-
The basic data used in this study were obtained from the established Tanggula site (91°52′E; 33°04′N and 5100 m) (Fig. 1) located in the hinterland of the QTP. The observation site is flat with open surrounding terrain, comprising grassy alpine meadow as the main land surface type with 30%–40% coverage (Yao et al., 2008; Li et al., 2019). Average annual air temperature is about −4.9°C, mean annual air pressure is about 538 hPa, and annual precipitation is about 436.7 mm (Gu et al., 2015). Active layer thickness is about 3.36 m (Li et al., 2012), and the soil is frozen from December to March. The active layer thaws from the ground surface beginning around the end of April. Generally speaking, the Tanggula site represents prevailing conditions of permafrost in the QTP hinterland (Gu et al., 2015).
Figure 1. Map of the study area and the locations of the monitoring sites along the Qinghai-Tibetan Highway. The frozen ground map data were derived from Zou et al. (2017).
Data used in this study were primarily drawn from meteorological and active layer hydrothermal data on the Tanggula site from 1 January 2005, to 1 January 2015, but 2009 data is missing. Active layer hydrothermal data used in this study included soil temperature and soil moisture. Soil temperature was measured from 2–300 cm below ground by 105 T thermocouple probes with ±0.1°C accuracy. A Stevens Hydro probe was employed to measure soil moisture content with ±3% accuracy. Measurements taken by these sensors were recorded with a CR1000 data logger (Campbell Scientific). All instruments were sampled every 5 minutes and data were averaged over 30 minutes (Li et al., 2019). Meteorological data included soil heat flux, snow depth, precipitation, vapor pressure, air temperature, radiation fluxes, and wind velocity. Soil heat flux was monitored using HFP01 at 5, 10, and 20 cm below the surface with ±3% accuracy. Snow depth was measured by SR50-L, and precipitation was monitored using a T-200B weighing rain gauge (Geonor, Norway) with accuracy ±0.1 mm (Yang et al., 2020). Air temperature was measured by HMP45C-L at 2, 5, and 10 m above ground. Downward short-wave, upward short-wave, downward long-wave, and upward long-wave radiation fluxes were measured using a four-component net radiometer at 2 m above ground. Wind velocity was obtained using a 05103-L at 2, 5, and 10 m above ground. Meteorological sensors were connected to a CR23X data logger (Campbell Scientific) and data were recorded in Local Standard Time (LST, LST = UTC + 8). Table 1 provides detailed information of the instruments related to the study at Tanggula site.
Observation item Instruments Height/depth Soil heat flux HFP01 5, 10, 20 cm Soil temperature 107 2, 5, 10, 20, 40, 80, 120, 160, 200, 240, 280, 320, 360 cm Soil moisture content CS616 5, 10, 20, 40, 70, 105, 140, 175, 210, 245, 280, 300 cm Air temperature HMP-45C 2, 5, 10 m Relative humidity HMP-45C 2, 5, 10 m Precipitation T200-B 1.5 m Snow depth SR-50 2 m Short wave radiation CM3 2 m Long wave radiation CM3 2 m Table 1. Observation items and instruments.
-
Net radiation combines effects from various radiation components and a factor that characterizes radiant heat exchange in the Earth’s climate system, reflecting net radiation budget at the ground surface, which has great impact on shallow soil temperature. Therefore, the changes of the air temperature and shallow soil temperature are in good agreement with the annual changes in Rn (Wang et al., 2019). As can be seen in Fig. 2, Rn increases sharply from March to April, with corresponding significant soil temperature increase (Fig. 3). However, Rn decreases rapidly from September to October with significant soil temperature cooling (Figs. 2 and 3). The energy jumps from March to April and September to October are related to the seasonal changes of the QTP. Overall, step increases in radiant energy during spring and sharp declines in autumn greatly promote rapid atmospheric circulation conversion from spring to summer and autumn to winter in the QTP (Ji et al., 2002).
Figure 2. Process variations for several factors at Tanggula, in 2010: (a) surface energy fluxes (units: W m−2), and (b) precipitation (units: mm).
Figure 3. Active layer (a) soil temperature (units: °C) and (b) moisture profiles (units: m3 m−3) at Tanggula in 2010.
Ground heat flux (G0) reflects heat transfer between underlying surface and topsoil, and variations are mainly due to the difference in soil temperature and the underlying surface. Fluctuations in G0 are very consistent with Rn, following its seasonal variation. Hence G0 exhibits significant seasonal pattern (Fig. 2a), and both exhibit unimodal changes peaking in summer and plunging in winter. G0 was positive from April to September and negative from January to March and October to December. Thus, the soil absorbs heat from the surface during summer and then spreads it downward, whereas it releases heat upward during winter. Figure 3a shows that positive and negative changes in G0 are basically consistent with freeze-thaw processes of the active layer. Vertical changes in the soil temperature in the active layer (Fig. 3a) suggests that the soil starts to freeze in October and the freezing process ends in November. Soil temperature is negative from November to March of the following year. Thus, the negative G0 period is consistent with the freezing period. Rn gradually increases due to seasonal variations in solar altitude, and soil temperature also increases due to surface energy accumulation (You et al., 2017). Soil temperature is positive from mid to late April, with the frozen surface (0°C line) gradually moving downward to reach maximum depth in late September. The active layer is thawing and G0 is positive during this period.
Rapid changes in Rn (compare Figs. 2a and 3a) promote cold and warm season transition, which is very closely related to freeze-thaw processes in the active layer. The active layer begins to thaw during the conversion process, whereas it begins to change from warm to cold season when local surface energy budget falls sharply from September to October, and the active layer soil begins to freeze. In general, the freeze-thaw cycle of the active layer is closely related to the surface energy budget (Gu et al., 2015).
Sensible and latent heat flux (H and LE, respectively) are major terms in the surface energy balance equation. Figure 2a also illustrates the annual trends of H and LE in Tanggula area in 2010. Seasonal trends in H and LE are obvious, with H relatively high in spring, reduced during summer, then gradually increasing in autumn. In contrast, LE exhibits the opposite seasonal pattern, and dominates the energy budget in summer and autumn, whereas H is dominant in winter and spring. H is consistent with annual variation of Rn. Rn is small during winter when surface soil is frozen. Therefore, H is smallest. The significant Rn increase in spring increases surface temperature, and the active layer begins to thaw. During this stage, radiation is mainly converted into H, which displays the highest value during the year. The summer monsoon arrival in QTP increases precipitation and hence increases soil moisture content (Fig. 2b), and vegetation begins to grow (Li et al., 2019). This makes the increase in LE particularly obvious in summer. In contrast, H decreases in summer and H changes in autumn are less obvious than for summer.
Seasonal variation trends for LE are more obvious and are more affected by rainfall and shallow soil moisture (compare Figs. 2 and 3), consistent with Wang et al. (2019). The rainy season at Tanggula is coincident with the summer monsoon, which usually starts in May and ends in October (Fig. 2b) (Yang et al., 2000). This period also corresponds to the thawing period of the freeze-thaw cycle of the active layer. As illustrated in Fig. 3b, shallow soil moisture content attained a maximum during this period. LE increases significantly from April due to increasing air temperature, land surface temperature, and soil moisture content, reaching a maximum in July and having the most impact on the surface energy balance. The soil begins to freeze in October to November with decreasing Rn and liquid water content in the topsoil layer. Precipitation also decreases sharply during this period, and hence LE also decreases. Four seasonal freeze-thaw regimes can be divided in the active layer above permafrost, which are the spring warming regime (SW), the summer thawing regime (ST), the autumn freezing regime (AF), and the winter cooling regime (WC) (Zhao et al., 2000; Hinkel et al., 2001). The different freeze-thaw regimes and monthly average values of H, LE, and Bowen ratio are summarized in Table 2. During the SW, the change in H is much larger than LE, i.e., heat exchange is mainly based on sensible heat transport. When the ST arrives, LE increases rapidly as precipitation increases, becoming equivalent to H. Bowen ratio gradually decreases at this stage. LE increases sharply from July to September with the arrival of the QTP rainy season, surpassing H and taking a dominant position. LE subsequently decreases rapidly during the AF, owing to reduced precipitation and surface soil moisture content, and H dominates surface energy budget during the WC.
Months Freeze-thaw regime H LE $ \beta $ 2 SW 44.17 4.83 13.92 3 SW 57.8 10.45 7.19 4 ST 67.92 15.27 7.57 5 ST 65.37 41.32 1.85 6 ST 74.3 61.08 1.26 7 ST 64.62 65.32 0.99 8 ST 63.19 67.38 0.94 9 ST 57.23 53.1 1.12 10 AF 42.52 31.3 1.51 11 AF 27.1 14.05 2.17 12 WC 25.12 7.92 3.69 1 WC 37.48 4.81 11.02 Note: SW, the spring warming regime; ST, the summer thawing regime; AF, the autumn freezing regime; WC, the winter cooling regime. Table 2. Monthly sensible heat flux H (W m−2), latent heat flux LE (W m−2), and Bowen Ratio β for the Tanggula region, 2010.
There is strong interaction between soil hydrothermal processes and surface energy regime throughout the active layer (Yang and Wang, 2019). The most important factors influencing surface energy budget variation from April to May are triggered by freeze-thaw processes (You et al., 2017). Rapidly increasing soil moisture during soil thawing causes a dramatic LE increase. The monsoon arrival over the plateau after May significantly increases soil moisture, and hence LE increases more rapidly to exceed H in the summer (Yao et al., 2020). Next, the active layer begins to freeze by the end of September (Zhao et al., 2000), and decreasing soil moisture suppresses LE increases. Reduced surface radiation reduces soil temperature, resulting in reduced H. Thus, energy flux variations are mainly affected by freeze-thaw processes and the monsoon, and different factors dominate in different seasons (Wang et al., 2019).
Daily trends and fluctuations for G0 exhibit good agreement with those for Rn. Thus, G0 was significantly affected by Rn on a daily scale. Figure 4 shows that the relationship between Rn and G0 in the Tanggula site is strongly linear. Regression relationships between the two other underlying surface types in QTP permafrost regions were estimated previously, as listed in Table 3. Changes in G0 and Rn are very consistent with similar linear correlations for different underlying surfaces in the permafrost region, as in Su (2002). This correlation is more significant on annual and monthly scales than daily scales. Therefore, G0 can be approximated by Rn on annual and monthly scales in the absence of observational data, consistent with Yang et al. (2019b).
Figure 4. Surface heat flux (G0, units: W m−2) and net radiation (Rn, units: W m−2) using 2010 daily average data at Tanggula.
Location Underlying surface type Regression equation Source Tanggula Alpine meadow G0 = 0.16Rn−11.29 This research Wudaoliang Alpine desert G0 = 0.18Rn−11.13 Li et al. (2007) Xidatan Alpine steppe G0 = 0.19Rn−11.63 Xiao et al. (2011) Table 3. Regression equations for different Qinghai-Tibetan Plateau regions.
-
Surface albedo is an important parameter for land–atmosphere energy balance, with small changes in albedo magnitude directly affecting the land surface energy budget (Strugnell and Lucht, 2001). Figure 5a shows that Tanggula area surface albedo was low in summer and high in winter, mainly due to differences in surface cover type (snow in winter and vegetation in summer) and in soil moisture. Although snowfall in QTP was mainly concentrated in the winter, there were still occasional snow events during the plateau monsoon season. Daily mean surface albedo was extreme after snowfalls due to strong solar radiation reflection by ice and snow (Gu et al., 2015; Yao et al., 2020). The impact of the snowfall process on the surface albedo was divided into two cases as follows. One was that there were large amounts of snow remaining for long periods, whereupon surface albedo immediately attained a maximum after snowfall, and then slowly reduced as the snow thawed. The other was a small amount of snowfall, causing the surface albedo to suddenly increase to extremes, which also thawed more quickly, whereupon albedo reduced more sharply forming a sharp peak. Figure 4b shows that for the first case, with snow already covering local surfaces, albedo increased significantly, and the daily average surface albedo could reach a maximum of 0.9; whereas the average surface albedo of Tanggula in 2010 was 0.22.
Figure 5a also shows that even without snow cover, albedo during winter was always higher than during summer, mainly due to two reasons. On the one hand, the soil was relatively dry during winter even with bare vegetation cover, hence generally high albedo; and on the other hand, the large albedo in the cold season is also related to the freezing of the surface soil. Due to the ice particles on the surface, the albedo during the frozen period is large. In contrast, surface vegetation growth and monsoon season during summer meant soil moisture content and vegetation leaf area indexes increased, greatly reducing surface albedo.
Maximum monthly average albedo occurred in November due to heavy snowfalls in mid-October (Fig. 5b) creating a thick snow cover through November, with daily maximum snow depth exceeded 6.5 cm. Snow cover can significantly impact surface energy distribution and radiation balance due to increased albedo and low thermal conductivity (Déry and Brown, 2007). Figure 5b shows that snow cover for the Tanggula site was relatively short for each season. Even during winter, when snowfall events were more frequent, snow cover did not last for the entire cold season, and snow cover changes in most eastern QTP areas had similar characteristics (Robinson et al., 1995; Xu et al., 2017). Snow cover has significant impact on surface energy budget and soil water and heat processes: Rn decreased almost 100%, H decreased after snowfall on day 305 (comparing Figs. 2 and 4), and G0 changed from positive to negative, transporting heat to the ground surface due to the phase change from snow. Rn increased sharply as snow began to melt and H and G0 also gradually increased. Heat to melt snow mainly comes from heat transfer from the soil to the surface. Soil and the surface lose heat during this process, and hence this has an overall cooling effect on the soil. Snow cover has particularly significant soil and atmospheric cooling in the Tanggula region due to short snow accumulation period, which is dominated by ablation (Li et al., 2021). The number of snow cover days in the QTP has consistently decreased, consistent with global warming, further weakening this cooling effect (Flanner et al., 2011).
There is also negative correlation between shallow soil moisture content and surface albedo. Albedo is generally less than 0.3 for the Tanggula area if snow cover effects are removed, hence this study only considered regions where albedo was below 0.3 to eliminate snow effects. Figure 6 shows the relationship between surface albedo and 5 cm soil moisture content at the Tanggula observation site in 2010. Surface albedo decreased with increasing water content in the shallow soil layer. Since the albedo for water is relatively low, higher soil moisture content will tend to reduce surface albedo, which is consistent with practical experience (Zhao and Sheng, 2019).
-
Soil heat flux is also an important component for surface heat balance. Soil absorbs heat and soil temperature rises when surface heat flux is positive; whereas soil gives off heat and soil temperature decreases when it is negative. We introduced the dimensionless soil heat balance coefficient K for convenience (Li et al., 2007),
where
${G}_{\mathrm{s}+}$ is total heat transferred from the land surface to the soil in a year, and$ {G}_{\mathrm{s}-} $ is the total heat transferred upward from soil to the land surface in a year.If K = 1, then heat absorbed by the soil balances with heat released by the soil during the year. K > 1 means the soil absorbs more heat than it releases, and hence soil temperature increases. Permafrost may degenerate if K > 1 for extended periods. If K < 1, heat absorbed by the soil in a year is less than heat released, and soil temperature reduces. Permafrost will develop if K < 1 for extended periods.
Figure 7 shows that K > 1 for the whole study region and all 10 years considered. Average K value is about 1.41 over the study period with maximum in 2010 and minimum in 2014. Thus, the soil was warming overall, leading to unstable frozen ground and tendency for permafrost to degrade. This was consistent with related studies regarding permafrost degradation in the QTP in the context of global warming (Zhang et al., 2012; Shen et al., 2016). Figure 8a shows that active layer thickness changes followed the same trend as for K, significantly increasing over the study period, and Fig. 8b confirms the trend for active layer thicknesses along the Qinghai-Tibetan Highway. Table 4 provides information on observation sites for active layer thickness data along the Qinghai-Tibet Highway.
Figure 8. Yearly mean active layer thickness (units: cm) for (a) the Tanggula study region, and (b) along the Qinghai-Tibetan Highway.
Location Station number Latitude (°N) Longitude (°E) Altitude(m) Kunlun pass CN06 35.62 94.07 4746 Suonandajie CN02 35.43 93.6 4488 Hoh Xil QT01 35.15 93.05 4734 Beiluhe1 QT02 34.82 92.92 4656 Beiluhe2 QT03 34.82 92.92 4656 Fenghuoshan CN01 34.73 92.9 4896 Kaixinling QT05 33.95 92.4 4652 Tongtianhe QT06 33.58 92.24 4650 Tanggula QT04 32.97 91.02 5100 Liangdaohe CN04 31.82 91.73 4808 Table 4. Location of the active layer measurements along the Qinghai-Tibetan highway.
-
Since freeze-thaw processes in the active layer are closely related to surface energy budget and soil thermal energy variations, as discussed above, changes in energy received at the surface eventually cause changes in soil heat. The most intuitive indicator of soil condition is soil temperature changes. With the periodic change of the solar radiation energy received at the surface during the year, freeze-thaw cycles occurred in the active layer (Li et al., 2011). Seasonal thawing process of active layers in permafrost regions was caused by the solar radiative heat reaching the ground and passing into the soil (Li et al., 2013). Generally, the thawing of the active layer in the northern QTP started from April to September. Therefore, this study considered April to September as the thawing period and analyzed surface energy the impacts on thaw depth (TD).
Figure 9 shows TD changes from the Tanggula Comprehensive Observation Field with respect to global radiation Q, Rn shortwave absorption radiation Sn, and ground heat source intensity HIS (Figs. 9a–d, respectively). Surface energy changes had great impact on TD, with active layer TD being minimized when local surface energy accumulation is 0.0 MJ m−2, and increasing with surface energy accumulation. The relationship between the two factors can be described by a power relationship,
Figure 9. Surface energy effects on active layer thawing depth (TD) at the Tanggula site: (a) global radiation Q, (b) net radiation Rn, (c) shortwave absorption radiation Sn, and (d) ground heat source intensity HIS.
where a and b are regression coefficients, and x represents each radiation accumulation.
Table 5 shows regression coefficients for each radiation component and overall correlation coefficient r is greater than 0.98 (p < 0.01). Figure 9 shows that dispersion was very low, and the relationship between energy changes and TD was significant. The process of soil thawing in the active layer is a heat absorption process, where energy and heat accumulation intensifies soil thawing due to the significant relationship between surface energy and thawing depth.
Radiation (MJ m−2) a b r Q 0.008 1.32 0.988 Rn 0.19 1.02 0.993 Sn 0.04 1.15 0.992 HIS 0.21 1.02 0.994 Table 5. Regression coefficients of Eq. (8) for different components.
Observation item | Instruments | Height/depth |
Soil heat flux | HFP01 | 5, 10, 20 cm |
Soil temperature | 107 | 2, 5, 10, 20, 40, 80, 120, 160, 200, 240, 280, 320, 360 cm |
Soil moisture content | CS616 | 5, 10, 20, 40, 70, 105, 140, 175, 210, 245, 280, 300 cm |
Air temperature | HMP-45C | 2, 5, 10 m |
Relative humidity | HMP-45C | 2, 5, 10 m |
Precipitation | T200-B | 1.5 m |
Snow depth | SR-50 | 2 m |
Short wave radiation | CM3 | 2 m |
Long wave radiation | CM3 | 2 m |