Journal of Geophysical Research: Atmospheres

Changes in active layer thickness over the Qinghai-Tibetan Plateau from 1995 to 2007

Authors

  • Qingbai Wu,

    1. State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Science, Lanzhou, China
    Search for more papers by this author
  • Tingjun Zhang

    1. State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Science, Lanzhou, China
    2. National Snow and Ice Data Center, Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author

Abstract

[1] The active layer over permafrost plays a significant role in surface energy balance, hydrologic cycle, carbon fluxes, ecosystem, and landscape processes and on the human infrastructure in cold regions. Over a period from 1995 to 2007, a systematic soil temperature measurement network of 10 sites was established along the Qinghai-Tibetan Highway. Soil temperatures up to 12 m depth were continuously measured semimonthly. In this study, we investigate spatial variations of active layer thickness (ALT) and its change over the period of record. We found that ALT can be estimated with confidence using semimonthly soil temperature profiles compared to those determined from available daily soil temperature profiles over the Qinghai-Tibetan Plateau. The primary results demonstrate that long-term and spatially averaged ALT is ∼2.41 m with a range of 1.32–4.57 m along the Qinghai-Tibetan Highway. All monitoring sites show an increase in ALT over the period of their records. The mean increasing rate of ALT is ∼7.5 cm/yr. ALT shows a closely positive correlation with the thawing index of air temperature on the plateau. We estimated ALT using the thawing index over a period from 1956 to 2005 near the Wudaoliang Meteorological Station in the northern plateau. ALT had no or very limited change from 1956 to 1983 and a sharp increase of ∼39 cm from 1983 to 2005. The magnitude of ALT increase is greater in the warm permafrost region than in the cold permafrost region. The primary control of increase in ALT is caused by an increase in summer air temperature, whereas changes in the winter air temperature and snow cover condition play no or a very limited role.

1. Introduction

[2] The active layer is defined as the top layer of ground subject to annual thawing and freezing in areas underlain by permafrost [van Everdingen, 1998]. It is primarily controlled by surface temperature, physical and thermal properties of the surface cover and subsoil, vegetation, soil moisture, duration, and thickness of seasonal snow cover [Brown et al., 2000; Hinkel and Nelson, 2003; Frauenfeld et al., 2004; Zhang et al., 2005; Smith et al., 2009]. The active layer plays a significant role in the surface energy balance, hydrologic cycle, carbon exchange between the atmosphere and the land surface, ecosystem, and landscape processes and on the human infrastructure in cold regions [Brown et al., 2000; Lemke et al., 2007]. Active layer thickness (ALT) generally ranges from a fraction of a meter in the Arctic to several meters at low and middle latitudes and is a relatively fast response layer over permafrost to climate variability. It is therefore critical to monitor changes in ALT at local and regional scales over decadal periods.

[3] In recent years, studies on ALT changes have received much attention in the world scientific community. Perhaps, the most comprehensive and systematic study is the Circumpolar Active Layer Monitoring (CALM) network, which was established in the early 1990s to observe the response of the active layer and near-surface permafrost to climate change [Brown et al., 2000; Nelson et al., 2004]. CALM currently has participants from 15 countries with more than 125 sites in both hemispheres [Brown et al., 2000; Nelson et al., 2004]. Using historical data and preliminary results from the CALM data, it has been found that ALT over the North American Arctic shows substantial interannual variability with no or an insignificant long-term trend since the 1980s [Brown et al., 2000; Nelson et al., 2004]. However, based on historical soil temperature measurements from 37 hydrometeorological stations over the period 1956–1990, the active layer exhibited a statistically significant thickening of ∼21 cm over Russian Arctic regions attributed mainly to the increase in summer air temperature, whereas seasonal freeze depth has decreased by ∼34 cm because of the increases in winter snow depth in nonpermafrost regions of Russia [Frauenfeld et al., 2004; Zhang et al., 2005; Lemke et al., 2007]. Evidence from European monitoring sites indicates that ALT has been increasing over the past decade. It was the greatest in the summers of 2002 and 2003, approximately 20% greater than in previous years [Harris et al., 2003].

[4] The Qinghai-Tibet Plateau plays important roles in the global climate system because of its unique topography [King et al., 2006]. Monitoring of active layer and borehole temperatures on the Qinghai-Tibetan Plateau is an important component of CALM and has the Global Terrestrial Network for Permafrost [Brown et al., 2000]. However, there are relatively little discussions in literature on the research of the active layer over the Qinghai-Tibetan Plateau because of scarcity of meteorological stations and lack of continuous long-term records in permafrost regions. Observed evidence shows that permafrost is warming, thawing, and degrading during the past few decades on the Qinghai-Tibetan Plateau [Cheng and Wu, 2007; Wu and Zhang, 2008]. Over the period 1995–2004, ALT has deepened by 10–40 cm in permafrost regions along the Qinghai-Tibetan Highway [Wu and Liu, 2004; Yang et al., 2004; Wu et al., 2006]. Depth of seasonally frozen ground has reduced, on average, by 22 cm, with an average decreasing rate of ∼0.71 cm/yr since the 1980s over the eastern Qinghai-Tibetan Plateau [Zhao et al., 2004; Wang et al., 2005].

[5] Soil temperatures in 10 shallow boreholes (up to 12 m) along the Qinghai-Tibetan Highway have been measured since 1995 [Wu and Zhang, 2008]. In this study, soil temperatures measured from these shallow boreholes are used to investigate changes in ALT and their response to climate change over a period from 1995 to 2007. First, we will validate the method to use semimonthly soil temperature data to determine thaw depth over permafrost. Second, we will study annual and interannual variations of ALT determined from semimonthly soil temperature data along the Qinghai-Tibetan Highway. Third, we will further validate the relationship between ALT and the thawing index over the Qinghai-Tibetan Plateau and document changes in ALT estimated from the thawing index of air temperatures since the late 1950s. Future perspectives of active layer and permafrost study over the Qinghai-Tibetan Plateau will also be discussed.

2. Data Sources and Site Description

2.1. Data Sources

[6] The data used in this study include daily and mean monthly air temperature at the Wudaoliang, Tuotuohe, and Anduo Weather Stations and the Beiluhe Permafrost Station and semimonthly soil temperatures from 10 sites along the Qinghai-Tibetan Highway over the period 1995–2007.

[7] Soil temperature was measured from 0.5 m to depths of up to 12 m with an interval of 0.5 m at 10 monitoring sites along the Qinghai-Tibetan Plateau. Soil temperature was measured using themisters made by the State Key Laboratory of Frozen Soil Engineering and digital multimeters (Fluke 180 series; Fluke, Everett, WA USA). Overall accuracy of the system in the laboratory is about ±0.05°C. The measurements were performed by well-trained technicians and professionals with an established uniform guideline, measured on the 5th and 20th of each month, respectively. These data are the longest continuous record of permafrost temperatures at present over Qinghai-Tibetan Plateau and are very useful for investigating the variation of ALT and thermal regime of permafrost.

[8] Daily air and soil temperatures up to 20 m were also measured at the Beiluhe Permafrost Station (Figure 1). Daily air temperature data at the Beiluhe Permafrost Station were used to the estimate thawing index. Daily soil temperatures up to 20 m were used to estimate thaw depth and ALT. Thawing index and ALT data were used to calibrate and validate (1) the relationship between the thawing index and ALT and (2) the method to estimate ALT using semimonthly soil temperature profiles. Daily and mean monthly air temperatures at the Wudaoliang Weather Station were used to estimate historical freezing and thawing indexes. We also use freezing and thawing indexes to study historical changes in ALT.

Figure 1.

Locations of the 10 borehole monitoring sites, meteorological stations, and Beiluhe Permafrost Station along the Qinghai-Tibetan Highway.

2.2. Site Description

[9] Soil temperatures were measured at 10 sites from the Kunlun Pass in the north to Anduo in the south along the Qinghai-Tibetan Highway (Figure 1). Geographical information of these sites is listed in Table 1. Soil temperatures were measured semimonthly up to 12 m over a period from 1995 to 2007.

Table 1. Geographical Data and Information of the 10 Monitoring Sites on the Qinghai-Tibetan Plateau
SitesAreasLocationAltitude (m)Distance between Site to Highway Right-of-Way (m)Observation Period
LatitudeLongitude
KM1Kunlun Mountains35°37′58″N94°04′18″E4770201995–2002
KM235°37′12″N94°04′13″E4759301995–2002
CM1Chumaer River35°30′50″N93°44′05″E4552301995–2002
CM235°23′49″N93°32′01″E4482301995–2007
WD1Wudaoliang35°13′48″N93°05′06″E4610301995–2002
WD235°07′36″N93°02′12″E4707301995–2007
FH1Fenghuo Mountains34°41′24″N92°53′30″E49381001995–2007
TG1Tanggula Mountains32°42′30″N91°52′16″E4997201998–2007
TM1Touerjiu Mountains32°29′33″N91°49′17″E4873501998–2007
AD1Near Anduo32°23′00″N91°42′27″E4786201998–2007

[10] Mean annual air temperature ranges from about −6.5°C at the Kunlun Pass to about −2.0°C at Anduo along the Qinghai-Tibetan Highway (Table 2). Long-term (1971–2000) mean air temperatures during the coldest month (January) were about −16.6°C at Wudaoliang, −16.2°C at Tuotuohe, and −14.6°C at Anduo. During the warmest month (July), they were 5.6°C, 7.5°C, and 7.8°C, respectively. Mean monthly air temperatures above 0°C last for 4 to 5 months per year over the plateau.

Table 2. Climate and Environmental Parameters at the 10 Monitoring Sites on the Qinghai-Tibetan Plateaua
SitesClimatic ConditionsSoil TypesVegetation Cover (%)Geomorphologic UnitPermafrost Conditions
MAAT (°C)Precipitation (mm)MAGT (°C)PT (m)FST
  • a

    MAAT, mean annual air temperature; MAGT, mean annual ground temperature at a depth of zero annual amplitude, usually at 10–15 m depth below the ground surface on the Plateau; PT, permafrost thickness; FST, frozen soil types: H, frozen soils with ice; B, saturated frozen soils; F, saturated frozen soils with excess ground ice.

KM1−6.0 to −6.5250–300Sandy clay10–15Top of the Kunkun Pass, flat surface−2.0 to −3.080–120H, B
KM2−6.0 to −6.5250–300Sandy clay10–15Gentle south-facing slope−2.0 to −3.080–120H, B
CM1−5.0 to −5.5230–250Sandy silt with gravel20–30Gentle slope plain−0.1 to −1.010–30F
CM2
WD1−5.5 to −6.0300–350Gravel and sandy silt15–20Top of hill, relatively flat−1.0 to −2.560–80F
WD2−5.5 to −6.0300–350Sandy silt with gravel60–80Gentle south-facing slope−1.0 to −2.560–80H, B
FH1−6.0 to −6.5250–300Sandy silt with gravel70–80Southeastern-facing slope−2.0 to −3.080–120H, B
TG1−6.0 to −6.5250–300Gravel and sandy silt20–30Gentle slope−1.0 to −2.590–120F
TM1−4.0 to −4.5350–400Gravel and sandy silt50–60Gentle slope−0.5 to −1.240–60B
AD1−2.0 to −3.0400–450Sandy silt90Gentle slope0.0 to −0.510–20H, B

[11] Climate over the Qinghai-Tibetan Plateau is extremely continental with annual total precipitation generally less than 300 mm, ranging from ∼250 mm in the north to 450 mm in the south. About 10% of annual precipitation falls as snow because of the impact of the monsoon climate [Wu and Zhang, 2008]. Snowfall occurs in the late autumn or early spring. Snow cover lasts for only a few days to 1 week with its thickness less than 10 cm. The majority of the plateau has a snow-free land surface in winter [Wu and Zhang, 2008]. There was no steady or winter-long snow cover on the ground at all study sites during the study period. The previous study shows that the lack of snow cover has a cooling impact on soils over the plateau because of the timing of snowfall, discontinuity, and thickness of snow cover [Zhou et al., 2000; Zhang, 2005].

[12] Overall, vegetation cover is sparse as relatively dry tundra (Figure 2); in most cases, it is even less sparse than vegetation in the Arctic tundra. Vegetation cover ranges from less than 30% in the north to above 90% in the south with vegetation height of ∼10–15 cm. Near-surface deposits are dominated by coarse materials such as gravel and sandy soils. Soil organic content is relatively low with no peat layer, which is in sharp contrast to thick accumulation of peat that may be found in most areas of the Arctic regions [Zhang et al., 1997; Brown et al., 2000; Smith et al., 2009]. Soil moisture content is generally low except over lowlands and river banks. Although elevation for all sites is close to or above 4500 m a.s.l., observational sites are located on either gentle slopes or flat plains. Detailed site-specific conditions are summarized in Table 2.

Figure 2.

In situ surface conditions at selected monitoring sites along the Qinghai-Tibetan Highway.

2.3. Methods

[13] There are two definitions of the active layer. According to Muller [1947], the base of the active layer corresponds to the depth of the maximum seasonal penetration of the 0°C isotherm. This definition is widely adapted [e.g., Hinkel et al., 1997; Nelson et al., 1998; Brown et al., 2000; Burn, 1998; Frauenfeld et al., 2004; Zhang et al., 2005]. In contrast, in two publications [Associate Committee on Geotechnical Research, 1988; van Everdingen, 1998], the active layer is defined as “the top layer of ground subject to annual thawing and freezing in areas underlain by permafrost.” There are some differences between these two definitions because of soil particles, salinity, and ground temperature [Burn, 1998; Brown et al., 2000; Zhang et al., 2005]. In this study, we use Muller’s original ALT definition of the maximum 0°C isotherm penetration depth directly related with measurements of soil temperatures. In other words, ALT is estimated as the maximum thaw depth in the late autumn through linear interpolation of soil temperature profiles between two neighboring points above and below the 0°C isotherm at all sites.

[14] Ideally, the maximum depth of 0°C isotherm should be determined using daily soil temperature measurements with depth [Brown et al., 2000]. However, a previous study shows that the difference in estimated ALT using daily and mean monthly soil temperatures is very small [Frauenfeld et al., 2004]. In this study, soil temperatures were measured semimonthly. We will also investigate the validity of using semimonthly soil temperatures to estimate ALT.

[15] The difference of ALT estimated by daily and semimonthly soil temperatures was studied based on daily soil temperature measurement along the Qinghai-Tibetan Railway. Daily soil temperatures at various depths from 27 sites along the newly constructed Qinghai-Tibetan Railway were measured since 2005. Because semimonthly soil temperature measurements along the Qinghai-Tibetan Highway were approximately conducted on the 5th and 20th day each month, we extracted soil temperatures on the 5th and 20th day each month at these 27 sites. Based on the daily and the extracted semimonthly soil temperature data, we estimated the maximum depth of 0°C isotherm at these sites using linear interpolation between two neighboring points above and below 0°C isotherm along the Qinghai-Tibet Railway. The ALT estimated using daily and semimonthly soil temperature is compared (Figure 3). ALT estimated by both daily and semimonthly soil temperatures is in good agreement (Figure 3a). The difference varies within 10 cm, and, most of all, the difference within 5 cm accounts for ∼85% in all observation sites (Figure 3b). These results indicate that ALT estimated by semimonthly soil temperature is reliable, in part supporting the conclusion by Frauenfeld et al. [2004] of using mean monthly soil temperature to estimate ALT. In the following study, we use ALT estimated from semimonthly soil temperature measurements along the Qinghai-Tibetan Highway.

Figure 3.

(a) Comparison of ALT estimated by daily and semimonthly soil temperature measurements from the 27 sites along the Qinghai-Tibet Railway. (b) Range of ALT difference determined using daily and semimonthly soil temperatures from the 27 sites along the Qinghai-Tibet Railway. The long dashed line represents the mean value of ALT difference from the 27 sites. Daily soil temperatures up to 10 m were measured at these 27 sites, from Xidatan in the north to Anduo in the south, along the Qinghai-Tibetan Railway.

3. Results

3.1. Features of Soil Seasonal Freezing and Thawing Processes

[16] Based on the multiyear average soil temperature data, the soil thermal regime including the position of the 0°C isotherm at a warm (CM2) and a cold (WD2) permafrost area is shown in Figure 4. Because of the difference in climate conditions, landscape, soil properties, and vegetation, freezing and thawing processes of the active layer have a pronounced difference for warm and cold permafrost over the Qinghai-Tibetan Plateau. Soils at 50 cm depth start to thaw at the beginning of May and to refreeze at the beginning of November over warm permafrost areas (Figure 4a). The maximum thaw depth occurs at the beginning of February, and the duration of the active layer in the thawing state is up to 9 months per year. Soils at 50 cm depth start to freeze at the beginning of November, and the duration of the freeze season is ∼3–4 months until the beginning of February. The duration of the cooling process is reduced to be shorter than 1 month. In cold permafrost areas (Figure 4b), soils at 50 cm depth start to thaw at the beginning of June and stop thawing at the end of October. The maximum thaw depth can be reached at the end of September, and duration of the active layer in the thawing state is only about 4 months, 5 months shorter than that in warm permafrost areas. Soils at 50 cm depth start to freeze at the end of October, and the duration of the freezing process (Figure 4b, AF) finishes within 15 days because of refreezing upward from the top of the permafrost table. The cooling process (Figure 4b, AF) is also longer than that in warm permafrost areas.

Figure 4.

Characteristics of soil freezing and thawing processes of the active layer for (a) warm and (b) cold permafrost on the Qinghai-Tibetan Plateau. SW, summer warming; ST, summer thaw; AF, autumn freezing; WC, winter cooling.

[17] Figure 5 illustrates the processes of soil freezing and thawing at observed CM2 sites from March 1996 to March 1997 (a) and from March 2005 to March, 2006 (j). Because of the effect of climatic and site-specific conditions on the processes of soil freezing and thawing, the first and last days and the duration of the active layer in the thawing state were different at 50 cm depth over the period of records. The first day of soil thawing at 50 cm ranged from 25 April 2005 to 12 May 2000 with an average around 4 May (Table 3) and an interannual variation of up to 17 days. The last day of soil thawing at 50 cm depth ranged from 6 November 1996 to 24 October 2005 with an average around 30 October (Table 3) and an interannual variation of up to 13 days. The duration of the active layer in the thawing state at 50 cm depth ranged from 172 to 185 days with an average variation of up to 13 days (Table 3). The date of maximum thaw over 1 year ranges from 11 February 1998 to 15 March 1999 with an average around 29 February, and the duration of the active layer in the thawing state ranges from 283 to 324 days with an average variation of up to 41 days. The first day of soil freezing at 50 cm depth ranges from 6 November 1996 to 24 October 2005 with an average around 30 October (Table 3). The last day of soil freezing ranges from 11 February 1998 to 15 March 1999 with an average around 27 February, and the duration of soil freezing process ranges from 98 to 137 days with an interannual variation of up to 39 days.

Figure 5.

Characteristics of the seasonal freezing and thawing processes of the active layer at the CM2 site from 1996 to 2006.

Table 3. Timing of Soil Seasonal Freeze and Thaw at 50 cm Depth for the CM2 Site
YearFirst Day of Soil ThawingLast Day of Soil ThawingDuration of Soil ThawingFreeze-Up Dates
19965 May6 Nov18512 Feb 1997
19977 May1 Nov17811 Feb 1998
19982 May29 Oct18115 Mar 1999
199912 May5 Nov17714 Mar 2000
200011 May29 Oct17221 Feb 2001
20019 May2 Nov17725 Feb 2002
200229 Apr26 Oct18126 Feb 2003
20033 May29 Oct1809 Mar 2004
200428 Apr25 Oct1816 Mar 2005
200525 Apr24 Oct18328 Feb 2006
Average4 May30 Oct179.527 Feb

3.2. ALT Variations

[18] Based on average semimonthly soil temperature from the 10 sites, the average ALT from 1995 to 2007 along the Qinghai-Tibetan Railway is 2.41 m (Figure 6a); average ALT varied from less than 1.3 m at the KM1 site to greater than 4.57 m at the TM1 site. The magnitude of interannual ALT variation ranges from less than 0.30 m at WD1 and TG1 sites to greater than 1.53 m at the TM1 site. Obviously, the magnitude of interannual ALT variations at CM2, TM1, and AD1 is greater than the other observed sites (Figure 6a), up to 0.92, 1.53, and 1.10 m, respectively. ALT continuously increases for all observed sites over the Qinghai-Tibetan Plateau (Figures 6b and 6c). The annual increase rate of ALT ranged from 2.1 cm/yr at TG1 to 16.6 cm/yr at TM1 with an average of 7.5 cm/yr (Figure 7).

Figure 6.

(a) Mean ALT and range over the period of record at each site. The open circle represents mean ALT thickness, and the bar represents the ALT range at each site. The dashed line indicates the mean ALT (2.39 m) over all observed sites. (b) Bar graph of ALT departure from its mean at each site. (c) Time series of ALT departure from its mean for each site over its period of record.

Figure 7.

Annual increase rate of ALT for all observed sites. The solid circle represents the mean increase rate, and the bar represents the range at each site.

3.3. Relationship Between ALT and Soil Temperature at 50 cm Depth

[19] ALT is closely related to the ground surface temperature, but the difference of surface temperature among sites commonly occurs because of difference in snow cover, vegetation, atmosphere radiation, and soil moisture conditions [Lunardini, 1981; Burn, 1992; Colin et al., 1999; Zhang et al., 2001; Zhang, 2005; Osterkamp, 2005; Smith et al., 2009]. This makes it difficult to directly establish the relation between ALT and surface temperature. In this study, we discuss the relationship between ALT and the average soil temperatures at the depth of 50 cm using these data of eight observed sites because of a lack of soil temperature data at 50 cm depth at TM1 and AD1. Based on soil temperature at the depth of 50 cm, the summer average soil temperatures during June, July, and August and the winter average soil temperatures during December, January, and February were used to establish the correlation with ALT (Figure 8). ALT is positively correlated with summer average soil temperatures at 50 cm depth with R2 = 0.67 and p < 0.001 (Figure 8b). ALT is also positively, in general, related to winter average soil temperatures at 50 cm depth with R2 = 0.28 and p < 0.001 (Figure 8a). This indicates that the variation of ALT strongly depends on soil temperature in summer and also less strongly depends on variation of soil temperatures in winter over the Tibetan Plateau. In other words, ALT is primarily controlled by summer climate conditions and modified by winter climate conditions as well, on the Qinghai-Tibetan Plateau. A previous study showed that besides impact of changes in summer air temperature, changes in winter snow conditions and air temperature also have a significant impact on ALT in Siberia [Frauenfeld et al., 2004; Zhang et al., 2005]. A noteworthy feature is that at the CM2 site, ALT is negatively correlated with average winter soil temperatures at 50 cm depth (Figure 8a). We do not have sufficient data to physically explain this phenomenon.

Figure 8.

Relationship between ALT and (a) winter and (b) summer average soil temperatures at 50 cm depth.

3.4. ALT Determined from Air Temperature Thawing Index

[20] ALT can be estimated from a variant of Stefan solution using a thawing index [Zhang et al., 2005], and the general validity of Stefan formulation has been demonstrated for northern Alaska by Romanovsky and Osterkamp [1995], Zhang et al. [1997], and Nelson et al. [1998]. The annual thawing index is defined as the cumulative number of degree-days above 0°C over 1 year. It is estimated by adding all the positive mean daily temperatures (°C) for a specific station during a calendar year. Generally, the higher thawing index implies a long and warmer thaw season, potentially producing a deeper active layer. In this study, we estimated thaw depth for each summer soil temperature profile at the WD2 site from 1995 to 2007. As stated earlier, soil temperatures were measured on the 5th and 20th each month. In this case, we have 10 thaw depth values each year starting from 20 May and ending on 5 October. Correspondingly, we estimated the thawing index of air temperature at the Wudaoliang Weather Station (35.22°N, 93.08°E, 4612 m a.s.l.) from the beginning of the calendar year to the date of each summer soil temperature measurements at the WD2 site. We then use the thawing index of air temperature and the thaw depth to analyze their relationship. Figure 9a shows the relationship between the mean thaw depth on the 5th and 20th of each month over 12 years and the corresponding square root of the thawing index. Bars in Figure 9a represent 1 standard deviation from its mean of thaw depth. Figure 9b shows the relationship between the thaw depth and the corresponding square root of the thawing index for all data points. The two variables are strongly linearly correlated with R2 = 0.99, p < 0.01 (Figure 9a) and R2 = 0.92, p < 0.01 (Figure 9b).

Figure 9.

Correlations between the thaw depth (TD) at the WD2 site and the square root of the thawing index of air temperature at the Wudaoliang Weather Station. (a) Relationship between the mean thaw depth on the 5th and 20th of each month over 12 years and the corresponding mean square root of the thawing index. Bars represent the standard deviation from its mean of thaw depth. (b) Relationship between the thaw depth and the corresponding square root of the thawing index for all data points.

[21] To further investigate the ALT estimated by the air temperature thawing index, we also studied the correlation between ALT and the thawing index at the Beiluhe Permafrost Station (Figure 10a). As a result, ALT is strongly correlated with the square-root thawing index of air temperature (R2 = 0.87, p < 0.01), and the estimated ALT using the correlation is close to the measured ALT results (Figure 10b).

Figure 10.

(a) Correlation between the ALT estimated from semimonthly soil temperature profiles and the square-root thawing index of air temperature. (b) Comparison of the ALT estimated from semimonthly soil temperature profiles and the ALT estimated by the equation in Figure 10a. Data are from the Beiluhe Permafrost Station.

[22] Based on results from Figures 9 and 10, we can estimate ALT using the thawing index of air temperature with confidence. Using daily air temperature from the Wudaoliang Meteorological Station, we calculated the annual thawing index for the period 1956–2005. Furthermore, we use the thawing index to estimate ALT over the same period using the equation in Figure 9a (Figure 11). ALT had a large interannual variation from about 1.4 m to greater than 1.9 m with no significant trend from 1956 to 1983 (Figure 11). After 1983, ALT experienced a significant increase at an average rate of ∼18 cm/decade, which is about 3 times greater than the average rate of ALT increase in Siberia over the period 1956–1990 [Frauenfeld et al., 2004; Zhang et al., 2005]. ALT varied from less than 1.6 m to greater than 2.0 m during this period (Figure 11).

Figure 11.

Interannual variations of the ALT estimated using the thawing index of air temperature from the Wudaoliang Meteorological Station (see Figure 10b) from 1956 to 2005.

4. Summary and Discussions

[23] This study examined variation of ALT along the Qinghai-Tibetan Highway using semimonthly soil temperature measurements from 1995 to 2007. Research results show that ALT has extensive temporal and spatial differences along the Qinghai-Tibetan Highway. Based on data from these 10 sites over 12 years and on information from this study, the average ALT along the Qinghai-Tibetan Highway is ∼2.41 m with a range of 1.32–4.57 m. The mean increasing rate of the ALT is ∼7.5 cm/yr with a range of 2.1–16.6 cm/yr from 1995 to 2007 (Table 4). Soil temperatures at 50 cm depth were closely related with ALT. ALT and their annual increasing rate in cold permafrost regions were larger than that in warm permafrost regions. The multiyear mean ALT in cold permafrost areas is ∼1.92 m, whereas in warm permafrost areas, ALT is up to 3.15 m (Table 4).

Table 4. Mean ALT, Increase Rate of ALT, and Permafrost Temperatures from the 10 Monitoring Sites Along the Qinghai-Tibetan Highway
Permafrost Temperature Lower Than1°C
SiteKM1KM2WD1WD2FH1TG1Mean
T at 6.0 m depth (°C)−3.02−2.91−1.57−1.71−3.43−1.13 
ALT (m)1.321.342.641.951.552.731.92
Increase rate of ALT (cm/yr)8.984.194.044.875.502.104.95
Permafrost Temperature Higher Than1°C
SiteCM1CM2TM1AD1  Mean
T at 6.0 m depth (°C)−0.87−0.42−0.19−0.25   
ALT (m)2.153.604.572.29  3.15
Increase rate of ALT (cm/yr)8.467.5416.6012.35  11.24

[24] Changes in ALT have a significant difference between cold and warm permafrost regions over the Qinghai-Tibetan Plateau. The average increasing rate of the ALT is ∼5.0 cm/yr in cold permafrost regions, whereas over warm permafrost areas, the mean increasing rate is up to 11.2 cm/yr (Table 4). At TM1 and AD1 sites, the ALT increasing rate is up to 16.6 and 12.4 cm/yr, respectively (Table 4). The magnitude of the rate of ALT increase is more than doubled in warm permafrost regions than in cold permafrost regions, although mean annual air temperature has increased ∼1°C uniformly across the plateau since the early 1960s [Frauenfeld et al., 2004]. We hypothesize that the higher rate of ALT increase in warm permafrost than in cold permafrost is mainly attributed to changes in unfrozen water content with temperature. For cold permafrost, much energy entering the active layer and permafrost may be largely consumed for latent heat fusion because of the increase in unfrozen water content and for specific heat consumption because of permafrost temperature increase. In terms of cold permafrost, it refers to mean annual ground temperature ranging from −1°C to −3°C when unfrozen water content increases sharply with permafrost temperature increase. For warm permafrost, much of the energy entering the active layer and permafrost is consumed for melting the remaining ground ice, thus increasing ALT. It should be noted that warm permafrost refers to mean annual ground temperature above −1.0°C, mostly higher than −0.5°C (Table 4). Much or a substantial fraction of ground ice in warm permafrost was converted into unfrozen water before 1995. As a result, for approximately the same amount of energy entering the active layer and permafrost, the magnitude of ALT change is greater for warm permafrost than for cold permafrost.

[25] For the difference in ALT change among sites within warm and cold permafrost regions, it may be explained by site-specific microclimate conditions and other site-specific parameters. For warm permafrost sites, changes in active layer thickness were be mainly controlled by changes in air temperature and soil moisture conditions. For example, the increase rate of ALT at AD1 and TM1 sites is approximately 50%–100% higher than the average increase rate of ALT at CM1 and CM2 sites (Table 2). This is because the increase rate of air temperature at and near AD1 And TM1 sites (Anduo station) are higher than that near CM1 and CM2 sites (Wudaoliang station) [see Wu and Zhang, 2008]. From 1995 to 2005, the rate of air temperature increase at the Anduo station was ∼0.65°C/decade, while at the Wudaoliang station, the rate was ∼1.35°C/decade [Wu and Zhang, 2008]. This may partly explain the higher increase rate of ALT at AD1 and TM1 sites. On the other hand, precipitation (especially summer rainfall) at the Anduo station is as about twice as that at the Wudaoliang station [Wang et al., 2007; Wu and Zhang, 2008], resulting in a wet surface and a near-saturated active layer with relatively well-developed vegetation. Both field measurements [Veldhuis et al., 2002] and numerical modeling (Z. Fan, personal communication, 2006) show that changes in ALT are greater at poorly drained sites (wet sites) than at well-drained sites (drier sites). This may also partly explain the higher rate of ALT changes in relatively wet sites (TM1 and AD1) than that at drier sites (CM1 and CM2). However, continuing monitoring of ALT change and comprehensive measurements of surface and soil energy fluxes are required for further investigation.

[26] It is widely hypothesized that ALT will increase in response to climate warming. It may be complex for the response of ALT to climate change; however, the seasonality of air temperature change is possibly predominant for the ALT variation over the Qinghai-Tibetan Plateau, and summer air temperature is the primary control on ALT, while winter climate conditions may also play a minor role. The relationship between soil temperature at the depth of 0.5 m and ALT can explain that this variation pertains to summer air temperature, which is essentially represented by the annual thawing index, as has been widely documented in the Arctic [Romanovsky and Osterkamp, 1995; Zhang et al., 1997; Nelson et al., 1998; Brown et al., 2000; Hinkel and Nelson, 2003].

[27] The magnitude of ALT change on the Qinghai-Tibetan Plateau is generally larger than that in the Arctic and Subarctic. The average magnitude of ALT increase over the Qinghai-Tibetan Plateau is up to 67 cm from 1995 to 2007. ALT has increased ∼21 cm from the early 1960s to 2000 in Siberia, as determined from mean monthly soil temperature profiles [Frauenfeld et al., 2004; Zhang et al., 2005]. Over Alaska and northern Canada, there is evidence of large interannual variations of ALT but no significant trend since the early 1990s [Brown et al., 2000; Nelson et al., 2004; Smith et al., 2009].

[28] The main reason to explain the large increase in ALT since 1995 over the Qinghai-Tibetan Plateau is probably attributed to the direct coupling of the active layer-permafrost system to climate. Over the plateau, there is no peat layer and essentially a bare ground surface that make climate forcing have direct influence on active layer-permafrost system. In contrast, over most areas of the Arctic and Subarctic, a thick peat layer over a continuous permafrost zone and a thick peat layer with heavy shrub vegetation cover in a discontinuous permafrost zone act as a strong buffer layer, significantly reducing the climate forcing effect on ALT [Woo et al., 2007; Smith et al., 2009]. Relative dry soil, coarse soil materials, and lack of winter snow cover may also play an important role in the response of ALT to climate change over the plateau.

[29] Over the circum-arctic region, there is no significant trend of ALT increase since the early 1990s. Thaw settlement may play an important role because of thawing of ice-rich permafrost near the permafrost surface [Tarnocai et al., 2004; Atkinson et al., 2006; Smith et al., 2009]. However, there is probably no or very little thaw settlement on the Qinghai-Tibetan Plateau mainly because of the relative low ground ice content and coarse soil materials. Compared to ground ice conditions in the Arctic, permafrost over the Plateau contains relative less ice as shown on the International Permafrost Association’s Circum-Arctic Map of Permafrost and Ground-Ice Conditions [Brown et al., 1997; Zhang et al., 1999]. From these 10 observational sites, there is relatively little excess of ice near the permafrost surface, although at some sites where ice-saturated permafrost exists.

[30] It is certain that there are errors in ALT determined from linear interpolation between two neighboring points above and below the 0°C isotherm. We have no field data to further evaluate the magnitude of errors generated from the method applied in this study. Based on modeling study, however, Riseborough [2008] found that the potential errors are up to 10 cm for ALT with 50 cm spacing of sensors, approximately 2%–10% of ALT for all sites during the entire study period. The errors were relatively larger during the mid-1990s when ALT was thinner and became smaller during the mid-2000s. In this case, the overall error is approximately less than 5%. Considering ALT increase from 33 to 153 cm with an average of 67 cm over the entire study period, the errors from the linear interpolation method are much smaller; thus, the detected trend of ALT increase from this study is significant.

Acknowledgments

[31] We express our gratitude to three anonymous reviewers for their constructive and insightful comments and suggestions. We thank Liu Yongzhi and Ma Zhixue for conducting field measurements over the years. This research was supported in part by the Outstanding Youth Foundation Project, Natural Science Foundation of China (grant 40625004) to Qingbai Wu, the Program for Innovative Research Group of Natural Science Foundation of China (grant 40821001), and the International Arctic Research Center, University of Alaska Fairbanks, through the U.S. NSF cooperative agreement OPP-0327664 to the University of Colorado for Tingjun Zhang.

Ancillary