markedly. It is projected that the degradation of permafrost is likely to accelerate, and substantial changes in the distributive features and thermal regimes of permafrost should be anticipated. However, regarding the relationships between degrading permafrost and the degradation of rangelands, it is still too early to draw reliable conclusions due to inadequate scientific criteria and evidence.

QTP; permafrost degradation; ground temperatures; change rates

1. Introduction

1.1. Degradation of permafrost

Permafrost is the result of heat and moisture exchange among the earth surface processes closely related to the atmosphere, lithosphere, pedosphere, hydrosphere, and biosphere, and generally as a result of the past cold periods in the late Quaternary. Temperature changes in the shallow(＜100 to 200 m) depths are sensitive to climatic variability and surface energy budgets at time scales up to decades and centuries (Lachenbruch and Marshall, 1986; Beltrami and Taylor, 1994; Pavlov, 1994; Allardet al., 1995; Osterkamp and Romanovsky, 1999; Romanovskyet al., 2002, Fukuiet al., 2007a; Zhaoet al., 2010a). Shorter-term changes in climate and those in surface conditions which are important for modifying thermal and moisture regimes of permafrost can also result in significant, and sometimes dramatic, changes in permafrost conditions.

Permafrost degrades when its thermal regimes are adversely changed by climate warming and anthropogenic activities. The degradation of permafrost is generally indicated by rising ground temperatures and by reduced thickness and areal extent of permafrost. Concurrently, because of changes in the physical and mechanical properties of frozen soils, many infrastructures in northern and alpine regions will be affected (US Arctic Research Commission Permafrost Task Force, 2003). The degradation of permafrost can also result in profound changes in the ecological and land surface processes in the northern countries and highlands(Jorgensonet al., 2001; Harris C and Mühll, 2003; Harris Cet al., 2009). Accordingly, when studying the degradation of permafrost, one should compare the changes in the thermal regimes of permafrost using extensive long-term monitoring data. Unfortunately, this is more difficult in the case of permafrost due to the paucity of data across the large area of the Qinghai-Tibet Plateau (QTP; see Appendix 1 for a list of acronyms used in this paper).

The cryosphere on the QTP responds to and shows significant feedbacks to changes in climate, hydrological processes, and ecosystems in High and Central Asia. Under a markedly warming climate, permafrost has been degrading extensively and its rates have been accelerating (Jinet al.,2009; Zhaoet al., 2010a). The thawing of permafrost and melting of large-volume ground ice (9,528 km3) on the QTP will in fact diminish climatic warming (Zhaoet al., 2010b).This may also result in lowered levels of ground water, or lateral or downward drainage of water to deeper aquifers,which may adversely impact the alpine ecosystems in the drylands, leading to land desertification. In turn, this would change the heat and moisture budgets on the land surfaces through the altered land surface properties, such as the albedo, and feed back to the climate systems.

The degradation of permafrost is a complicated process due to its spatiotemporal variability and its variable time lag after the changes in air and surface temperatures. The time lag, the degree and extent of permafrost degradation, and the depths of permafrost resulting from the climate changes depend on the areal extents, patterns, rates, and amplitudes of the changes in the climate indicators, such as air temperatures and precipitation, are closely related to soil types,surface covers, ground ice, geothermal flows, and temperatures of the permafrost soils.

The large-scale changes in the southern and lower limits of permafrost are generally regarded as taking a long time.For example, in North America, the southern limit of permafrost (SLP) retreated extensively northwards in response to the post-glacial climate warming and this is still continuing (Halseyet al., 1995). The SLP is now 130 km farther north than it was 50 years ago in the James Bay region of northern Quebec, Canada (Thibault and Payette, 2010). A similar response occurred in the Eurasian continent (Jinet al., 2000a, b), particularly on the QTP and in the Himalayas(Fukuiet al., 2007b; Jinet al., 2007a). The SLP retreated 50-120 km in northeastern China during the last 60 years due to a combined influence of climate change and human activities (Jinet al., 2007b).

Recent monitoring data indicate that the warming and thawing of permafrost have occurred, are occurring, and are projected to continue to take place under a warming climate.However, there is considerable regional and temporal variability in the onset, magnitude, and rates of climate warming,and degradation of permafrost, as reflected in rising ground temperatures, increasing thaw depths, and disappearing permafrost patches. The warming of permafrost has not been observed in some other places such as in the central Yukon of Canada (Burn, 1998; Harris S, 2009), and actual cooling was observed in the eastern Canadian Arctic from the late 1980s to the mid-1990s (Allardet al., 1995).

1.2. Degradation of permafrost in central Asia

The alpine permafrost in central Asia occupies approximately 3.5 million km2(about 15%) of the total areal extent of permafrost in the Northern Hemisphere. It mainly occurs on the QTP and in the Tianshan, Altai, Khangai, Hovsgol,and Kentii Mountains (Figure 1) (Marchenkoet al., 2006;Zhaoet al., 2010a).

The International Permafrost Association (IPA) and geocryologists from Russia, Mongolia, Kazakhstan, China,Japan, and some other central Asian states have been mapping the permafrost in central and eastern Asia (Marchenkoet al., 2006). Monitoring and modeling research during the past decades indicates that climate warming has affected the thermal regimes of permafrost in central and eastern Asia.The transition zone between the mountain and latitudinal permafrost zones is found in the Hovsgol Mountains in northern Mongolia, where permafrost has been warming at a rate of 0.02 to 0.04 °C/yr under the combined influences of climate change and anthropogenic activities. In comparison with those in the 1970s and 1980s, permafrost degradation has accelerated significantly since the 1990s. A warming of 0.01 to 0.02 °C/yr has taken place in alpine permafrost, accompanied with a deepening of 20%-25% of the active layer thickness in the Tianshan Mountains during the past three decades. Modeling suggests that the lower limit of permafrost (LLP) has risen by 150 to 200 m during the 20th century and resulted in an areal reduction of about 18%(Marchenkoet al., 2007). However, some observations and numerical simulations conclude that permafrost can still occur or the relict permafrost can be preserved in coarse block fields or in the dense needle-leaved forests with above-zero (up to 4 °C) mean annual air temperatures(MAATs), or where the elevations, such as 1,800-1,900 m a.s.l., are much lower than the LLP, due to the thermal offsets of about 2.5 to 4.0 °C (Harris S and Pedersen, 1998;Gorbunovet al., 2004).

Mongolia has an areal extent of elevational and latitudinal permafrost of almost 1 million km2, and it has been subjected to remarkable changes (Kynickyet al., 2009).However, the permafrost in Mongolia and Trans- and Outer-Baikalia, particularly the Xingan-Baikal permafrost, is more sensitive to human disturbances to the surface conditions than to climatic changes (Jinet al., 2007b). The inactive ice wedges formed during the Last Glaciation Maximum (LGM)and the Neoglaciation are still preserved in many areas as far south as Yitulihe and Wuma in northeastern China, and the LLP in western Mongolia has moved upwards by only about 300 m since the LGM compared to 500-1,200 m in western China (Zhenget al., 1998). Although latitudinal permafrost in northeastern Mongolia is less sensitive to climate warming, the permafrost at high elevations is generally sensitive to climate warming. Significant degradation of permafrost has been observed during the last half-century, particularly in the Altai and the Khangai Mountains, and somewhat in the Kentii Mountains and the Xingan Mountains (Jinet al.,2007b; Kynickyet al., 2009). However, it is believed that cold air drainage and the winter atmospheric temperature inversion, together with surface conditions, further complicate the features of permafrost and their sensitivity to climate changes (Romanovskiiet al., 1991; Fukuiet al.,2007a).

Figure 1 Distribution of permafrost in central and eastern Asia, and locations of Thermal State of Permafrost (TSP) boreholes(revised from Marchenko et al., 2006)

1.3. Degradation of permafrost on the Qinghai-Tibet Plateau

The persistent climatic warming on the QTP observed since the mid-1980s has significantly and directly impacted the permafrost and cold environments (Jinet al., 2008a,2009). Rising air temperature has resulted in changes in the distributive features of permafrost, and has also led to changes in the permafrost soil itself, such as rising of ground temperatures, increasing in the thickness of the active layer,thinning of permafrost, melting of ground ice, and modificating of structures and textures of the permafrost soils(Wang S, 1993, 1997; Wang Set al., 1996; Jinet al., 2000a,2000b; Wang S and Jin, 2000; Yang and Ding, 2004; Wu Tet al., 2005, 2008). During the last few decades, increasingly more attention has been paid to a series of cold-regions environmental problems resulting from climate warming and subsequent permafrost degradation. As a result, their environmental impacts have become the focus of research programs on cold-regions environments (Wang Set al., 1996,1999, 2003; Cheng and Zhao, 2000; Wu Qet al., 2001, 2005;Nelson, 2003; Cheng and Wu, 2007; Jinet al., 2008a, 2009).Although the results are mixed, permafrost degradation may have contributed significantly to the degradation of rangelands (Harris R, 2010).

The authors systematically collected and analyzed the long-term data on the air and ground surface temperatures from six meteorological stations in the permafrost regions on the QTP, ground temperatures from 60 boreholes, and the permafrost table as revealed from 710 hand-dug pits and shallow boreholes. These data were obtained mainly by geocryologists and permafrost engineers during field explorations and by long-term monitoring along some major transects, such as that along the Qinghai-Tibet Highway(QTH, or G109, from Golmud to Lhasa). The data period is generally started from the 1970s, but the data were scarce and less accurate in the early periods. Since the 1990s, the data volume and quality have been improved greatly. In this study, decadal averages of the mean annual air, ground sur-face and ground temperatures, and their rates of inter-decadal changes were then computed for various periods of the instrument-measured 50 years. The temperatures of various permafrost types were divided into categories on the basis of the statistics of the rates of changes in ground temperatures in different periods and at various depths. Concurrently, evidence of changes in the permafrost table was obtained from the statistics for different periods and permafrost types. Thus, the spatiotemporal variability of permafrost degradation on the QTP under a warming climate was analyzed and discussed here, together with the correlations of the change rates in the annual averages of air, ground surface and ground temperatures, and the permafrost table.

2. Methods

2.1. Measurements and accuracy

Air temperatures are generally measured at a height of 1.5 m in the air, and ground surface temperatures at a depth of 0.05 m below the ground surface, at standard or automatic meteorological stations, with an accuracy of ±0.1 °C or better.

Prior to and during the 1970s and 1980s, ground temperatures were measured using Assmann thermometer chains encased in composite tubes. The thermometers were put in boreholes for at least 6 hours before they were skillfully and quickly brought up to surface. The changes in temperature readings were minor and were minimized by the Vaseline? (petroleum jelly) coating around the thermometers. The measurements generally had an accuracy of about ±0.1 °C. After the 1980s, ground temperatures were generally measured using thermistor cables, with an accuracy of at least ±0.1 °C. Since the 1990s, measurements were obtained using thermistor cables with platinum sensors and automatic data loggers (some still manually operated),with an accuracy of ±0.05 to ±0.1 °C.

The readings of electrical resistance, or more directly,temperatures, were taken by meters (such as Fluke 87/89)from the thermistor cables permanently installed in the steel or PVC-cased boreholes, or frozen in soils. This was generally done when the equilibrium of thermal regimes had been reestablished after the cables were emplaced in the boreholes.This generally took at least 6 hours for the long-established boreholes. Borehole temperatures were generally measured at least 3-6 months after their drilling, depending on the depths of the boreholes. Ground penetrating radar (GPR)was sometimes used for detecting the spatial and temporal changes in permafrost distribution, thawed depths, and moisture contents. Most reliable methods for exploring the occurrence of permafrost include direct drilling or excavations of hand-dug pits or trenches, and subsequent measurements in ground temperatures and ice contents, either for one-time or for long-term, periodic or repetitive processes.Water contents (i.e., the total water weight divided by the total soil weight) were generally observed by naked eye in the field in the descriptions, and then measured for the bulk density in situ. Soil samples from the excavations or wellbores were then taken for the dry density measurement when the samples were dried to constant weight in laboratories,and for other analyses.

2.2. Computation methods

If not otherwise mentioned, all averaging is the arithmetic average. When studying the trends of change, generally the linear regression is used. The slopes of the regressed lines are regarded as the trend rates of change.

The freezing index (FI) or thawing index (TI) are defined as the cumulative number of degree-days when air or ground surface temperatures are below or above 0 °C, respectively. The FI and TI have been widely used to predict permafrost distribution and to estimate the maximum depth of thaw in frozen ground (e.g., Harris S, 1981; Nelson and Outcalt, 1987; Nelson and Anisimov, 1993; Nelsonet al.,2002). The geothermal heat flux, or the product of the thermal gradient and the heat conductivity of soils, is defined as the amount of heat moving steadily outward from the interior of the earth through a unit area in unit time. It is generally obtained from the measurement of ground temperatures at deeper depths. The coefficient of soil heat balance is the ratio of the heat input to heat output into a defined domain. The rates of change (in °C/yr) in ground and air temperatures are derived using the change of temperature (in °C) over the period (in years, or year) at various depths or in the atmosphere.

2.3. Criteria for classifications

The classification of frozen ground in China (Qiuet al.,2002) is largely based on the areal extent of permafrost in a given region and differs slightly from those in Russia (the former USSR) and in North America, as follows: continuous permafrost (areal extent of permafrost: 70%-90%); discontinuous permafrost (30%-70%); sporadic and patchy permafrost (＜30%); and mountain permafrost. Other classifications are: deep seasonally frozen ground (＞1 m); shallow seasonally frozen ground (＜1 m); short-lived frozen ground(less than two weeks); and unfrozen ground. On the QTP,the zones of permafrost are delineated as follows: mountain permafrost in the Altun-Qilian Mountains; seasonally frozen ground in the Qaidam Basin; continuous permafrost in the northern part of the southern Qinghai and northern Tibet Plateau; discontinuous permafrost in the southern part of the northern Tibet Plateau; mountain permafrost in the Himalayas; and sporadic and patchy mountain permafrost on the eastern peripheries of the QTP (Zhouet al., 2000).

Thermal stability can reflect the changes in frozen ground in its natural state and with human activity, and it has a close relationship with mean annual ground temperature(MAGT), temperature at the permafrost table, and seasonal thaw depth (Wu Qet al., 2002). Accordingly, permafrost is classified as thermally stable (MAGT＜-3 °C), metastable(-3 °C≤MAGT＜-1 °C), unstable (-1 °C≤MAGT＜-0.5 °C),and very unstable (≥-0.5 °C) on the QTP (Zhouet al., 2000),in contrast to the classification of permafrost in North America: stable (＜-5 °C), metastable (-5 °C≤MAGT≤-2°C), and unstable (＞-2 °C), which is for a much larger spatial scale (Harris S, 1986). This classification largely corresponds to the classification on the basis of the areal extent of permafrost. Permafrost soils are also classified on the basis of ice content: ice-poor, ice-medium, ice-rich, ice-saturated,and ice layer with soil inclusions (PRC Ministry of Construction, 2001), but the volumetric ice contents of soils may vary considerably even within a same soil category. The degradation modes of permafrost generally include downward, upward, and lateral, which can be one or a combination in different zones of permafrost distribution or vertical profiles (Jinet al., 2006a, b; Wu Jet al., 2009).

2.4. Modeling and forecasting

In mountains and on plateaus, the modeling, mapping,and predicting of elevational permafrost have been more active in recent years. Most of these models, either empirical or process-based, aim at describing the distribution of permafrost in the rugged terrains, with various degrees of successes and deviations. This is attributed to the fact that the boundary conditions and parameters needed for developing reliable models are either difficult to obtain or are unevenly distributed in space and in time. Although these models and predictions are still quite preliminary, they are being upgraded almost monthly using rapidly developing remote sensing, geographic information systems (GIS), and other space-detecting and simulating technologies, which can simultaneously cover much larger areas and with better resolutions.

On the QTP, most of these permafrost models fall into the following categories on the basis of MAAT, FI and TI,MAGT, the digital elevation model (DEM) of LLP, and the equivalent elevational model (EEM). Due to the complicated boundaries conditions, many assumptions have to be made, such as the neglected lag time for changes in the permafrost in response to climate change; no changes for other geographical components such as deserts, glaciers, and lakes;the statistical relationships between the LLP and MAAT; and the atmospheric elapse rate. In the simulations, many models choose to use the projected climate change scenarios presented in Intergovernmental Panel on Climatic Change(IPCC) reports, or by other General Circulation Models(GCMs), and heat transfer models and numerical solutions are generally applied in studying the responses of permafrost to climate change, in the vertical profiles. These models have projected the various degrees of degradation of permafrost on the QTP during the next century, such as 45% to 50% in areal extents (Li and Cheng, 1999; Nanet al., 2003;Cheng and Wu, 2007; Cheng and Liu, 2008), assuming that the present-day warming rate continues constantly for that period of time.

3. Study region

The major concerns of this paper include the spatial and temporal distribution and degradation of permafrost on the central and eastern QTP during the past 50 years. The pertinent data were generally obtained along the Qinghai-Tibet Highway (QTH, or G109) or the Qinghai-Tibet Railway(QTR), the Qinghai-Kang (Western Sichuan) Highway(QKH, or G214) from Xining to Yushu, and the Xinjiang-Tibet Highway (XTH, or G219) from Yecheng, southeastern Xinjiang Uygur Autonomous Region to Lhasa, Tibet Autonomous Region (Figure 2).

The climate on the QTP displays a three-dimensional zonation of latitude, longitude, and elevation. The general characteristics of the geomorphology of the Plateau include northwestward-rising mountains, high plateaus, and intermontane basins alternating with wide valleys at elevations averaged at 4,500 m a.s.l., and surrounded by high mountains with deeply incised valleys. The major geomorphologic units include, from north to south: Altun to Qilian Mountains, Qaidam Basin, southern Qinghai to northern Tibet Plateau, southern Tibet Plateau, and the Himalayas.Numerous high mountains and deep gorges are present on the eastern and southeastern edges and in the Hengduan Mountains.

The rapid uplift of the QTP during the Quaternary greatly influenced the atmospheric and land surface environments on the Plateau, its contiguous regions, and eastern, southern, and central Asia. It profoundly impacted the distribution and degradation of permafrost. The uplifting helped the QTP achieve a general elevation of 4,000 to 4,500 m a.s.l. by the Late Pleistocene. The high elevation and large areal expanse provide the crucial low temperatures for the development of the permafrost. In the cold season (October to the following April), the Plateau is under the control of the Westerlies and the Siberian High. In the warm season (May to September), it is greatly influenced by the Indian and Pacific monsoons. In particular, the moisture-laden air masses bring significant precipitation to the southern and southeastern Plateau, which declines northwestwards. As a result, the heat-moisture regimes and consequent climate and vegetation display horizontal and vertical zonation, with a successive spectrum from subtropical forests in the southeast, to meadows, steppes, semi-deserts,and deserts in the northwest.

The 1,150-km-long QTH (G109) traverses representative permafrost areas in the interior of the QTP. The 650-km section from Xidatan in the vicinity of the Highway Maintenance Squad Station 60 (HMSS 60) to Liangdaohe(HMSS 125) is in the permafrost area. Continuous permafrost occurs along the 560-km-long section from the Kunlun Mountain Pass to the Amdonanshan Mountains (near HMSS 116). From Xidatan to the Kunlun Pass, and from the Amdobeishan Mountains to Liangdaohe, sporadic permafrost occurs at elevations above 4,125-4,250 m a.s.l. and 4,640-4,680 m a.s.l., respectively.

Figure 2 Schematic map of the study region indicating the distribution of permafrost and major establishment of transects, study sites, and experimental stations on the interior and eastern QTP. The coordinates and locations of boreholes and other less known sites are included in the map in Appendix 2.

The QKH (G214) skirts the permafrost zone on the northeastern edges of the Plateau, traversing many mountain ranges, such as the ?la, An?emaqên, and Bayan Har Mountains, and intermontane valleys and basins, such as the Gonghe, Madoi, and Yushu Basins. These topographical features result in alternating distribution belts of discontinuous, sporadic, and patchy permafrost and seasonally frozen ground. Elevational permafrost is found mainly in the sections passing the HekaNanShan, ?la, Jiangluling,An?emaqên, Bayan Har, and eastern Tanggula Mountains.The total length of permafrost along the QKH is about 330 km. Permafrost is best developed in the section of Bayan Har Mountains, with a total length of the QKH of 124 km.The LLP rises southwards along the QKH from 3,840 to 3,900 m a.s.l. in the HekaNanShan Mountains in the north,3,850 to 3,900 m a.s.l. in the ?la Mountains, and 4,000 to 4,050 m a.s.l. in the Jiangluling Mountains, to 4,150 to 4,250 m a.s.l. in the Bayan Har Mountains and 4,500 to 4,600 m a.s.l. in the Eastern Tanggula Mountains (Wanget al., 1999;Fanget al., 2009).

Active fault activities differentiate the development and evolution of permafrost on the edges of the QTP. As a result,permafrost is relatively thin and warm with numerous taliks.To the northwest, alpine permafrost is thicker and colder than that on the high plains and plateaus. However, the thermal stability of permafrost is complicated because of the combined influences of soil types, moisture, surface cover,and resultant differences in the thermal regimes on the ground surfaces and soil thermal conductivity.

The primary focus of this paper is the spatial and temporal variability in the degradation of permafrost along the QTH in the interior QTP (including the southern Qinghai Plateau and northern Tibet Plateau), where presently alpine semiarid climate prevails at elevations generally of 4,500-4,800 m a.s.l. on plateaus and 5,000-6,000 m a.s.l. in mountainous areas during the past 50 years. The pertinent data along the QKH and XTH, and from scientific expeditions to other areas of the Plateau, indicate that the MAATs range from -0.6 to -9.8 °C and the annual precipitation from 20 to 500 mm.

The latest data indicate that the total area of permafrost on the Plateau is 1.26 million km2(48.4% of the Plateau area)(Cheng and Wu, 2007). The western area of the QTP is dominated by alpine desert and semi-desert landscapes, and the eastern area by steppes and meadows. More than 70%-80% of the interior QTP has continuous permafrost where the MAATs are generally colder than -4 °C and the observed MAGTs range from -0.1 to -3.5 °C. Sporadic or patchy permafrost may have MAGTs as warm as +0.5 to +1°C due to its small thickness, usually indicating degrading permafrost and/or changing surface or subsurface conditions,surrounding the areas of continuous permafrost.

The observed thicknesses of permafrost range from 10 to 175 m but can be highly variable, especially in areas of patchy permafrost generally less than 20 m in thicknesses,whereas in the very high mountains they can be more than 200-300 m. The observed depths of measurable annual change in ground temperature vary from 10 to 20 m in the natural state. There are three situations in the continuous permafrost zone. In hilly or alpine areas where the MAGTs range from -1 to -4 °C, the permafrost thicknesses range from 30 to 130 m; in the high plateau river valleys of the sources of the Yangtze River, where the MAGTs vary from-1 to -1.5 °C, permafrost is generally thinner than 60 m;and in other river valleys with MAGTs from 0 to -1 °C,permafrost is generally less than 50 m thick. The MAGTs in talik zones beneath large rivers generally range from +0.5 to+1.0 °C and are dependent on both the widths and, to some extent, the depths of the rivers.

Along the QTH, the LLP is at 4,150-4,250 m a.s.l. in the north and 4,450-4,560 m a.s.l. in the south. Along the NW-SE direction XTH in the Western Kunlun Mountains,the isolated patches of permafrost are found in Dahongliutan at an elevation of 4,250 m a.s.l. and with an MAAT of -2.6°C, which can be regarded as the northern LLP. About 30 km southward, the elevation rises to about 4,450 m a.s.l. and permafrost becomes continuous. The XTH traverses southeastwards across the western QTP, and the southern boundary of continuous permafrost is reached at an elevation of 4,500 m a.s.l. and with an MAAT of -5.0 °C. About 30 km to north of Domar (4,630 m a.s.l.), at an elevation of about 4,430 m a.s.l., the southern LLP (i.e., the occurrence of isolated patches of permafrost) is reached. Therefore, a total length of 331 km along the XTH is in the permafrost zone(Li Set al., 1998).

On the eastern QTP, alpine permafrost occurs in the Burhan Budai, An?emaqên, and Bayan Har Mountains of the Eastern Kunlun Mountains, with LLPs at 3,840-3,900 m a.s.l. in the HekaNanShan Mountains, 4,000-4,050 m a.s.l.in the Anyêmaqên Mountains, 4,150-4,250 m a.s.l. in the Bayan Har Mountains, and 4,250-4,300 m a.s.l. in Yushu County, Qinghai Province. For every 1-degree southward decline in northern latitude, the rise of LLP is 120-130 m along the QTH (32°N-36°N, 92°E-94°E), 130-145 m along the QKH (33°N-37°N, 97°E-100°E), but only about 80 m along the XTH (34°N-37°N, 78°E-80°E) (Zhou, 1965;Zhouet al., 2000). The LLP also decreases eastwards. For example, it is at 4,500 m at the Karakorum and Western Kunlun Mountains, 4,300-4,350 m a.s.l. in the Muztag Mountains (36°N-37°N, 87°E-88°E) (Li Set al., 1998),4,200 m a.s.l. at Xidatan, and 3,900 m a.s.l. in the HekaN-anShan Mountains (i.e., for every increase of 1°E, the LLP declines by about 27 m). In the Kunlun Mountains, the MAGTs drop with elevation at rates of 7-8 °C (MAGT)/km(elevation) from Xidatan to the Kunlun Mountain Pass,while the average values on the QTP are 5-8 °C/km. The thickness of permafrost increases with elevation at rates of 200 m (thickness)/km (elevation) along the QTH and 100-200 m/km along the QKH.

In summary, on the QTP:

(1) High elevation and topography, intensive uplifts, the rich variety of sediments, cold climate, strong solar radiation,and the complex distribution of heat-moisture budgets control or influence the formation, distribution, and changes of permafrost.

(2) Regionally, permafrost is predominantly controlled by elevation and topography, as well as by the latitudinal and longitudinal zonation of climate.

(3) At smaller spatial scales, local factors such as vegetation, active faults, and lithology become increasingly important in controlling the distribution and ground temperatures of permafrost. Some local factors, such as vegetation and snow cover, and surface soil moisture and sand layers, could have dual impacts (increase or decrease the ground temperatures) on the thermal regimes of permafrost (Jinet al.,2008b; Lüet al., 2008). More often, these factors work together, further complicating the distribution of and changes in ground temperatures.

(4) Because of the tectonically young Plateau, active faults and tectonic movements have shaped the formation,storage, and discharge of ground and surface waters as well as the distribution of geothermal fluxes. These disrupt the three-dimensional zonation of permafrost and thermal regimes; they have formed numerous taliks of various genetic types; and they have facilitated the formation of many periglacial manifestations, such as icings, pingos, and thermokarst ponds. At the local scale, the distribution and thermal regimes of ground ice and permafrost are largely controlled by the geological structures.

(5) The main body of permafrost was formed during the Last Glaciation Maximum (LGM) in the Late Pleistocene.During the Holocene, the climate has fluctuated and permafrost also changed accordingly in areal extent, thickness, and temperature, particularly on the margins of the Plateau where shallow permafrost has been subjected to several long-term freeze-thaw cycles. The observed degradation of permafrost on the Plateau is a continuation and acceleration of the general trend of changes in permafrost since the LGM,the Neoglaciation, and the Little Ice Age (LIA).

4. Climate changes

4.1. Changes in air temperatures

The climatic environment and its changes are the primary external driver for the formation and evolution of permafrost. From the viewpoint of energy equilibrium, air temperature reflects an average thermal state of the atmosphere during a certain period in a given region. On large scales, air temperatures control the formation, evolution, and distribution of permafrost.

The decadal trends of air temperature at the six meteorological stations in the permafrost regions on the QTP during the period from 1961 to 2010 are analyzed in this paper (Table 1 and Figure 3). The results indicate that the decadal average of the MAATs in the studied regions was-4.2 °C in the 1960s and 1970s, rose to -3.9 °C in the 1980s, rose further to -3.2 °C in the 1990s, and finally rose to -2.6 °C in the 2000s. During the last 50 years, the decadal average of MAATs in the studied regions rose by 1.3°C, with an average increase of 0.03 °C/yr. This warming rate was higher than those (about 0.02 °C/yr) in the adjacent regions affected by only the seasonally frozen ground(Zhao, 2003), and was far greater than the average warming rate in China (0.005 °C/yr) (Qin, 2002). After the 1980s the decadal averages of MAATs surpassed the 50-year averages at each station. Since the 1990s, the decadal averages have been increasing more rapidly, particularly since 2001. Spatially, Madoi, Qümarlêb, and Amdo are on the eastern and southern edges of the patchy permafrost zone on the QTP. They have had increase rates of about 0.03 to 0.04 °C/yr (Table 1).

Table 1 Decadal averages (DA) and rates of change in the MAATs at six meteorological stations in the Plateau permafrost regions from 1961 to 2010

Figure 3 Variations of the MAATs at six meteorological stations in the Plateau permafrost regions during 1961-2010

4.2. Changes in ground surface temperatures

The thermal regime of permafrost is the result of the interactions of surface radiation, geothermal fluxes,evapotranspiration, and many other processes. Ground temperature controls and reflects the thermal regime of permafrost and plays important roles in studying soil heat fluxes and the permafrost dynamics. As an upper boundary condition, ground surface temperature is one of the most important indicators for the occurrence and development of permafrost, as well as an important parameter in thermo-moisture budgets and spectral characteristics in terrestrial processes, such as in the modeling, monitoring, and mapping of permafrost. The roughness and albedo of ground surfaces and moisture content and thermal conductivity of the soil can cause dynamic changes in surface temperatures.Permafrost degradation, land desertification, and decrease of vegetative coverage have altered the albedo of ground surfaces, the heat capacity of soils, and the sensible and latent heat transfer between the atmosphere and soil, resulting in rising surface temperatures on the QTP (Liuet al., 2009).

Using an analysis of major components, Wu T (2005)conducted a statistical study on the mean annual warm season (May to October), and cold season (November to the following April) ground surface temperatures and their annual amplitudes at the 101 meteorological stations on the QTP and periphery regions during the period from 1961 to 2000. Using these data, the interdecadal changes of these parameters were analyzed. The average of mean annual ground surface temperatures (MAGSTs) on the Plateau increased by 0.8 °C during the 40 years, with an average increase rate of 0.019 °C/yr. In the warm season, 88% of the stations had a trend of warming; in the cold season, 87%.During the 40 years, the average of surface temperatures in the warm season increased by 0.7 °C, with an average increase rate of 0.018 °C/yr, and by 0.8 °C in the cold season,with an average increase rate of 0.020 °C/yr.

Zhao (2003) studied changes in air and ground surface temperatures in regions affected only by seasonally frozen ground on the QTP during 1967-1997 (Table 2). The MAAT average of the QTP during the period rose by 0.6 °C, with an average increase rate of 0.019 °C/yr. The MAGST average rose by about 0.8 °C, with an average rate of 0.027 °C/yr,which was about 42% larger than that in MAATs. The climate warming mainly occurred in the cold season, with an average warming trend of 0.024 °C/yr, while the rate in the warm season was 0.012 °C/yr. The ground surface warming had a very distinct seasonality, with a warming rate of 0.020°C/yr in the cold season and 0.032 °C/yr in the warm season(Table 2).

Table 2 Increases of seasonal and annual averages in air and surface temperatures (°C) in the seasonally frozen ground areas on the margins of the QTP from 1967 to 1997 (revised from Zhao, 2003)

Decadal averages and interdecadal change trends of MAGSTs were computed for five meteorological stations in the permafrost regions on the QTP that had continuous measurements for surface temperatures from 1961-2008(Table 3). The results indicate that, on average, MAGSTs increased by 1.3 °C during the 48 years, with an average rate of increase of 0.034 °C/yr, which was slightly higher than those (0.028 °C/yr) of adjacent regions affected only by seasonally frozen ground (Table 2). Ground surface temperatures increased more rapidly (at an average rate of 0.027 °C/yr) than air temperatures (at an average rate of 0.019 °C/yr). This might be attributed to the relatively higher elevations of permafrost regions, higher insolation,and less and dwarfed vegetation. From Table 3, the MAGSTs at Madoi, Tuotuohe, and Amdo were subzero during the 1960s and 1970s; they gradually changed to above-zero during the 1970s-1980s and have remained positive since. Soon afterwards, the permafrost underlying the three stations thawed. However, the permafrost is colder and thicker in the Wudaoliang area. The changes in heat source intensity and ground surface solar radiation components indicate that, evidently, the ground surfaces have been receiving more heat since 1997 (Li Ret al.,2005).

Table 3 Decadal averages (DA) and increases of MAGSTs (°C) at five meteorological stations in the Plateau permafrost regions from 1961 to 2008

The FI and TI on the ground surface can indicate the extents of heat dissipation and accumulation on the ground surface (at a depth of 5 cm) and their variations are indirect,but reliable, indicators for permafrost changes. It is evident from Table 4 that the FI on the QTP was on the decline during 1961-2005, whereas the TI increased. Ground surface temperatures were increasing both in warm and cold seasons,resulting in regional permafrost decay. In the zones of patchy and sporadic discontinuous permafrost and on the margins of the continuous permafrost zone, such as in Madoi, Tuotuohe,and Amdo, the ratio of TI/FI on ground surface has increased from 1.8 to 2.4, suggesting a heat accumulation. This trend was eventually translated to the change in surface temperatures from subzero to above-zero; rising ground temperatures;and warming, downward thawing, and eventual disappearance of permafrost. In the zone of continuous permafrost, such as Wudaoliang, the decadal average of MAGSTs increased from -1.9 °C in the 1960s to -0.5 °C in 2001-2005. Concurrently, the TI/FI ratio rose from 0.9 to 1.3 (Wu Tet al., 2008).Although permafrost temperatures are still low at present, they have been increasing according to the trend analysis of the heat budgets on the ground surface.

Table 4 Statistics on MAGSTs: DA (°C), FI (°C·d), and TI (°C·d), and the ratio of TI/FI at five stations in the Plateau permafrost regions(Revised from Wu T et al., 2008)

5. Degradation of permafrost

Direct responses of permafrost to climate warming include rising ground temperatures, a deepening active layer,and shrinking areal extent of permafrost. Consequently, declining moisture contents of soils in the active layer will inevitably have some environmental impacts on soils and vegetation, and will lead to a series of changes in permafrost hydrology and ecology (Wanget al., 1996; Jinet al., 2000a,b, 2009; Zhanget al., 2004).

5.1. Changes in ground temperatures

In response to the fluctuations in air and ground surface temperatures, permafrost is subject to periodic changes or variations in ground temperatures, but with a time lag and attenuation with depths until approaching undetectable temperature variations at the depth of annual zero amplitude(less than instrument measurable accuracy, generally within the range of ±0.1 °C), generally at 10-20 m on the QTP.However, variations in permafrost temperatures can also be caused by the superposition of the thermal effects resulting from changes in local surface conditions and physical properties of the overlying soil layers. Ground temperatures are geologically controlled by changes in the geothermal heat flow, but their effects are negligible at a decadal timescale. Changes in ground temperatures can be a combined result from changes of various periodicities in air and surface temperatures, which have different amplitudes and impacting depths. Often, the response processes of earlier temperature changes have not yet completed when the responses of the next begin. The present distributive features and thermal regimes of the Plateau permafrost,generally several tens of meters in thickness, were largely formed under the control and influences of the cold climate during the LGM (35,000-10,800 yrs B.P.), the Neoglaciation(4,000-1,000 yrs B.P.), and the LIA (500-100 yrs B.P.) (Jinet al., 2007a). Permafrost has been on a general degradation since, and its degradation has continued and intensified under a persistently warming climate during the last century(Jinet al., 2006b).

The rates of change in ground temperatures at various periods and depths were computed on the basis of relatively continuous measurements of permafrost temperatures in 60 boreholes on the QTP since 1975 (some boreholes since 1961). Most of the measured depths are greater than 10 m(38 boreholes greater than 20 m, and 3 boreholes down to depths of 60-70 m). (Note: Coordinates of boreholes and important but unfamiliar sites used in this paper are provided in Appendix 2). Analysis of the averaged annual ground temperatures curves indicate that most of the observed curves fall in the type of degrading permafrost (Figures 4, 5,and 6) (Jinet al., 2000a), which is characterized by higher temperatures above the depth of zero annual amplitude in comparison with the depths below. The ground warming will be first detected at the depths of 5-10 m. Close to the depth of zero annual amplitude, the ground temperature curves have started to show the transitory trends (Figure 6).

Figure 4 Curves of mean annual soil temperature (MAST) along the QTH (G109) in 1999.Coordinates for the boreholes can be found in Appendix 2.

Figure 5 Curves of ground temperatures of the degrading permafrost on the QTP. (a) Borehole No. 1 at the undisturbed site near the HMSS 66 on the Chumarhe High Plain; (b) Borehole K2956 at the western side of the Qingshuihe River along the QTH; (c) Borehole 113-5 on the southern piedmont of the Taoerjiu Mountains; (d) Borehole at the Northern Outlet of the Jingxiangu Valley in western Xidatan.Coordinates for the boreholes can be found in Appendix 2.

Figure 6 Curves of ground temperatures of transitory permafrost on the QTP. (a) Boreholes No. 1 and No. 2 at the Liangdaohe; (b) Borehole Xidatan-3; (c) Borehole Xidatan-4; (d) Borehole 103-II-1 in the Wenquan Valley; (e) Boreholes in the Liangdaohe Valley(April 16, 1976). Coordinates for the boreholes can be found in Appendix 2.

Statistics indicate that ground temperatures started to increase in 1985, but the initial rates of increase were small (about 0.01-0.03 °C/yr). Since 1996, the rates of increase have been accelerating (Table 5). They have generally risen to the range of 0.02-0.06 °C/yr, occasionally to 0.07 °C/yr. Accordingly, the recent acceleration of ground warming should be closely related to the rapidly increasing air temperatures. In similar climatic environments, ground temperatures increase with the decreasing MAGTs in segments with cold (＜-1 °C) permafrost, such as at the Kunlun Mountain Pass, on the northern slope of the Hoh Xil Mountains, and at the Fenghuoshan Mountain Pass. The rates of ground warming at depths of 5-10 m are generally 0.03-0.07 °C/yr, which are higher than those in other segments with warm (≥-1 °C) and thin(＜10-20 m) permafrost. The upward thawing from the permafrost base has resulted in higher warming rates in the vicinity of the permafrost base than those in the upper layers of permafrost, such as those observed at Xidatan(Table 5).

Table 5 Increases in ground temperatures at various depths along the QTH (G109) and at Huashixia along the QKH (G214) during 1996-2008

Measurements of the heat fluxes at 5 cm in the permafrost regions along the QTH during 1996-2004 (Table 6)indicate that the heat flux curves show a distinct seasonal pattern. The heat fluxes during March to September are positive (i.e., the ground receives heat from the atmosphere;heat is released from the ground during October to the following February). However, the annual average of heat fluxes is also positive, indicating that, overall, heat has been accumulating. Consequently, ground temperatures are on the rise and permafrost is degrading. From Table 6, it is clear that the difference between air and ground temperatures, and the annual average soil heat fluxes and resultant heat accumulation in the ground, are larger in areas with lower MAGTs (Li Ret al., 2005). Therefore, the areas with colder permafrost have been experiencing more pronounced ground warming,both in the absolute increases and the warming rates.

Table 6 Annual average heat fluxes of soil at 5-cm depth in permafrost regions along the QTH (Li R et al., 2005)

Large amounts of observational data have proved that ground temperatures at shallow (＜5 m) depths in permafrost regions are greatly affected by seasonal variations in air temperatures, with frequent fluctuations and complex patterns of variation. However, those at greater (＞5 m) depths have better correlations with longer-term (≥10 yrs) variations of air temperatures in comparison with interannual variations. Although ground temperatures at depths of 5-10 m change according to variations in MAATs, they also reveal a multiyear trend, particularly decadal fluctuation. With much-reduced amplitudes and increasingly prolonged time lags with depth, ground temperatures at depths greater than 20 m respond to climate changes at timescales longer than 20 years. Accordingly, change patterns of ground temperatures are classified into those at three depths (5-10 m, 10-20 m, and ＞20 m) to facilitate the discussions.

Based on MAGTs (Table 5), the Plateau permafrost can be categorized into three types: cold (MAGT＜-2 °C), transitory (-2 °C≤MAGT≤-1 °C), and warm (MAGT＞-1 °C)permafrost. When comparing and contrasting the response of permafrost to climate change, and the time lag of ground temperature after air and surface temperatures are taken into consideration, the time series of changes in ground temperatures are deliberately delayed for 5 years compared to those of air and ground surface temperatures. As a result, a division of three decades were delineated: 1976-1985,1986-1995, and 1996-2008. Prior to 1975, there were not many borehole measurements of permafrost temperatures,and few changes were observed in those measurements. The thermal types of permafrost, the time periods, the ranges of depths, and the results on air, surface, and ground temperatures, and their change rates, are summarized in Table 7 for comprehensive analysis of spatiotemporal variations in the degradation of permafrost on the Plateau in response to a warming climate.

From Tables 1, 3, and 7 and Figure 3, it is evident that although air and surface temperatures had started to increase during 1976-1985, mean annual soil temperatures (MASTs)had largely remained unchanged, apart from slight fluctuations in a range of -0.001 to 0.015 °C/yr in the changing rates of ground temperatures at depths of 5-10 m in some boreholes in warm and transitory permafrost regions. The ground warming is generally observed in the areas underlain by coarse-grained soils and covered by dry and sparsely vegetated surfaces. The thermal states of the Plateau permafrost were relatively stable during this period.

With continued rising air and surface temperatures during 1986-1995, permafrost was warming and degrading extensively. However, increases in ground temperatures were slow, and MAGTs gradually rose by only about 0.1-0.2 °C, occasionally up to 0.3 °C, with average increase rates of 0.01 to 0.032 °C/yr.

Due to remarkable increases in air and surface temperatures after the 1990s, the downward heat transfer was strengthened and the rates of change in ground temperatures at depths of 5-10 m reached 0.03-0.07 °C/yr during 1996-2008 (Table 7), doubling those during the previous decade (1986-1995). As a result, MAGTs generally increased by 0.2-0.3 °C, and even by 0.4 °C in some areas.This is a period of accelerated regional degradation of permafrost.

It was during this period that vertical detachment of permafrost from seasonal frost action was identified in many areas in the warm permafrost regions. The warming of cold permafrost was accelerated, and its warming rates had evidently exceeded those of warm permafrost. This is attributed to the fact that annual average heat fluxes of soils in cold permafrost regions were greater than those in warm ones(Table 6), and consequently permafrost received more heat being transported downwards. Additionally, when heated,cold permafrost first raises its temperature, but when its temperature reaches and stays at the thawing/freezing point of water, ground ice starts partial melting, which absorbs a large amount of heat for the phase change while the temperature remains unchanged. Therefore, under similar external conditions, the warming rates and amplitudes of cold permafrost are relatively larger.

By the end of the 20th century, the warming of perma-frost had reached depths of about 60 m on the QTP. For example, the ground temperature at the depth of 59.8 m in Borehole CK7 in the Liangdaohe Valley increased by 0.1 °C in 1998 compared to that in 1976. This indicates that the climate warming during the previous few decades had reached a depth of about 60 m, and probably lower.

In Table 7, in the row "Warm permafrost" and columns"1986-1995" and "1996-2008", rates of change in ground temperatures at depths greater than 20 m are larger than those at shallower depths. This phenomenon was generally observed in the margins of permafrost regions where permafrost was generally 15-30 m in thickness (Table 5). At present, the rate of upward degradation of permafrost is faster than that of downward degradation (Jinet al., 2006a). The QTP is relatively young, with active tectonic activities, and the geothermal gradients in the unfrozen ground beneath the base of permafrost (qu) are generally greater than those in permafrost. Based on the statistics on the Plateau geothermal fluxes,geothermal gradients of permafrost (qf) generally range from 0.04 to 0.07 °C/m, although they can be as small as 0.02°C/m, in comparison to those of the unfrozen ground which generally vary from 0.05 to 0.10 °C/m, occasionally being as small as 0.03 °C/m. The ratios of the geothermal gradients of unfrozen to frozen ground (qu/qf) have varied from 1.1 to 2.0. In recent years, with rising air and ground temperatures,the curves of warm permafrost temperatures are approximating the roughly zero geothermal gradient. However, the geothermal gradients of unfrozen ground and the ratios(qu/qf) are enlarged after the warming of unfrozen ground under the permafrost. This has increased the upward heat fluxes to the permafrost layer and thus intensified the warming of permafrost, and subsequently accelerated the upward decay of permafrost. As a result, the rise of permafrost temperatures has been more rapid due to the two- or three-directional warming and degradation of permafrost and the resultant reduced lag in the response time.

Table 7 Rates (°C/yr) of increase in the MAAT, MAGST, and MAST in cold, transitory, and warm permafrost regions on the QTP during the last three decades

5.2. Changes in the positions of the permafrost table

In the natural state and when there are no substantial changes in surface and hydrological conditions, the extensive and persistent lowering of the permafrost table (i.e.,deepening of the active layer thickness) is a reliable evidence for the response of permafrost to a warming climate.The statistical data on the changes in the permafrost table on the QTP since the 1970s (as revealed by the 710 shallow boreholes and hand-dug pits) are summarized in Table 8 according to the regions and time periods. Although spatiotemporal variations in the permafrost table positions are stochastic processes due to their numerous contributing factors, there is a clear and logical deepening trend under the warming climate. General increases in the active layer depth range from 25 to 50 cm, and even to 70-80 cm in some areas underlain by ice-poor warm permafrost. The deepening of the active layer is largely correlated to the rising air temperature, and it also depends on soil types and moisture contents in the active layer. The lower the ice content, the greater the increase in the active layer thickness, particularly for that in ice-poor permafrost areas in comparison with that in permafrost areas with an average ice content (Table 8).

Air temperatures on the QTP were largely fluctuating without much overall trend from 1976 to 1985. This resulted in the permafrost table showing fluctuating changes in the range of -1 to +2 cm/yr (Table 8). However, the active layer was generally deepening at rates of +1 to +3 cm/yr from 1986 to 1995. From 1996 to 2008, the rates of increase in the thickness of the active layer (+2 to +10 cm/yr) were more than doubled in comparison with that of the previous decade. On the margins of the permafrost zones, taliks were expanding and permafrost became detached from seasonal frost action in many places. This evidence also points to the trend of an accelerated regional degradation of permafrost.

Table 8 Comparisons of the permafrost table (cm) under the natural state along the QTH (G109)

5.3. Changes in the areal extent of permafrost

The QTP has the largest and highest expanse of what may scientifically be called "elevational permafrost". During the last 30 years, air and ground temperatures have been increasing persistently and extensively. The thin (several meters thick) permafrost has been thawed and converted to either seasonally frozen ground or talik. The LLP has extensively risen by about 40-80 m (i.e., the LLPs have been retreating towards the center of the mass of elevational permafrost). For example, the southern LLP along the QTH has shifted northwards by about 12 km, while the northern LLP has moved only 3 km southwards. In the vicinity of Madoi on the eastern QTP, the LLP has shrunk westwards by at least 15 km (Jinet al., 2009).

The continued degradation of permafrost has resulted in a gradual shrinkage in the areal extent of permafrost(Table 9). Although various researchers have used different base maps of various scales and methods for estimating the total area of permafrost on the QTP, the change in permafrost area during the last 30 years should be significant. The total area of permafrost decreased from 1.50×106km2in 1975 to 1.26×106km2in 2006, that is, a reduction of about 0.24×106km2(16%) during the last three decades. It is worth noting that shrinkage of about 0.14×106km2of permafrost, roughly 60% of the total reduced area during the 30 years, occurred during the period of 1996-2006. This suggests that the degradation of permafrost has been accelerating. For example, the permafrost accounted for about 20%in the surveyed patchy permafrost region at the section from HMSSs 115 to 117 along the southern part of the QTH in 1975. Repeated surveys concluded that the areal extent of permafrost became 13% in summer of 1996, and has become 5% at present.

Table 9 Changes in the areal extent of permafrost on the QTP

The degradation of permafrost has been one of the major internal driving forces in accelerating eco-systematic successions on the Plateau. The ecological responses to climate warming have intensified as shown by the lowering ground water table, shrinking lakes and wetlands, and degenerating and desertification of rangelands (Xuet al., 2008; Li Let al.,2010). During the last 15 years, as a result of permafrost degradation, the areal extent of alpine paludal meadows declined sharply by 28.11%, and that of alpine meadows by about 8% (Wang Get al., 2006). Declining vegetative coverage alters the terrestrial heat exchange processes through the changed albedo, and will in fact enhance climate warming (Liuet al., 2009), and subsequently further permafrost degradation. As a result, the degradation of permafrost and climate warming may facilitate each other and become self-perpetuating.

5.4. Modes of permafrost degradation

Permafrost describes a specific physical state of the ground (i.e., ground of subzero temperature for no less than two consecutive years), and therefore means a result of climate cooling or a cold climate. Under a warming climate, or when the periglacial environments conducive to the formation and/or preservation of permafrost are disturbed or removed, permafrost will be subject to destabilization or degradation. First, the MAST and MAGT will rise until the permafrost starts to thaw downwards from the permafrost table, upwards from the base of permafrost, laterally due to the horizontal heat fluxes, or their combinations (Jinet al.,2006a; Wu Jet al., 2009). At a certain point, taliks can be created between the degraded permafrost and the seasonally frozen ground because the frost action cannot reach the new permafrost table, or between areas of or inside degrading permafrost.

Regarding spatial distribution, permafrost becomes more discrete in the discontinuous permafrost regions and the boundaries of permafrost shift polewards, that is, northwards or southwards for latitudinal permafrost in the Northern or Southern Hemisphere and upwards for the elevational permafrost. Regional differentiation in the degradation of permafrost can be quite striking, or even controversial. Some scholars suggest that climatic warming can be more substantial in polar or alpine regions, and permafrost there would be degraded more rapidly. The thickness of permafrost can be largely determined by the range and amplitudes of MAATs,MAGSTs, MASTs, and MAGTs. They are subject to the climate changes at various spatial and temporal scales.

However, due to the strong influences of some local factors, such as cold air drainage, rich moisture content of soil surfaces and rich organic or peat soils, blocky stone fields (Gorbunovet al., 2004), and changes in snow cover,some patches of sporadic permafrost can be quite stable or robust due to their unique stabilizing environs, even under a very dramatic climate warming (Harris S, 2005, 2009).Sometimes the favorable microclimate due to the blocky debris or organic soils can result a thermal offset of as much as 2 to 6 °C (Harris S and Pedersen, 1998; Gorbunovet al.,2004), which can result in the occurrence of relict permafrost well below the LLP or much farther south than the SLP of latitudinal permafrost (Nekrasov and Klimovskii, 1978;Jinet al., 2007b).

6. Spatiotemporal variability of permafrost degradation

6.1. Complexity of permafrost degradation on the QTP

In the earth surface systems, the cryosphere is the most sensitive, rapid, and pronounced zone in response to climate changes. In particular, as the major component and with huge areal extents and volumes, permafrost responds and feeds back to climate changes in a complicated way. These interactions are difficult to systematically and economically detect and evaluate, and to forecast, even using sophisticated models. In comparison with the latitudinal permafrost in the Arctic and Subarctic, the elevational permafrost on the QTP is warm in temperature and therefore it may respond more readily to climate warming, and is more profoundly influenced by the shifts and modifications in the Pacific and Indian monsoons. As a result, the degradation of permafrost on the eastern and southern QTP, such as in the Hengduan Mountains and the Sources Areas of the "Three Rivers"(Yangtze, Yellow and Lancang-Mekong) (SATR), and the Himalayas, is remarkable (Fukuiet al., 2007b; Jinet al.,2009). The permafrost is relatively more thermally stable in the continental climate regions under the stronger influences of the Westerlies, such as in the interior and on the western and northern QTP. However, there are only very limited measurements and modeling of these regions; further research is needed for comprehensive assessment and forecast of their trends (Jinet al., 2009; Wu Jet al., 2009).

Although many classifications of frozen ground do not include the precipitation or aridity, they play important roles in the responses of permafrost and its feedbacks to the climate systems, as it is well known that a great amount of heat exchange occurs during raining and snow-melting and the subsequent infiltration and evapotranspiration. However, the observations and analyses on the trends of precipitation on the QTP during the last 50 years present mixed signals. Although a number of Chinese scholars have claimed that the QTP has experienced a general drying trend (e.g., Yang Met al., 2004), some others believe that the plateau is showing evidence of warming and wetting (e.g., Zhanget al., 2009;Li Let al., 2010). Other scholars have concluded that currently there is no persuasive evidence that recent climate change has led to reduced precipitation generally on the QTP (e.g., Harris R, 2010). Therefore, at present it is difficult to evaluate the impacts of changes in precipitation patterns, amounts, and days on the thermal regimes of permafrost and the degradation of permafrost.

Although the existing permafrost on the Plateau is the result of permafrost development under climate fluctuations of many periodicities and amplitudes, the responses of permafrost temperatures to sub-level climate changes were later overwritten by the results of the top-level climatic events(such as the LGM, or the Megathermal). Therefore, in a given period, the degrees of permafrost degradation vary significantly across the QTP. Even at the same location, the rates of permafrost degradation at different depths could also be variable. Since the permafrost is a thermal state of soils,the change rates in ground temperature, that is, the range of amplitudes at the affected depths, which impacted by climate changes of the various periodicities, are apparently the drivers of the amplitudes and modes change rates of the degradation of permafrost.

6.2. Spatial variability

6.2.1 Regional variability

The degradation of permafrost on the interior and eastern QTP, particularly along the QTH and QTR, has been extensively reported. The eastern and southern margins of the Plateau permafrost regions are profoundly influenced by the monsoonal climate and have a generally flatter topography,resulting in more pronounced degradation of permafrost over larger areal extents, increased rates and amplitudes of changes of ground temperatures, as well as greater environmental impacts of the warming and degrading permafrost.

On other edges of the QTP, permafrost degradation has also been observed. In the Khumbu Himal of the Nepal Himalayas, the LLP was estimated to be at 4,900-5,000 m a.s.l.on northern slopes and 5,200-5,300 m a.s.l. on southern slopes in 1973 (Fujii and Higuchi, 1976). Using seismic reflection soundings, it was estimated to be at 5,400-5,500 m a.s.l. on southern slopes in 1991 (Barsch and Jakob, 1993).Thus, it was possible that the LLP rose by 100-300 m between 1973-1991. It was proven by ground temperature measurements that it was 5,400-5,500 m a.s.l. in 2004.Therefore, during the last decade, the LLP seems to have been stable. The rise of 100-300 m exceeds that of 40-80 m on the QTP under a warming rate of 0.2-0.4 °C from the 1970s to the 1990s (Jinet al., 2000a; Fukuiet al., 2007b).

In the eastern Qilian Mountains on the northern edge of the QTP, the LLP has risen by about 80-100 m during the last 50 years while the MAGTs have warmed by about 0.3 to 1.0 °C, and the thicknesses of the active layer have deepened by about 0.3 to 0.9 m (Wu Jet al., 2007a, b). In the western QTP, some short-term investigations have revealed the signs of permafrost degradation, such as along the XTH, but these observations need further monitoring and mapping data to validate.

6.2.2 Changes of ground temperatures at depths

Our data indicate that shallow (＜10 m) permafrost experienced degradation first, which was characterized by the short time lags behind the changes in air temperatures, the rapid response, and large amplitudes and increased rates of change in ground temperatures. Due to the exponential attenuation of the heat transfer processes with depth, the amplitudes and increase in rates of rise in ground temperature decrease with elongated lags. Due to the presence of the"zero curtain effect", permafrost needs much longer time than the unfrozen soil to change its temperatures. The warming rates of the Plateau permafrost, the affected depths, and time lags in areas of changing air temperatures have been greater than those found in the latitudinal permafrost under similar climate warming conditions. This may be attributed to the relative larger geothermal gradients on the QTP.

Based on the statistics of ground temperature measurements in boreholes during the years since the onset of the 21st century (Table 7 and Figures 4, 5, and 6), the degradation of permafrost at shallow depths (＜20 m) has been accelerating, and the MAGTs have extensively increased by 0.2 to 0.4 °C. The ground temperatures at depths of 20-50 m have increased by 0.1 to 0.2 °C. In some areas, ground temperatures at depths of 50 to 70 m also have been on the rise.This may suggest that the effects of climate warming on the Plateau from 1961 to 2008 have reached depths of 50 to 70 m in most places. The vertical distribution of the time lags of the response of permafrost (i.e., ground temperatures) to the observed warming of air temperatures are as follows: 2 to 5 years at depths of 5-10 m, 4 to 10 years at depths of 10-20 m, 8 to 25 years at depths of 20-30 m, 20 to 35 years at depths of 30-50 m, and 30 to 45 years at depths of 50-70 m(Li Set al., 1996).

6.3. Temporal variability

6.3.1 Permafrost evolution during the late Quaternary

The main body of the existing permafrost on the QTP was formed by permafrost development during the LGM(about 35,000 to 10,800 yrs B.P.), followed by the relatively cold period subject to frequent and abrupt climate changes during the early Holocene (10,800 to about 8,500 yrs B.P.),the Neoglaciation in the middle Holocene (3,000 to 1,000 yrs B.P.), and the more recent Little Ice Age (LIA, 500 to 100 yrs B.P.). During the warm periods in the Holocene,such as the Megathermal (8,500 to about 3,000 yrs B.P.) and the warming during the later Holocene (1,000 to 500 yrs B.P.), permafrost degraded accordingly, and extensive taliks and the deeply buried permafrost (permafrost table ＞10 m or even to 25 m) were formed on the eastern and southern margins of the QTP (Jinet al., 2007a). In contrast to the conditions during the LIA, the permafrost has been gradually degrading during the last 100 years, but there are significant spatiotemporal variations across the Plateau.

6.3.2 Recent permafrost degradation

During the last 50 years (1961-2010), the MAAT of the permafrost regions on the QTP increased by 1.3 °C, with an average warming rate of 0.03 °C/yr. Recent publications suggest that climate change varied across the Plateau, with far greater variability on the eastern QTP (Liuet al., 2006;Li Let al., 2010). At the same time, the MAGST rose by 1.3 °C, with an average warming rate of about 0.03 °C/yr.These rates greatly exceed their respective global averages.The most intensive and persistent climate warming during the last 100 years should be the primary cause for the degradation of permafrost on the QTP. In particular, since the late 1990s, the air and ground surface temperatures have been increasing at much greater rates, which have further accelerated the degradation of permafrost.

Due to the climate warming since the end of the 1980s,the Plateau permafrost started to degrade first in the warm(＞-1 °C) permafrost regions, and gradually spread to the regions of transitory (-1 to -2 °C) and cold (＜-2 °C) permafrost. By the end of the 1990s, the permafrost had reached a stage of regional degradation. Permafrost of various types showed different modes of degradation. In the warm permafrost regions, this is displayed as a reduction in the areal extent of permafrost and shrinkage of isolated patches of permafrost, that is, the horizontal distribution of permafrost becomes more discrete and, vertically, permafrost either thins or simply disappears. This is particularly evident in the areas of patchy permafrost. The major evidence of permafrost degradation in the regions of transitory and cold permafrost is the rising ground temperature.

6.3.3 Future permafrost degradation

According to the Fourth Assessment Report by the Intergovernmental Panel on Climatic Change (IPCC) (Shen,2007), the present global climate is undergoing a profound change characteristic of significant warming. The warming during 11 of the 12 years from 1995 to 2006 was among the greatest since 1850. The average linear climate warming rate of 0.13 °C/yr during the past 50 years was almost double the average of that during the last 100 years (Shen, 2007).

Responses of permafrost temperatures to changes in air temperatures are multifaceted, with time lags closely related to rates of change in air temperatures, lithology and ground ice contents of soils, and vegetation. On the basis of spatial and temporal analysis, permafrost started to warm first at depths shallower than 10 m. The extensive degradation of permafrost then reached the depths of 10 to 20 m (i.e., the range of the zero annual amplitude) by the end of the 1980s. In particular,the warming rates at all depths almost doubled from 1996 to 2008 in comparison to the preceding decade (Table 7). The abovementioned spatiotemporal variability in the degradation of permafrost may reflect consequences of the persistent climate warming since the mid-1980s and the significant rise in air temperatures since the late 1990s on the QTP.

Many scholars have conducted numerical modeling and prediction of future changes of the permafrost on the QTP.All these predictions invariably assumed that the warming trend currently observed would continue uninterrupted for at least another century. The numerical simulations and modeling provided some mixed results on the trends of permafrost degradation under the projected climate warming.

(1) Numerical simulations of Li Set al. (1996) suggested that after 50 years the changes in the areal extent of permafrost would be insignificant, assuming a constant climate warming of about 0.04 °C/yr, although rising ground temperatures, deepening active layer, slight thinning in permafrost, and detachment from the seasonal frost action may happen.

(2) Using altitudinal and Gaussian models, Li and Cheng(1999) predicted that permafrost would experience insignificant areal change (less than 19%) from 1999 to 2019, but by the year 2099 the areal extent reduction of permafrost would exceed 58%, assuming an increase of 2.91 °C in MAATs. Under this scenario, almost all the permafrost on the southern and eastern Plateau may disappear.

(3) According to Wu Qet al. (2000, 2001), during the next 50 years, regardless of whether the air temperature keeps rising or is unchanged, the degradation rates of all types of permafrost at various depths would still accelerate,resulting in significant shifts in the spatial distribution of permafrost. Along the QTH, they predicted that by 2099: (a)The areal extents of very stable permafrost (MAGTs＜-5 °C)would shrink from 5.6% to about 0.6%; stable permafrost(-3 to -5 °C in MAGTs) from 16.3% to 3.3%; and metastable permafrost (-1.5 to -3 °C in MAGTs) from 25.5% to 17.4%. (b) The transition zone (-0.5 to -1.5°C in MAGTs)would enlarge from 22.8% to 31.0% and the unstable permafrost zone (MAGTs＞-0.5 °C) from 10.8% to 27.5%; and the southern and northern LLPs, and other boundaries will shrink toward the interior of the Plateau by about 5 to 10 km during the period of 2009-2049. (c) The continuous permafrost would become discontinuous and would be confined to very high elevations.

(4) Using a numerical model, Nanet al. (2002, 2003,2005) predicted the changes of permafrost during the next 50-100 years under two climate warming scenarios of 0.02°C/yr (low value by IPCC-AR3,i.e., 2.5 °C by 2100), and 0.052 °C/yr (more dramatic change predicted by Qinet al.(2002), 2.2-2.6 °C for the next 50 years). They predicted that in the less dramatic warming scenario about 9% of permafrost,mainly extremely warm (＞-0.11 °C in MAGTs) may be converted to seasonally frozen ground during the next 50 years.Very warm (＞-0.5 °C) permafrost, about 13.4% of total permafrost area, may disappear by 2100. However, in the dramatic warming scenario, this case would be realized in about 50 years. More remarkable degradation of permafrost,about 46% in areal reduction, would occur in the next 100 years, and all permafrost warmer than -2 °C would be lost.

(5) Wu Tet al. (2005) used a 50-MHz GPR for detecting the boundary of permafrost in the Xidatan region, in the vicinity of the northern LLP on the Plateau. A GIS-aided map of the permafrost distribution on the basis of this survey indicated that the permafrost in Xidatan was reduced by 12% in area and the northern LLP was elevated by 25 m compared with that in 1975.

(6) Using numerical models on the basis of MAAT,DEM, and the downscaled projected climate warming scenario (A1B), Cheng Z and Liu (2008) simulated the distri-bution of permafrost on the QTP for three periods of 1980-1999, 2030-2049, and 2080-2099. In this model,MAGST no warmer than -1 °C was used as the criterion for delineating the boundary of permafrost, while it was +0.5 °C in Cheng G (1984) and -0.5 °C in Nanet al. (2002). The simulation for 1980-1999 indicates a total permafrost area of about 1.28 million km2, which agrees well with the mapped distribution of permafrost. It is projected that by 2030-2049, the areal extent of the Plateau permafrost would be 0.87 million km2(a reduction of about 32%), and by 2080-2099, it would be 0.69 million km2(46% reduction),and the rate of permafrost degradation would be slow with rising elevation.

(7) The Huashixia Permafrost Station (35.06534°N,98.69882°E; 4,100 to 4,300 m a.s.l.) on the northeastern QTP is in the Huashixia Valley, where the MAAT is about-4.5 °C, the MAGST is -2 °C, and the thermal gradient is 18 °C/km. Continuous permafrost with MAGTs at -0.9 to-1.2 °C is present and the permafrost table is at 1.2 m in depth in a typical alpine meadow. The strata include 3.3 m of organic- and ice-rich silt and clay on the top, then 4.3 m of gravel, underlain by bedrocks. Assuming a climate warming of 0.04 °C/yr, the prediction for this specific site by Li Det al. (2008) suggests the permafrost of 55 m in thickness on average could be quite stable. The permafrost table would be lowered slightly from 1.2 to 1.5 m during the next 110 years but a supra-permafrost talik would not result, although the MAGT would rise from an initial -1.4 to-0.3 °C and the base of permafrost would rise from an initial 55 m to 15 m.

7. Conclusions and prospects

7.1. Conclusions

On the basis of these analyses, the following conclusions can be reached:

(1) 1961 to 2010 was a period of climate fluctuations,with the longest and greatest warming occurring when the average of MAATs in the permafrost regions on the QTP rose by 1.1 °C, with a warming rate of 0.03 °C/yr. The persistent increases in MAATs began in the mid-1980s, in which the decadal increase rate was 0.03 °C/yr in the 1980s,0.07 °C/yr in the 1990s, and 0.06 °C/yr in the 2000s. During the same period, the average MAGSTs rose by 1.3 °C, with an average increase rate of 0.03 °C/yr, in which the decadal increase rate was 0.03 °C/yr in the 1980s, 0.04 °C/yr in the 1990s, and 0.05 °C/yr during 2001-2008. The average increase rates in the ground temperatures of various permafrost types ranged from -0.01 to 0.015 °C/yr in 1976-1985,from 0.01 to 0.032 °C/yr in 1986-1995, and from 0.01 to 0.07 °C/yr in 1996-2008.

(2) Since 1975, the permafrost table has been lowered by 25 to 50 cm, and in some places as much as 80 cm. The average increase rates in the active layer thickness ranged from-1 to +2 cm/yr for the period of 1976-1985, from +1 to +3 cm/yr for 1986-1995, and +2 to +10 cm/yr for 1996- 2008.The total areal extent of permafrost on the QTP has shrunk from 1.50×106km2in 1975 to 1.26×106km2in 2006 (a reduction of about 16%), in which about 0.14×106km2, or 60% of total area of the degraded permafrost, occurred during 1996-2006.

(3) Spatially and temporally, the degradation of permafrost first started in the warm (＞-1 °C) permafrost regions and gradually expanded to the regions of transitory (-1 to -2°C) and cold (＜-2 °C) permafrost. Ground temperatures started to increase at shallow depths, with attenuated amplitudes and lagged time phases, and have now reached depths of 50 to 70 m. The persistent climate warming has resulted in the transition of the thermally stable permafrost to degrading permafrost. In comparison with that on the northern and western parts of the QTP, the permafrost degradation on the eastern and southern parts and in the warm permafrost regions has been accelerated during the last decade. Regionally, the permafrost has been degrading more remarkably and more rapidly, and its ecological and hydrological impacts are much more severe on the eastern and southern parts due to the strong influences of the monsoonal climate.Permafrost on the northern and western Plateau might be more thermally stable, but it is too early to draw reliable conclusions because of the paucity of long-term monitoring data on permafrost temperatures.

(4) The response processes, mechanisms, and the spatiotemporal variability of the Plateau permafrost to climate change are crucial in understanding the interactions between the cryosphere and the atmosphere, in improving the global circulation and permafrost forecast models, in deepening the understanding of the heat and moisture transfer processes, in better predicting and adapting to the cold regions ecological and hydrological changes, and in analyzing and assessing the design principles and long-term stability of engineered infrastructures in the permafrost regions. In the long run,further research on the degradation needs to monitor the heat fluxes at three dimensions in many important sites, and in particular, at the positions of the permafrost table and base and inside the permafrost layer(s). The changes in heat fluxes can better reflect the long-term trend of thermal regimes of frozen ground.

7.2. Unresolved issues

Although there have been many studies, the uncertainties in climate and environmental changes, including paleo-reconstruction, the evaluation of present status, and for the prediction of the future, remain great. This makes a more accurate forecast of the spatiotemporal response of permafrost to climate change on the QTP more difficult. The prevailing projections for climate change are for climate warming, and those for precipitation and other climate variables are very inadequate. It is difficult to form a general conclusion on climate change. Additionally, human impacts on the Plateau environments have been increasing due to the re-sources development and the increasing technological capabilities. The cold regions environments, represented by the alpine meadows, steppes, and others, will experience more anthropogenic disturbances such as overgrazing, cultivation,and engineering activities.

The understanding of and observational data on the mechanisms of permafrost response to climate and environmental change are still very limited, and these observations are limited too along major engineering corridors, such as the QTH and the QKH. In areas with more anthropogenic influences, observations on the permafrost environment and its change may be skewed in comparison with those in the natural state, and the actual changes on the Plateau may not be as dramatic as those along the engineering corridors.

Although there are some modeling and numerical simulations on the permafrost evolution, distribution of permafrost, and future changes, and some of them incorporate GIS and remote sensing data, they are very rudimentary because of the paucity of data on the physical processes, oversimplification, and difficulties in model parameterization. Therefore, the accuracy of these projections or predictions is compromised and would need more research to improve the models. Since the mechanisms of the interactions between the climate change and permafrost degradation are largely unclear, and their predictions are inaccurate, the environmental impacts and the adaptive measures to the changing permafrost environments are more difficult to understand and evaluate. So far, many analyses and speculations on the relationships among the changing climate, degradation of permafrost, and degradation of alpine ecosystems are either misunderstood or overestimated. It is possible that even the so-called degradation of the rangelands on the QTP should be more closely and systematically scrutinized. The most urgent issues now are to establish more systematic and well coordinated monitoring systems on the QTP, to improve communications and sharing of observational data, and to more reasonably evaluate the present status and future trends.

Both the stability and degradation of permafrost and alpine ecosystems have been extensively reported, but many of these studies have not provided adequate or convincing evidence and/or logical analysis, therefore no reliable conclusions can be drawn thereupon. Many ecological rehabilitation programs, either based on these reports or on preliminary studies, and generally hastily implemented,would inevitably have only mixed, or at best, limited (if not adverse) effects on alleviating the so-called deteriorating environments, which also needs adequate scientific evaluation for itself.

7.3. Prospects

Since the spatial and temporal variability of permafrost degradation and its environmental impacts have great implications to regional sustainable development, more investment and efforts are deemed necessary for understanding the amplitudes, rates, and mechanisms of the past, present, and future changes of the cold regions environment and for adaptation of human-earth systems. A multidisciplinary approach is indispensable to the improvement of the relevant research. On the present thermal states of permafrost, more cutting-edge technologies, such as remote sensing and GIS,permafrost surveys, mapping and monitoring at fixed sites,transects, and periodic investigations for comparisons using environmental geophysical methods, such as the multi-channel ground penetrating radar (GPR), should be integrated with other ongoing study networks in climate,hydrology, and ecology. These studies should also be merged with international and regional research and monitoring projects to provide global and regional perspectives,and be more closely collaborated with governmental efforts on environmental remediation and management/protection programs.

The ongoing permafrost ecological transect of about 1,200×600 km2spreading from Nagqü south of the Tanggula Mountains to the Qilian Mountains may play a critical role in providing reliable and badly needed observational data for the trends of changes in elevational permafrost, hydrology, soil, and ecology on the "roof of the world", but only if the progress and evaluation results are well and timely communicated with and adequately utilized by policy makers. The foci of long-term programs should include ecology, hydrogeology, and geocryology in the Sources Areas of the "Three Rivers" (Yangtze, Yellow, and Lancang-Mekong) and along the QTH, QKH, and XTH, and in the Qilian Mountains. However, this will require decades of well-designed and painstakingly implemented programs for a successful scientific evaluation.

The research was supported by the China Key Research Project for Global Change (No. 2010CB951404) "Changes in the Cryosphere in the Northern Hemisphere and their Impacts on Climatic Environments, and their Adaptation",and the China National Science Foundation (No.40821001) "Frozen Ground and Cold Regions Engineering". Professor Stuart A. Harris gave many insightful and constructive comments on the manuscript. Mr. Geoffrey Gay and Mr. Anda Divine provided generous assistance in English editing. Two unidentified reviewers provided many comments for improving the quality of the paper.Their help is greatly appreciated and thus acknowledged herewith.

Allard M, Wang BL, Pilon JA, 1995. Recent cooling along the southern shore of Hudson Strait, Quebec, Canada, documented from permafrost temperature measurements. Arctic and Alpine Research, 27: 157-166.

Barsch D, Jakob M, 1993. Active rock glaciers and the lower limit of discontinuous alpine permafrost in the Khumbu Himalaya, Nepal. In: Proceedings of the Six International Conference on Permafrost. South China University of Technology Press, Wuhan, China. 27-31.

Beltrami H, Taylor AE, 1994. Records of climatic change in the Canadian Arctic: Combination of geothermal and oxygen data yields high resolution ground temperature histories. EOS, Transactions, AGU, 75(44): 75.

Burn CR, 1998. Field investigations of permafrost and climatic change in northwest North America. In: Proceedings, 7th International Conference on Permafrost, Yellowknife, Canada, June 1998, Université Laval Collection Nordicana, 57: 107-120.

Cheng GD, 1984. Discussion on the zonation of elevational permafrost in China. Acta Geographica Sinica, 39(2): 185-193.

Cheng GD, Zhao L, 2000. Permafrost problems in the development of the Qinghai-Tibet Plateau. Quaternary Science, 20(6): 521-531.

Cheng GD, Wu TH, 2007. Responses of permafrost to climate change and their environmental significance, Qinghai-Tibet Plateau. Journal of Geophysical Research, 112: F02S03. DOI: 10.1029/2006 JF000631.

Cheng ZG, Liu XD, 2008. A preliminary study of future permafrost distribution on the Qinghai-Tibet Plateau under global warming conditions.Areal Research and Development, 27(6): 80-85.

Fang JH, Yong GWC, Li DQ, Zhang K, Xu AH, Li HQ, 2009. Disease investigation and analysis on cement pavement and asphalt pavement of Qing-Kang Highway. In: Huo M, Ma W and Niu FJ (eds.). Recent Development of Research on Permafrost Engineering and Cold Regions Environment. Lanzhou University Press, Lanzhou, Gansu, China. 280-285.

Fujii Y, Higuchi K, 1976. Ground temperature and its relation to permafrost occurrences in the Khumbu Himal and Hidden Valley. Seppyo, 38:125-128.

Fukui K, Fujii Y, Mikhailov N, Ostanin O, Iwahana G, 2007a. The lower limit of mountain permafrost in the Russian Altai Mountains. Permafrost and Periglacial Processes, 18(2): 129-136. DOI: 10.1002/ppp.585.

Fukui K, Fujii Y, Ageta Y, Asashi K, 2007b. Changes in the lower limit of permafrost between 1973 and 2004 in the Khumbu Himal, the Nepal Himalayas. Global and Planetary Change, 55: 251-256. DOI:10.1016/j.glopacha.2006.06.002.

Gorbunov AP, Marchenko SS, Seversky EV, 2004. The thermal environment of blocky materials in the mountains of Central Asia. Permafrost and Periglacial Processes, 15(1): 95-98. DOI: 10.1002/ppp.478.

Halsey LA, Vitt DH, Zoltai SC, 1995. Disequilibrium response of permafrost in boreal continental western Canada to climate change. Climate Change,30: 57-73.

Harris C, Mühll DV, 2003. Permafrost and climate in Europe. In: Visconti Get al. (eds.), Global Change and Protected Areas: Climate Change,Mountain Permafrost Degradation and Geotechnical Hazard. Kluwer Academic Publishers, Amsterdam. 71-82.

Harris C, Arenson LU, Christiansen HH, Etzelmüller B, Frauenfelder R,Gruber S, Haeberli W, Hauck C, H?lzle M, Humlum O, Isaksen K, K??b A, Kern-Lütschg MA, Lehning M, Matsuoka N, Murton JB, N?tzli J,Phillips M, Ross N, Sepp?l? M, Springman SM, Mühll DV, 2009. Permafrost and climate in Europe: Monitoring and modelling thermal, geomorphological and geotechnical responses. Earth-Science Review. DOI:10.1016/j.earscirev.2008.12.002.

Harris RB, 2010. Rangeland degradation on the Qinghai-Tibet Plateau: A review of the evidence of its magnitude and causes. Journal of Arid Environments, 74: 1-12. DOI: 10.1016/jaridenv.2009.06.014.

Harris SA, 1981. Climatic relationships of permafrost zones in areas of low winter snow-cover. Arctic, 34: 64-70.

Harris SA, 1986. Permafrost distribution, zonation and stability along the eastern ranges of the Cordilera of North America. Arctic, 39(1): 29-38.

Harris SA, Pedersen DE, 1998. Thermal regimes beneath coarse blocky material. Permafrost Periglacial Processes, 9: 107-120. DOI: 10.1002/(SICI)1099-1530(199804/06)9:2＜107:AID-PPP277＞ 3.0.CO;2-G.

Harris SA, 2005. Thermal history of the Arctic Ocean environs adjacent to North America during the last 3.5 Ma and a possible mechanism for the cause of the cold events (major glaciations and permafrost events). Progress in Physical Geography, 29(2): 218-237.

Harris SA, 2009. Climate change and permafrost stability in the eastern Canada Cordillera: The results of 33 years of measurements. Science in Cold and Arid Regions, 1(5): 1-15.

Jin HJ, Li SX, Wang SL, Zhao L, 2000a. Impacts of climate change on the permafrost and cold regions environments in China. Acta Geographica Sinica, 55(2): 161-172.

Jin HJ, Li SX, Cheng GD, Wang SL, Li X, 2000b. Permafrost and climatic change in China. Global and Planetary Change, 26: 187-404.

Jin HJ, Zhao L, Wang SL, Guo DX, 2006a. Evolution of permafrost and changes in cold regions environments on eastern and interior Qinghai-Tibet Plateau. Quaternary Science, 26(2): 198-210.

Jin HJ, Zhao L, Wang SL, Jin R, 2006b. Thermal regimes and degradation modes of permafrost along the Qinghai-Tibet Highway. Science in China,49(D11): 1170-1183.

Jin HJ, Chang XL, Wang SL, 2007a. Evolution of permafrost on the Qinghai-Tibet Plateau since the end of the Pleistocene. Journal of Geophysical Research, 112: F02S09. DOI: 10.1029/2006JF000521.

Jin HJ, Yu QH, Lü LZ, Guo DX, Li YW, 2007b. Degradation of permafrost in the Xing’anling Mountains, Northeastern China. Permafrost and Periglacial Processes, 18(2): 245-258. DOI: 10.1002/ppp.589.

Jin HJ, Yu QH, Wang SL, Lü LZ, 2008a. Changes in permafrost environments along the Qinghai-Tibet engineering corridor induced by anthropogenic activities and climate warming. Cold Regions Science and Technology, 53(3): 317-333. DOI: 10.1016/j.coldregions.2007.07.005.

Jin HJ, Sun LP, Wang SL, He RX, Lü LZ, Yu SP, 2008b. Dual impacts of local factors on ground temperatures on the Qinghai-Tibet Plateau (I):Snow cover and vegetation. Journal of Glaciology and Geocryology,30(4): 535-545.

Jin HJ, He RX, Cheng GD, Wu QB, Wang SL, Lü LZ, Chang XL, 2009.Change in frozen ground and eco-environmental impacts in the Source Area of the Yellow River (SAYR) on the northeastern Qinghai-Tibet Plateau, China. Environmental Research Letters, 4: 045206. DOI:10.1088/1748-9326/4/045206.

Jorgenson MT, Racine CH, Walters JC, Osterkamp TE, 2001. Permafrost degradation and ecological changes associated with a warming climate in central Alaska. Climatic Change, 48: 551-579.

Kynicky J, Brtnicky M, Vav?í?ek D, Barto?ová R, Majigsuren U, 2009.Permafrost and climatic change in Mongolia. Online at:http://www.ta3.sk/bioklima2009/Papers/Kynicky%20et%20al.pdf.

Lachenbruch AH, Marshall BV, 1986. Changing climate: Geothermal evidence from permafrost in the Alaskan Arctic. Science, 234: 689-696.

Li DQ, Chen J, Meng QZ, Liu DK, Fang JH, Liu JK, 2008. Numeric simulation of permafrost degradation on the eastern Tibetan Plateau. Permafrost and Periglacial Process, 19: 93-99. DOI: 10.1002/ppp.611.

Li L, Yang S, Wang ZY, Zhu XD, Tang HY, 2010. Evidence of warming and wetting climate over the Qinghai-Tibet Plateau. Arctic, Antarctic, and Alpine Research, 42(4): 449-457. DOI: 10.1657/1938-4246-42.4.449.

Li R, Ji GL, Li SX, Yang W, Zhao JQ, Zhou XP, Lü LZ, 2005. Discussions on the thermal regimes of soils in the Wudaoliang area in the interior of the Qinghai-Tibet Plateau. Acta Energiae Solaris Sinica, 26(3): 299-303.

Li SD, 1996. Map of Permafrost on the Qinghai-Tibet Plateau (1:3,000,000)and its Annotations (annotations in Chinese). Gansu Culture Press, Lanzhou, China.

Li SD, He YX, Wang YQ, Wang SL, 1998. Permafrost and periglacial environments in the Karakorum and Kunlun Mountains. In: Su Z, Xie Z,Wang Z (eds.). Glaciers and Environment in the Karakorum and Kunlun Mountains. Science Press, Beijing, China. 181-215.

Li SX, Cheng GD, Guo DX, 1996. The future thermal regime of numerical simulating permafrost on the Qinghai-Xizang (Tibet) Plateau, China,under a warming climate. Science in China, 39(D4): 434-441.

Li X, Cheng GD, 1999. Response model of high-altitude permafrost to global change. Science in China, 29(D2): 185-192.

Liu XD, Yin ZY, Shao XM, Qin NS, 2006. Temporal trends and variability of daily maximum and minimum, extreme temperature events, and growing season length over the eastern and central Tibetan Plateau during 1961-2003. Journal of Geophysical Research, 11: D19109. DOI:10.1029/2005JD006915.

Liu YP, Lü SH, Li SS, Gao YH, 2009. Numerical simulations on the impacts of the grassland degradation in the sources area of the Yellow River on local climate environments. Plateau Meteorology, 28(2): 327-334.

Lü LZ, Jin HJ, Wang SL, Xue X, He RX, Yu SP, 2008. Dual impacts of local factors on ground temperatures on the Qinghai-Tibet Plateau. (I): Surface soil moisture and sand cover. Journal of Glaciology and Geocryology,30(4): 546-555.

Marchenko SS, Ishikawa M, Sharkhuu N, Jin HJ, Li X, Zhao L, Yabuki H,Brown J, 2006. Distribution and monitoring of permafrost in central and eastern Asia. Online at: http://polaris.nipr.ac.jp/～ipy/usr/sympo/procfiles/57-Marchenko_and_Ishikawa.pdf.

Marchenko SS, Gorbunov AP, Romanovsky VE, 2007. Permafrost warming in the Tien Shan Mountains, central Asia: Northern Eurasia regional climate and environmental change. Global and Planetary Change, 56(3-4):311-327. DOI: 10.1016/j.gloplacha.2006.07.023.

Nan ZT, Li SX, Liu YZ, 2002. Mapping of permafrost on the Qinghai-Tibet Plateau on the basis of mean annual ground temperatures and its application. Journal of Glaciology and Geocryology, 29(2): 142-148.

Nan ZT, Gao ZH, Li SX, Wu TH, 2003. Permafrost changes in the northern lower limit of permafrost on the Qinghai-Tibet Plateau during the last 30 years. Acta Geographica Sinica, 58(6): 817-823.

Nan ZT, Li SX, Cheng GD, 2005. Prediction of permafrost distribution on the Qinghai-Tibet Plateau in the next 50 and 100 years. Science in China,48(D6): 797-804.

Nekrasov IA, Klimovskii IV, 1978. Permafrost along the Baikal-Amur Main Railway in Russian. Siberia Branch of the Science Press, Yakutsk, Yakutia, Russia.

Nelson FE, 2003. (Un)frozen in time. Science, 299(5613): 1673-1675. DOI:10.1126/science.1081111.

Nelson FE, Anisimov OA, 1993. Permafrost zonation in Russia under anthropogenic climatic change. Permafrost and Periglacial Processes, 4(2):137-148.

Nelson FE, Anisimov OA, Shiklomanov NI, 2002. Climate change and hazard zonation in the circum-arctic permafrost regions. Natural Hazards,26: 203-225.

Nelson FE, Outcalt SI, 1987. A frost index number for spatial prediction of ground frost zones. Arctic and Alpine Research, 19: 279-288.

Osterkamp TE, Romanovsky VE, 1999. Evidence for warming and thawing of discontinuous permafrost in Alaska. Permafrost and Periglacial Processes, 10(1): 17-37.

Pavlov AV, 1994. Current changes of climate and permafrost in the Arctic and sub-arctic of Russia. Permafrost and Periglacial Processes, 5: 101-110.

PRC Ministry of Construction, 2001. PRC National Code for Engineering Geological Investigation of Frozen Ground. China Planning Press, Beijing, China,.

Qin DH, Ding YH, Wang SW, Wang SM, Dong GR, Lin ED, Liu CZ, She ZX, Sun HN, Wang SR, Wu GH, 2002. A study of environmental change and its impacts in western China. Earth Science Frontiers, 9(2): 321-328.

Qiu GQ, Zhou YW, Guo DX, Wang YX, 2002. Maps of Geocryological Regions and Classifications. Digitized by Zhang T. Boulder, CO, USA,National Snow and Ice Data Center/World Data Center for Glaciology.Digital media.

Romanovskii NN, Zaitsev VN, Volchenkov SY, Zagryazkin DD, Sergeyev DO, 1991. Alpine permafrost temperature zonality, northern trans-Baikal region, USSR. Permafrost and Periglacial Processes, 2(3): 187-195. DOI:10.1002/ppp.3430020303.

Romanovsky VE, Burgess M, Smith S, Yoshikawa K, Brown J, 2002. Permafrost temperature records: Indicators of climate change. EOS, AGU Transactions, 83(50): 589-594.

Shen YP, 2007. The Fourth Assessment Report of the IPCC WGI: Major scientific points on global climate change. Journal of Glaciology and Geocryology, 29(1): 156.

Thibault S, Payette S, 2009. Recent permafrost degradation in bogs of the James Bay area, northern Quebec, Canada. Permafrost and Periglacial Processes, 20(4): 383-389. DOI: 10.1002/ppp.660.

US Arctic Research Commission Permafrost Task Force, 2003. Climate Change, Permafrost, and Impacts on Civil Infrastructure, Special Report,(01-03): 1-72. US Arctic Research Commission, Arlington, Virginia,USA. Online at: http://www.arctic.gov/publications/permafrost.pdf.

Wang GX, Li YS, Wu QB, Wang Y, 2006. Relationships of permafrost and vegetation on the Qinghai-Tibet Plateau and its impacts on the alpine ecosystems. Science in China, 36(D8): 743-754.

Wang SL, 1993. Changes of permafrost along the Qinghai-Tibet Highway during the last few decades. Arid Regions Geography, 16(1): 1-8.

Wang SL, 1997. Degradation of permafrost on the Qinghai-Tibet Plateau.Advances in Earth Sciences, 12(2): 164-167.

Wang SL, Jin HJ, 2000. Permafrost degradation on the Qinghai-Tibet Plateau and its environmental impacts. Permafrost and Periglacial Processes,11(1): 43-53.

Wang SL, Huang YZ, Guo DX, Huang YZ, 1996. Responses of permafrost on the Qinghai-Tibet Plateau to climate change. Journal of Glaciology and Geocryology, 18(suppl.): 157-165.

Wang SL, Lin Q, Zhao L, 1999. Permafrost along the Qing-Kang Highway(National Highway No.214). Arid Land Geography, 22(2): 42-49.

Wang SL, Li SX, Zhao L, 2003. Thermal stability of roadbed in permafrost region along Qinghai-Tibet Highway. Cold Regions Science and Technology, 37: 25-34.

Wang T (ed.), 2006. Map of Glaciers, Permafrost and Deserts in China(1:4,000,000), and Annotations. Science Press, Beijing, China.

Wu JC, Sheng Y, Yu H, Li JP, 2007a. Permafrost in the middle and eastern Qilian Mountains, (I): Distribution of permafrost. Journal of Glaciology and Geocryology, 29(3): 418-425.

Wu JC, Sheng Y, Yu H, Li JP, 2007b. Permafrost in the middle and eastern Qilian Mountains, (II): Characters of permafrost. Journal of Glaciology and Geocryology, 29(3): 426-432.

Wu JC, Sheng Y, Wu QB, Wen Z, 2009. Degradation processes and modes of permafrost on the Qinghai-Tibet Plateau. Science in China, 39(D11):1-9.

Wu QB, Li X, Li WJ, 2000. Prediction of permafrost changes along the Qinghai-Tibet Highway, China. Permafrost and Periglacial Processes, 11:371-376.

Wu QB, Li X, Li WJ, 2001. Response of permafrost along the Qinghai-Tibet Highway to climate change and its impacts on engineering infrastructures. Journal of Glaciology and Geocryology, 23(1): 1-6.

Wu QB, Zhu YL, Liu YZ, 2002. Evaluation model of permafrost thermal stability and thawing sensibility under engineering activity. Cold Regions Science and Technology, 34(1): 19-30.

Wu QB, Dong SF, Liu YZ, 2005. Analysis on the responses of permafrost along the Qinghai-Tibet Highway to climate change and the impacts of permafrost on engineering infrastructures. Journal of Glaciology and Geocryology, 27(1): 150-154.

Wu TH, 2005. Response of Permafrost on the Qinghai-Tibet Plateau to Global Climate Change. Doctoral dissertation, Graduate School of the Chinese Academy of Sciences, Beijing, China. 1-83.

Wu TH, Li SX, Cheng GD, Nan ZT, 2005. Using ground-penetrating radar to detect permafrost degradation in the northern limit of permafrost on the Tibetan Plateau. Cold Regions Science and Technology, 41(3): 211-219.

Wu TH, Zhao L, Li SX, Xie CW, Pang QQ, Zhang WG, 2008. Freezing/thawing index variation during the last 40 year over the Tibet Plateau. In:Kane DL, Hinkel KM (eds.). Proceedings of the 9th International Conference on Permafrost, Fairbanks, Alaska, USA, June 29 to July 3, 2008, Vol.2: 1969-1973. University of Alaska Fairbanks, Fairbanks, AK, USA.

Xu XZ, Guo DX, 1982. Compilation of the map of permafrost distribution in China (1:4,000,000). Journal of Glaciology and Geocryology, 4(2):18-26.

Xu XK, Chen H, Levy JK, 2008. Spatiotemporal vegetation cover variations on the Qinghai-Tibet Plateau under global climate change. Chinese Science Bulletin, 53(6): 915-922.

Yang JP, Ding YJ, 2004. Changes of ecological environments in the Sources Area of the Three Rivers during the past 40 years. Journal of Glaciology and Geocryology, 26(1): 7-16.

Yang MX, Wang SL, Yao TD, Gou XH, Lu AX, Guo XJ, 2004. Desertification and its relationship with permafrost degradation on the Qinghai-Tibet Plateau. Cold Regions Science and Technology, 39: 47-53.

Zhang SQ, Wang YG, Zhao YZ, Huang Y, Li YG, Shi WD, Shang XG, 2004.Degradation of permafrost in the sources area of the Yellow River and its environmental implications. Journal of Glaciology and Geocryology,26(1): 1-6.

Zhang WG, Li SX, Pang QQ, 2009. Spatiotemporal changes in precipitation on the Qinghai-Tibet Plateau during the last 40 years. Advances in Water Science, 20(2): 168-176.

Zhao L, 2003. Changes of permafrost on the Qinghai-Tibet Plateau and its environmental and engineering impacts. In: Zheng D (ed.). Formation, Environments and Development of the Qinghai-Tibet Plateau. Hebei Science and Technology Press, Shijiazhuang, China. 143-150.

Zhao L, Wu QB, Marchenko SS, Sharkhuu N, 2010a. Thermal state of permafrost and active layer in central Asia during the International Polar Year. Permafrost and Periglacial Processes, 21: 198-207. DOI:10.1002/ppp.688.

Zhao L, Ding YJ, Liu GY, Wang SL, Jin HJ, 2010b. Estimates of the reserves of ground ice in permafrost regions on the Tibetan Plateau. Journal of Glaciology and Geocryology, 32(1): 1-9.

Zheng Z, Yuan BY, Petit-Maire N, 1998. Paleoenvironments in China during the Last Glacial Maximum and the Holocene Optimum. Episodes, 21(3): 152-158. Online at: http://www.episodes.org/backissues/213/152-158%20Zheng.pdf.

Zhou YW, 1965. Permafrost along the Qinghai-Tibet Highway. In: Monograph on the Permafrost along the Qinghai-Tibet Highway. Science Press,Beijing, China. 1-10.

Zhou YW, Guo DX, 1982. Major features of permafrost in China. Journal of Glaciology and Geocryology, 4(1): 1-19.

Zhou YW, Guo DX, Qiu GQ, Cheng GD, Li SD, 2000. Frozen Ground in China. Science Press, Beijing, China. 367-376.

Appendix 1: Acronyms used in this paper

DA—Decadal Average

DEM—Digital Elevational Model

EEM—Equivalent Elevation Model

FI—Freezing or Frost Index

FMP—Fenghuoshan Mountain Pass

GIS—Geographic Information System

GCM—General Circulation Model

GPR—Ground Penetrating Radar

HMSS—Highway Maintenance Squad Station

HPS—Huashixia Permafrost Station

ILwSI—Ice layer with soil inclusions

IMP—Ice-medium permafrost

IPCC—Intergovernmental Panel on Climatic Change

IPP—Ice-poor permafrost

IRP—Ice-rich permafrost

ISP—Ice-saturated permafrost

KMP—Kunlun Mountain Pass

LGM—Last Glaciation Maximum

LIA—Little Ice Age

LLP—Lower Limit of Permafrost

MAAT—Mean Annual Air Temperature

MAGT—Mean Annual Ground Temperature (at the depth of zero annual amplitude)

MAGST—Mean Annual Ground Surface Temperature (generally measured at the depth of 5 cm)

MAST—Mean Annual Soil Temperature (at any depth)

QKH (G214)—Qinghai-Kang (Western Sichuan) Highway (G214 is from Xining to Jinghong, Yunnan)

QTH (G109)—Qinghai-Tibet Highway from Xining, via Golmud,to Lhasa (G109 is from Beijing to Lhasa)

QTP—Qinghai-Tibet Plateau, sometimes also called Tibetan Plateau

QTR—Qinghai-Tibet Railway from Xining, via Golmud, to Lhasa

SAYR—Source Area of the Yellow River

SLP—Southern Limit of Permafrost

SALR—Source Area of Lancang-Mekong River

SATR—Source Areas of the "Three Rivers"

STH (G318)—Sichuan-Tibet Highway from Chengdu to Lhasa

TI—Thawing Index

TMP—Tanggula Mountain Pass

TSP—Thermal State of Permafrost

TTOP—Temperature at the Top of Permafrost

XTH (G219)—Xinjiang-Tibet Highway from Yecheng, Xinjiang to Lhasa, Tibet

10.3724/SP.J.1226.2011.00281

*Correspondence to: Professor HuiJun Jin, State Key Laboratory of Frozen Soils Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, No. 320, West Donggang Road, Lanzhou, Gansu 730000, China. Email: hjjin@lzb.ac.cn

16 March 2011 Accepted: 20 June 2011

Ap pend ix2b:T ab leof coord in atesof imp ortantb oreh olesand sites in this p ap er