Advertisement

Climate Dynamics

, Volume 51, Issue 5–6, pp 2209–2227 | Cite as

Spatiotemporal variations of annual shallow soil temperature on the Tibetan Plateau during 1983–2013

  • Fuxin Zhu
  • Lan Cuo
  • Yongxin Zhang
  • Jing-Jia Luo
  • Dennis P. Lettenmaier
  • Yumei Lin
  • Zhe Liu
Article

Abstract

Soil temperature changes in cold regions can have great impacts on the land surface energy and water balance, and hence changes in weather and climate, surface and subsurface hydrology and ecosystem. We investigate the spatiotemporal variations of annual soil temperature at depths of 0, 5, 10, 15, 20, and 40 cm during 1983–2013 using observations at 85 stations on the Tibetan Plateau (TP). Our results show that the climatological soil temperatures exhibit a similar spatial pattern among different depths and they are generally higher than surface air temperature at the individual stations. Spatially averaged soil temperature show that the TP has experienced significant warming trends at all six depths during 1983–2013, and the soil at 0-cm depth has the fastest warming rate among all the six layers and the surface air temperature. The first leading mode of joint empirical orthogonal function (EOF) analysis exhibits a spatially prevailing warming pattern across the six depths. This plateau-wide soil warming correlates very well with surface air temperature and sea surface temperature in response to increasing radiative forcing caused by greenhouse gases. The joint EOF2 displays a southeastern-northwestern dipole pattern on the TP in the interannual-decadal variability of soil temperature at all layers, which appears to be related to the warm season precipitation and anomalous atmospheric circulations. The spatial difference of soil warming rates across stations on the TP is associated primarily with the spatial distribution of precipitation (mainly rainfall), with vegetation, snowfall and elevation playing a rather limited role.

Keywords

Soil temperature Tibetan Plateau Climate change Joint empirical orthogonal function (joint EOF) 

1 Introduction

Soil temperature is a key state variable that describes both land surface processes and regional environmental and climate conditions. Shallow soil temperature plays an important role in land–atmosphere interactions by modulating the exchanges of energy and water between the atmosphere and the earth’s surface (e.g., Cuo et al. 2015; Gao et al. 2008; Zhan et al. 2014). Soil temperature change is also an excellent indicator of climate change (Zhang et al. 2001), e.g., rising soil temperature in permafrost regions is an important indicator of permafrost degradation (e.g., Peng et al. 2017; Cheng and Wu 2007). Permafrost thaw induced by rising soil temperatures will inevitably alter surface and subsurface hydrological processes (e.g., Yang et al. 2014; Cuo et al. 2015), disturb the fragile ecosystem (e.g., Li et al. 2012), and enhance the rate of global warming by releasing stored carbon from frozen soil into the atmosphere (e.g., Davidson and Janssens 2006; Melillo et al. 2002; McGuire et al. 2016). Furthermore, soil temperature anomalies directly affect the growth and yield of agricultural crops (Hu and Feng 2004a). Clearly, the investigation of soil temperature contributes to the understanding of the changes in regional climate and environmental conditions.

The increase in soil temperatures has been demonstrated to be a prominent phenomenon in the past decades (e.g., Hao et al. 2014; Oelke et al. 2004; Yesilirmak 2014), with increases in anthropogenic greenhouse gases being a major reason (Solomon 2007). Some previous studies have suggested that soil temperatures increase along with rising surface air temperatures (e.g., Cheon et al. 2014; Helama et al. 2011). For instance, using a 27-year soil temperature record at five depths (0.05–1 m) in the Netherlands, Jacobs et al. (2011) found that the annual mean soil temperature showed a statistically significant increase, consistent with the increasing air temperature. Hu and Feng (2005) found that surface air temperatures are correlated with soil temperatures at 153 stations in Eurasia.

However, changes in air temperature alone cannot explain the changes in soil temperatures (Zhang et al. 2001), especially in high-latitude cold regions (Brown and DeGaetano 2011; Jungqvist et al. 2014). For instance, at eight sites in northwestern Canada, Woodbury et al. (2009) found a significant increasing trend in surface air temperature but no obvious upward trend in surface soil temperature. García-Suárez and Butler (2006) suggested that soil temperature variations are related to changes in air temperature but are also influenced by changes in precipitation. Qian et al. (2011) reported that the thermal insulating effect of snow is the major factor masking the increasing trend in annual mean soil temperatures that could have resulted from the widely warming trend in annual mean air temperatures across Canada. Also, changes in the timing and intensity of snowfall and snow density can have significant feedback effects on soil temperature dynamics in cold regions and seasons (e.g., Lawrence and Slater 2010; Ling and Zhang 2006; Sinha and Cherkauer 2008; Zhang 2005a, b). Additionally, Mihalakakou et al. (1997) found that an increase of wind speed led to a reduction of soil temperature. Soil temperature could also be affected by vegetation dynamics (Hu et al. 2009; Tesař et al. 2008).

Most of the previous studies have showed that the soil temperature warming is significant in Arctic and subarctic regions, Siberia, North American boreal area, and other high-latitude cold regions (e.g., Beltrami and Kellman 2003; Helama et al. 2011; Jungqvist et al. 2014; Ling and Zhang 2006; Oelke et al. 2004; Park et al. 2014; Qian et al. 2011; Rankinen et al. 2004; Zhang et al. 2001; Zhang 2005b). However, variations in soil temperature on the Tibetan Plateau (TP), which has the largest cryosphere extent outside the polar region and is also the source region of Asia’s nine major rivers that support about 1.65 billion people downstream (Cuo et al. 2014; Cuo and Zhang 2017), have been less systematically investigated, mainly because soil temperature observation data are generally scarce. Besides, a noteworthy feature is that snow in high-latitude cold regions and snow in the TP play a different role in ground thermal regime, mainly due to the different timing, duration and depth of snow between the two regions (e.g., Zhang 2001). The impact of snow on soil temperatures in the TP may differ from that in the high-latitude snow-dominated regions. Peng et al. (2017) examined the distribution of soil freeze depth (SFD) across China and found distinctive features of SFD on the TP compared to several other parts of China, presumably related to characteristics of soil temperature changes and snow impact on the TP. Therefore, it is important to investigate soil temperature changes on the TP.

The objective of this study is twofold: (1) to illustrate the spatial and temporal variations of annual mean shallow soil temperatures on the TP and (2) to explore the possible mechanisms behind the variations. We place our primary focus on the annual mean soil temperatures out of the considerations that the annual mean time series not only is one of the key climatological variables but also has the advantage of aggregating soil temperature seasonality and accounting for the lagging effects caused by the soil properties. Nevertheless, we also examine the long-term changes of soil temperature in terms of four seasons and warm (growing) season/cold season and contrast the seasonal variations with the annual variations. The structure of this paper is as follows. Section 2 describes the data and methods used in this study. In Sect. 3, we examine the spatiotemporal characteristics of annual mean soil temperatures at six shallow depths, and their relation to surface air temperature, global SST, precipitation and geopotential height. We also discuss the effects of precipitation, vegetation, snow and elevation on soil temperature on the TP in Sect. 3. Section 4 presents our conclusions.

2 Data and methods

2.1 Data and quality control

Observed daily mean soil temperatures at shallow depths of 0, 5, 10, 15, 20, and 40 cm at China Meteorological Administration (CMA) stations were obtained from the Meteorological Bureaus in Qinghai Province (QP) and Tibet Autonomous Region (TAR); the two provinces together occupy the major portion of the TP (Fig. 1a). Soil temperature data were collected at 88 stations where operational observations started in different years at different depths and stations, with the observations at some stations initiated as early as the 1950s. There are missing or abnormal values at some depths and stations, especially in the early period. For this work, soil temperature records were thoroughly checked and quality controlled at daily, monthly, and annual timescales following the methods described in Hu and Feng (2003, 2004b), Hu et al. (2002), and Vickers and Mahrt (1997).

Fig. 1

a Locations and elevations of stations in the area of this study. Red lines delineate the boundary of the Tibetan Plateau in China; black lines delineate the boundary of Qinghai Province (QP) and Tibet Autonomous Region (TAR), China. Black solid dots represent 85 stations that have soil temperature values at 0-cm depth; blue hollow triangles indicate 69 stations with soil temperature observations available at the depth of 5, 10, 15, and 20 cm; and red hollow squares denote 23 stations with soil temperature observations available at 40-cm depth. b Temporal variation of the numbers of stations on the TP at six shallow depths where quality-controlled annual soil temperatures are available

First, the observed daily time series were plotted at each depth for individual stations to visually spot the apparent abnormal and missing values. Obviously incorrect values in daily soil temperature series were identified by setting a threshold (± 50 °C), and these values were treated as missing. Second, to identify the drift in a soil temperature series, we displayed data at each station and compared the variation of the target time series with those at neighboring depths or sites. Then drifts in soil temperatures were corrected using a regression analysis with neighboring depths or sites. Third, monthly mean value of soil temperature at each depth was calculated for a given month and station if (1) the daily data have fewer than ten missing values and missing values are scattered in the given month, and (2) the daily series have less than five consecutive missing values. Otherwise, the monthly mean value was considered as missing. Finally, the annual mean value was calculated from the monthly mean values that must cover all 12 months in a year. Otherwise, the corresponding annual mean soil temperature was considered as missing.

Figure 1b shows the temporal variation of numbers of stations on the TP at six soil layers with quality-controlled annual mean soil temperature. It can be seen that annual mean soil temperature records for the period before the early 1980s are only available at fewer than half of the total number of stations (88). But after 1983, the numbers of the stations with available records at the depth of 0–20 cm exceed 69. Considering the adequate length and continuity of records and the representativeness of spatial coverage, we selected 1983–2013 as the study period. In total, there are 85 stations at soil surface (0 cm), 69 stations at 5, 10, 15, and 20 cm depths, and 23 stations at 40 cm depth where annual mean soil temperature observations are obtained (Fig. 1b). Geographic locations and elevations of the selected stations at various depths are shown in Fig. 1a. Despite the fact that there are only three stations located in the western TAR, our selected stations are relatively representative and adequate to depict the characteristics of soil temperature in the main part of the TP, especially at 0 cm depth (Fig. 1a). The supplementary Table S1 lists the basic information of the selected stations in the study.

Surface air temperature (2 m above the land surface) and precipitation data obtained from the CMA (http://data.cma.cn/site/index.html) were used to explore the relationship between the shallow soil temperature and surface air temperature/precipitation. Air temperature and precipitation data also underwent the quality control using the procedures described above. Snowfall data were derived from precipitation and temperature data by the method described in Cuo et al. (2013). In this study, daily precipitation with daily mean air temperature below 0 °C is treated as water equivalent of snowfall (snowfall, for short). This threshold temperature method is relatively simple but effective, and the threshold (0 °C) is commonly used to separate rainfall and snowfall. National Oceanic and Atmospheric Administration (NOAA) Optimum Interpolation (OI) Sea Surface Temperature (SST) V2 data and geopotential height from NCEP/NCAR Reanalysis data (Kalnay et al. 1996), available at http://www.esrl.noaa.gov/psd/, were used to examine the relationships between soil temperature changes on the TP and changes in global SST and geopotential height. NCEP/NCAR Reanalysis data were used by plenty of previous studies (e.g., Park and Ahn 2016; Ruiz-Barradas et al. 2017; Schlichtholz 2016; Sun et al. 2017). The normalized difference vegetation index (NDVI), a satellite-derived greenness vegetation index from the advanced very high resolution radiometer (AVHRR), was interpolated into the 85 stations by selecting cells covering the stations to explore the relationship between soil temperature and vegetation condition on the TP. AVHRR NDVI data were produced by the global inventory modeling and mapping studies group (GIMMS) at a spatial resolution of 8 km through applying the 15-day maximum-value composition technique to observations by the AVHRR onboard the NOAA satellites. The preprocessing of the NDVI data is described in detail by Shen et al. (2014). For all these datasets, the time coverage is 1983–2013.

2.2 Methods

The Mann–Kendall test for a trend and Sen’s slope estimate (Sen 1968), a nonparametric method, were employed to detect and estimate trend magnitudes in soil temperature and other variables. Empirical orthogonal function (EOF) analysis is a useful method for expressing the spatial and temporal variability of time series data. It was first applied to weather prediction (Lorenz 1956). Joint EOF analysis, a variant of EOF analysis, was performed based on the covariance matrix constructed from annual mean soil temperature at all six soil depths to highlight how the temperature at various depths co-varies with one another (e.g., Luo and Yamagata 2002). In principle, the joint EOF analysis of the soil temperatures at all different depths together is able to extract leading EOF modes that depict the dominant spatial patterns that co-vary among different soil depths for each EOF mode. The results can be used to compare the similarity and difference of soil temperature change among the different depths. In order to examine the impacts of elevation, vegetation and snow on the soil temperature changes on the TP, we adopted a linear regression analysis between the trends in soil temperature and those factors at the corresponding stations. In this study, trends and regression coefficients are considered to be statistically significant by the Student’s t-test when the values reach the confidence level of 95% (p < 0.05).

3 Results and discussions

3.1 Spatiotemporal variation of soil temperature

3.1.1 Climatology of soil temperature

Climatology of the annual mean shallow soil temperatures at six depths (0, 5, 10, 15, 20, and 40 cm) was calculated at individual stations over the 31-year period (1983–2013). The results are shown in Fig. 2 and are also listed in the supplementary Table S1. Figure 2 shows that the spatial patterns of the climatological annual mean soil temperatures on the TP are similar across the different depths, and also resemble that of surface air temperatures. Such a spatial distribution of the climatological soil temperature at each depth on the TP indicates strong dependence on latitude and elevation of the stations. For all the six depths, stations located at higher elevations display relatively lower soil temperatures compared to the median (i.e., 9 °C), whereas stations located in the low-elevation river valleys in the southeastern TAR, Qaidam Basin in the northwestern QP and Huangshui River valley in the northeastern QP exhibit relatively higher soil temperatures (Fig. 2). Also, at similar elevations, stations located in higher latitudes show lower temperatures compared to those in lower latitudes.

Fig. 2

The spatial distribution of climatological annual mean surface air temperature and soil temperatures at the depths of 0, 5, 10, 15, 20, and 40 cm on the TP during 1983–2013. The size of circle denotes the magnitude of the climatological mean values

Across the TP, the climatological annual mean soil temperature ranges from − 0.6 °C at Wudaoliang (52908, 4612.2 m above sea level, ASL) located in the high altitude of the QP interior, to 15.4 °C at Jiacha (56307, 3260 m ASL) located in the low altitudes of the southeastern TAR. Both the maximum and minimum values occur at 0-cm depth. At the 5, 10, 15, and 20 cm depths, both the climatological annual mean soil temperatures and their spatial temperature ranges (i.e., the differences between the maximum and minimum values across the stations) are similar. But the spatial ranges of climatological annual mean soil temperatures at the depths of 5, 10, 15, and 20 cm (from around 1.70–14.50 °C, or 12.80 °C in difference) are smaller than that at 0 cm depth (from − 0.60 °C to 15.40 °C, or 16.00 °C). The spatial range of the soil temperature at 40 cm depth is from 1.93–13.34 °C, smaller than those of the upper depths. The fact, that the spatial ranges of the climatological annual mean soil temperature are gradually damped from the surface to the deeper layers, indicates that the shallow soil temperature changes are mainly caused by atmospheric boundary layer or land surface processes; internal heat sources within the 40-cm depth might be negligible on the TP.

Figure 2 and the supplementary Table S1 also show that the climatological annual soil temperatures at the depth of 0–40 cm are higher than the climatological annual surface air temperatures at the corresponding stations on the TP. To further demonstrate this, we calculate regional means of the climatological soil temperatures at each depths and surface air temperature by averaging over all stations at each depth. The results are listed in Table 1 (marked with ‘Mean’ values). The regionally averaged climatological annual mean soil temperatures at each soil layer are 3–4 °C higher than the corresponding annual mean air temperature on the TP. Similar phenomenon was also found at stations in Russia (Zhang et al. 2001), Canada (Beltrami 2001; Zhang 2005b), The Netherlands (Helama et al. 2011), and South Korea (Cheon et al. 2014).

Table 1

The climatological mean values (Mean) and warming rates (WR) of air and soil temperatures averaged over 85 stations (black dots in Fig. 1b) at 0 cm depth (a), 69 stations (blue hollow triangles in Fig. 1b) at the depths of 0, 5, 10, 15, and 20 cm (b), and 23 stations (red hollow squares in Fig. 1b) at the depths of 0, 5, 10, 15, 20, and 40 cm (c) on the TP during 1983–2013

 

Ta

Ts-0

Ts-5

Ts-10

Ts-15

Ts-20

Ts-40

(a) 85 stations

 Mean (°C)

3.37

7.11

     

 WR (°C/year)

0.0519

0.0718

     

(b) 69 stations

 Mean (°C)

4.27

7.98

7.93

7.96

8.03

7.99

 

 WR (°C/year)

0.0511

0.0711

0.0578

0.0577

0.0524

0.054

 

(c) 23 stations

 Mean (°C)

3.12

6.87

7.15

7.16

7.11

7.12

7.32

 WR (°C/year)

0.0511

0.0761

0.0521

0.0516

0.0532

0.0529

0.0489

All of the linear trends are statistically significant (p < 0.05)

3.1.2 Temporal changes of soil temperature

Figure 3a shows the anomalies of annual mean surface air temperature and soil temperature at 0 cm depth averaged over 85 stations (black dots in Fig. 1a), soil temperature at the depths of 5, 10, 15, and 20 cm over 69 stations (blue hollow triangles in Fig. 1a), and soil temperature at 40 cm depth over 23 stations (red hollow squares in Fig. 1a) on the TP during 1983–2013. The anomalies are computed by removing the long-term (1983–2013) mean temperature at each station and then averaging over all stations on the TP. The inter-annual variations of the spatially averaged surface air temperature and soil temperatures at individual depths are quite similar, showing a significant warming trend superimposed with interannual-decadal fluctuations. A regime shift to a warmer mean-state after 1997/98 when a major El Niño occurred can also be seen in all the time series, consistent with changes in the Northern Hemisphere mean surface air temperature (e.g., Luo et al. 2011). Also, there is slowing down of the warming trend after 2005 in Fig. 3a, which is about 5–6 years delayed compared to the global warming hiatus after late 1990s (e.g., Easterling and Wehner 2009; Trenberth and Fasullo 2013). Figure 3b shows the anomalies of spring (March–May), summer (June–August), autumn (September–November), winter (December–February), and annual mean soil temperature at 0-cm depth averaged over the 85 stations on the TP. Clearly, both the seasonal mean and the annual mean soil temperatures exhibit significant and similar warming trends during 1983–2013, although some differences in peaks and troughs related to inter-annual variability are noted. Figure 3b suggests that the temporal changes in the annual mean soil temperature closely reflect the temporal changes in the seasonal mean soil temperature and therefore it is suitable to use annual mean soil temperature to explore the long-term changes in soil temperature.

Fig. 3

a Time series of the anomalies of the annual mean soil temperature at the surface layer (0 cm) and surface air temperature averaged over 85 stations, 69 stations at the depths of 5, 10, 15, and 20 cm, and 23 stations at 40 cm depth on the TP during 1983–2013. b Anomalies of the annual, spring, summer, autumn, and winter mean soil temperature at 0-cm depth averaged over 85 stations on the TP

We calculate the spatial averages of the annual mean climatological surface air temperature and soil temperatures as well as their long-term trends over 85 stations at 0 cm depth (Table 1a), over 69 stations for each of the upper five depths (i.e., 0, 5, 10, 15 and 20 cm, Table 1b), and over 23 stations for each of the six depths on the TP (i.e., 0, 5, 10, 15, 20 and 40 cm, Table 1c), respectively. Although the spatially averaged climatological annual mean temperatures exhibit some differences because of the different number of stations, there are much small differences in the spatially averaged warming rates among the results based on 85 stations, 69 stations and 23 stations for each depth (Table 1). This suggests that our results based on the maximum number of data-available stations at different depths adopted in this study are robust for long-term change analysis. The warming rates of the annual mean soil temperature during 1983–2013 gradually decrease from 0 to 40 cm depth, with statistically significant rates of 0.0718, 0.0578, 0.0577, 0.0524, 0.0540, and 0.0489 °C/year at 0, 5, 10, 15, 20, and 40 cm depth, respectively (Table 1). These warming trends in the annual mean soil temperatures are consistent with the results of Gao and Wu (2015), who showed that soil temperatures at stations laid out along the Qinghai-Xizang Highway have increased during 1996–2012.

It is worth noting that the warming rates during 1983–2013 of the spatially averaged soil temperature at depths from 0 to 20 cm are larger than that of surface air temperature when averaged over the same number of stations. This is due to the fact that the air temperature cools (− 0.048, − 0.049, and − 0.068 °C/year for the averages of 85, 69 and 23 stations, respectively) more than the soil temperature does (> around − 0.03 °C/year) during the period after 2005. Note that, before 2005, the warming trend in the air temperature is comparable to that of surface soil temperature, but is higher than that of the temperature at deeper layers (Fig. 3a). Obviously, the greater cooling trend in the air temperature after 2005 reduces the air temperature warming during the entire period of 1983–2013.

The greater warming in the shallow soil temperature than that in the air temperature is largely consistent with the finding reported by Zhang et al. (2007). Similarly, Hu and Feng (2003) also found that in the North America soil temperature at 10 cm depth has warmed at an average rate of 0.031 °C/year during 1967–2002, greater than the average rate of 0.010 °C/year for surface air temperature. But in Canada, due to the effect of snow cover, the annual soil temperature at 20 cm depth increased less than the annual mean air temperature did during 1901–1995 (e.g., Zhang 2005b). However, in our case, the 23-station averaged soil temperature at 40 cm depth, where the influence from atmospheric climate change becomes weaker compared to that at the upper soil layers, has warmed less compared to the surface air temperature during 1983–2013 on the TP.

Figure 4 shows the spatial distribution of the trends of the annual mean surface air temperature and shallow soil temperature at the six depths (i.e., 0, 5, 10, 15, 20 and 40 cm) at individual stations on the TP during 1983–2013. Almost all of the stations exhibit statistically significant (p < 0.05) warming trends at all the six depths, except for Linzhi (56312) at 0, 5, 10, 15, 20, and 40 cm depths, Xining (52866) and Lenghu (52602) at 5, 10, 15, 20, and 40 cm depths, and Nanmulin (55572) at 0 cm depth where statistically non-significant weak cooling trends are displayed. The spatial distributions of the soil temperature trends at the six depths are generally similar to one another, and resemble that of the surface air temperature trends. But the warming rates of the annual mean soil temperatures at the top 20-cm depths are greater than those of surface air temperatures at most of stations on the TP as seen from the regional averages listed in Table 1.

Fig. 4

The spatial distribution of Mann–Kendall trends of annual mean air temperature and soil temperatures at the depths of 0, 5, 10, 15, 20, and 40 cm on the TP during 1983–2013. The size of triangle denotes the magnitude of trend. Solid triangles represent statistically significant trends (p < 0.05), and the hollow ones represent non-significant trends. Similar symbols are used hereinafter

The magnitude of the trends of the annual mean soil temperature varies considerably across the region. The soil temperatures have increased rapidly during 1983–2013 in the northern and western TP, the arid and semi-arid regions with less precipitation, where the warming rates of the soil temperature at some stations are greater than 0.100 °C/year. For example, Tongde (52957) in the northeastern QP shows the most rapid soil warming at 0, 15, and 20 cm depths, with the warming rates of 0.140, 0.112 and 0.121 °C/year, respectively. The greatest soil warming rates at both 10 and 40 cm depths appear at Shiquanhe (55228) in the western TAR, with the warming rates of 0.113 and 0.138 °C/year, respectively. At 5 cm depth, the greatest soil warming rate is 0.113 °C/year at Xiaozaohuo (52707) in the northwestern QP. On the contrary, in the relatively humid southeastern TP, the warming of the soil temperatures is modest, and the weak warming rates at some stations even fail to pass the significance test (p < 0.05). We have also examined the spatial distributions of Mann–Kendall trends of warm (May–September) and cold (October–April) season mean soil temperatures at the six depths (not shown). Compared to the annual means, both warm and cold season mean soil temperatures also present predominantly warming trends across the TP, with generally larger warming in the northwestern than the southeastern TP, suggesting that the annual mean soil temperature trends are indicative of the seasonal trends.

In order to better understand the soil temperature changes, we performed a joint EOF analysis to examine the co-varying spatial patterns of the soil temperature at the depths of 0, 5, 10, 15, 20, and 40 cm on the TP during 1983–2013. Figure 5 shows the spatial pattern and principal component of the first joint EOF mode of the annual mean soil temperature at the six depths (0–40 cm), accounting for 70.1% of the total variance. There is a similar spatial pattern among different depths, showing uniform positive signals over almost all stations at individual depths, with strong positive signals located in the northern and western TP and weak positive signals appearing in the southern TP. The strong signals at most stations and weak and negligible signals at several stations almost occur across all the soil depths. For instance, Tongde (52957), Shiquanhe (55228), and Xiaozaohuo (52707) have strong positive signals across all the six depths. And Linzhi (56312), Lenghu (52602) and Nanmulin (55572) have weak signals across all the depths. The spatial pattern of the first joint EOF mode is in good agreement with that of the Mann-Kendall trends of the soil temperatures at each depth (Figs. 4, 5). The principal component (PC1) of the first joint EOF mode of the soil temperature shows a striking upward trend (Fig. 5, bottom panel), with the temporal variation being quite similar to those of the spatially averaged soil temperature on the TP at the different depths (Fig. 3a). Therefore, the first joint EOF mode mostly represents the plateau-wide warming trend in the soil temperature on the TP, notwithstanding that the magnitude of the soil warming rates varies with space.

Fig. 5

Spatial pattern (top panels) and principal component (bottom panel) of the first joint EOF mode of the annual mean soil temperature at six depths (0–40 cm), accounting for 70.1% of the total variance. The black dots in top six panels denote station locations

Figure 6 shows the spatial pattern of the second joint EOF mode of the soil temperature at all the six depths and the corresponding principal component (PC2). This mode explains 8.5% of the total variance. In contrast to the monopole pattern of the first joint EOF mode, the second joint EOF mode reveals a remarkable dipole pattern with an out-of-phase variation between the northwestern and southeastern TP at all the six depths. In addition, the PC2 shows a noticeable decadal variability without a noticeable linear trend. In general, the second joint EOF mode of the soil temperature delineates the northwestern and southeastern TP spatial pattern difference in the soil temperature variations and reveals soil temperature inter-annual to decadal variability that is apparently not related to global warming.

Fig. 6

Spatial pattern (top panels) and principal component (bottom panel) of the second joint EOF mode of the annual mean soil temperature at six depths (0–40 cm), accounting for 8.5% of the total variance. The black dots in top six panels denote station locations

3.2 Possible factors influencing temporal and spatial soil temperature change

3.2.1 Long-term trend of soil temperture change

Figure 7a shows the regression coefficients of the annual surface air temperature at 85 stations onto the joint EOF PC1 of annual mean soil temperature at the six depths on the TP. The regression coefficients are positive at all 85 stations on the TP. Significant positive regression coefficients across all stations with a spatial pattern that resembles that of the joint EOF1 of the soil temperatures indicate that changes in the soil temperature at 0–40 cm depth on the TP are closely related to those in surface air temperature at each station. Such a good in-phase relation between the shallow soil temperature and surface air temperature change at most of the stations on the TP is distinct from that in the snow-dominated Arctic and subarctic regions, where the insulation effect of snow cover is significant (e.g., Qian et al. 2011; Zhang 2005a).

Fig. 7

Regression coefficients of the surface air temperature at 85 stations (a) and global annual SST anomalies (b) onto the PC1 of joint EOF analysis for annual soil temperature at six depths on the TP. Regression coefficients of the warm season precipitation (c) and geopotential height anomalies (d) onto the PC2 of joint EOF for annual soil temperature at six depths on the TP. The crosses in all panels represent the statistically significant regression coefficients (p < 0.05)

Change in global SST is a good indicator of the global climate change. Here we calculate the regression coefficients of the global annual mean SST anomalies during 1983–2013 onto the joint EOF PC1 of annual mean soil temperature at the six depths on the TP. The spatial pattern of the regression coefficients of annual mean SST anomalies onto the PC1 of the soil temperature on the TP (shown in Fig. 7b) resembles that of trends in SST during 1982–2006 identified by Barbosa and Andersen (2009). Figure 7b shows that the majority of global ocean areas display statistically significant positive regression coefficients (i.e., SST warming), especially in the tropical Indian Ocean, Western Pacific, and North Atlantic. In the eastern tropical Pacific, a La Niña-like cooling pattern appears. Patchy cooling also appears in the southern ocean near Antarctic. The global SST warming over the past 2–3 decades have been extensively studied and mainly ascribed to internal variability and external radiative forcing due to the increased greenhouse gases (e.g., Kosaka and Xie 2013; Luo et al. 2012). Such positive relationships between the soil temperature and the air temperature on the TP and the SST over the major parts of global oceans suggest that the annual mean soil warming on the TP is yet another piece of evidence of the regional manifestation of global warming due to the increased emissions of greenhouse gases (e.g., Portmann et al. 2009; Zhou et al. 2009).

3.2.2 Dipole spatial pattern of soil temperature variability

In order to explore possible reasons for the out-of-phase interannual-decadal variability of the soil temperature between the northwestern and southeastern TP, we calculate the regression coefficients of the warm season (May–September) precipitation (mainly rainfall) anomalies at 85 stations onto the joint EOF PC2 of annual mean soil temperature at the six depths on the TP (shown in Fig. 7c). The regression coefficient spatial pattern is similar to that of the second joint EOF2 mode of the soil temperature but with an opposite polarity (Fig. 6). The negative regression coefficients of the warm season precipitation onto the PC2 are mainly located at stations in the northern and western TP, where the signals of the second joint EOF mode of the soil temperature are positive. On the contrary, in the southeastern TP, there are prevailing positive regression coefficients, where the signals of the joint EOF2 of soil temperature are negative (cf., Figs. 6, 7c). Such an opposite relationship between the soil temperature and warm season precipitation suggests that the out-of-phase interannual-decadal variations of the soil temperature between the northwestern and southeastern TP may be associated with warm season precipitation variations on the TP. We also calculate the regression coefficients of the cold season (October–April) and annual precipitation anomalies at 85 stations onto the joint EOF PC2 of annual mean soil temperature at the six depths on the TP (not shown). There is no significant relationship between cold season precipitation anomalies and the second mode of the joint EOF for annual soil temperature. But the relationship between annual precipitation anomalies and the second mode of the joint EOF for annual soil temperature is quite similar to that of warm season, likely because annual precipitation changes are mainly dominated by warm season precipitation changes on the TP.

Figure 7d shows the regression coefficients of 300-mb geopotential height anomalies onto the joint EOF PC2 of the annual mean soil temperature at the six depths on the TP. A dipole pattern with negative coefficients over the southeastern TP and positive coefficients over the northwestern TP is evident (Fig. 7d). This dipole pattern can be detected at 200–400 mb as well (not shown). This spatial pattern of the regression coefficients closely matches that of the joint EOF2 for soil temperature, although most of the regression coefficients fail to pass the significance test (p < 0.05) likely because of limited time period. Good in-phase relation between the shallow soil temperature and the geopotential height on the TP suggests that the difference in soil temperature variations between the northwestern and southeastern TP may be linked to atmospheric circulations although the exact mechanisms warrant future studies. The regressions between EOF PC2 of annual soil temperature and the NDVI and snowfall at 85 stations fail to match the dipole pattern (figures not shown), indicating that the NDVI and snowfall are not related to the dipole pattern of soil temperature interannual-decadal variability.

3.2.3 Spatial difference of soil warming trends

Soil temperature is a measure of soil thermal state that is determined mainly by the energy budget on the land surface (Chen et al. 2003; Hillel 2013; Lal and Shukla 2004; Qi et al. 2016). The shallow soil temperature variations also integrate the influence of almost all of the meteorological variables and processes occurring at and above the ground and in the soil (Subin et al. 2013). As important factors that affect the energy and water exchanges between the atmosphere and the land surface, precipitation, vegetation and snow conditions and topography could exert effects on the soil temperature variations (Helama et al. 2011; Lawrence and Slater 2010; Tesař et al. 2008; Zhang 2005a). Here, we explore the effects of precipitation, vegetation, snowfall and elevation on the spatial pattern of soil temperature trends on the TP.

Figure 8a shows that the climatological warm season (May–September) precipitation decreases from the southeast TP to the northwest TP, ranging from about 600 mm to less than 20 mm. Precipitation increases at most stations during 1983–2013, but only a few stations exhibiting statistically significant trends (Fig. 8b). Figure 9 presents the scatter plot of the climatological precipitation and the soil warming trends during the warm season of 1983–2013 at each of the six depths among all stations on the TP. This figure shows that regression coefficients are negative and are statistically significant at all layers except 40 cm, indicating that higher precipitation causes lower soil warming at the shallow soil layers. However, none of the regression coefficients of precipitation trends and soil temperature trends during the warm season of 1983–2013 at all six layers among the stations on the TP passes the significance test (not shown). As there are 76 out of 85 stations where the warm season precipitation amount is more than 80% of the annual precipitation amount on the TP, the spatial regression coefficients of annual mean soil temperature trends and annual climatological precipitation (not shown) are similar to those based on the warm season climatological precipitation (recall Fig. 9). The above results suggest that the spatial pattern of climatological precipitation, instead of precipitation change, can explain the spatial difference of soil warming trends between the southeastern and northwestern TP.

Fig. 8

The spatial distribution of the climatology and trends of warm season precipitation (a, b), growing season NDVI (c, d), and cold season snowfall (e, f) during 1983–2013 on the TP. The color of circle in panels (a, c, e) denotes the magnitude of the climatological mean values. Solid triangles represent statistically significant trends (p < 0.05), and the hollow ones represent non-significant trends

Fig. 9

Scatter plot of the trends in annual mean soil temperature and precipitation at the corresponding stations for each shallow soil depth on the TP during 1983–2013. Each circle represents a station (the same for the following figures). Solid circles indicate the stations where the trends of annual soil temperatures are statistically significant, and hollow ones indicate non-significant trends. Red line represents the best linear fit to the station data with the slope (b) and p-value (p) indicated in each panel. Note that the red solid line indicates a significant regression. Similar symbols are used hereinafter

The negative coefficients could be attributed to the evaporation-cooling effect of precipitation on soil. For instance, a higher climatological precipitation can enhance the surface evaporation that takes more heat away from ground, resulting in a reduced soil warming rate. In addition, soil temperature changes also depend on the soil specific heat capacity. Given equal net energy flux, the warming rate is greater for soil with lower heat capacity than that for soil with higher heat capacity. Hanks (2012) reported that variations in soil heat capacity can be primarily explained by changes in soil water content. As soil water content increases, soil heat capacity increases as well. Therefore, the climatological wet conditions (high precipitation) and the associated high heat capacity of soil over the humid and semi-humid region of the southeastern TP would slow down the warming rates of soil temperatures there.

The mean NDVI during the growing (warm) season from May to September is considered as the indicator of vegetation conditions on the TP (e.g., Shen et al. 2014). The climatology and trends of the NDVI during growing season of 1983–2013 at 85 stations on the TP are shown in Fig. 8c, d, respectively. In general, the climatological growing season NDVI decreases from the eastern TP to the western TP (Fig. 8c). Concerning the long-term trends of the growing season NDVI, 53 (including 15 stations where the trends pass the significance test) out of the 85 stations have positive NDVI trends. The growing season NDVI at the remaining 32 stations has a negative trend, but the trends at only 5 out of the 32 stations pass the significance test. In order to explore the relationship between the variations in the shallow soil temperature and the vegetation on the TP, scatter plots of soil temperature trends at the depths of 0, 5, 10, 15, 20, and 40 cm and the climatological NDVI at the corresponding stations during the growing season of 1983–2013 are shown in Fig. 10. This figure shows weak negative regression coefficients between the soil warming at individual depths and the climatological NDVI. Although the regression coefficients fail to pass the significance test (p < 0.05), the negative coefficients between the soil temperature trends and the climatological NDIV during the growing season imply that as the climatological NDVI increases, the soil warming rates slightly decrease across the TP. Note that the NDVI trends also have non-significant negative regression coefficients with the soil temperature trends during the growing season at all six layers (not shown). These weak regression coefficients imply that vegetation conditions slightly negatively affect soil temperature changes during the growing season of 1983–2013 on the TP.

Fig. 10

As in Fig. 9, but for the relationship between the trends in the soil temperature at the six depths and the climatological annual NDVI at individual stations on the TP during 1983–2013

Vegetation variation exerts influence on climate change through changing albedo and evapotranspiration (e.g., Chapin et al. 2005). On one hand, previous studies have shown that the increased vegetation conditions can accelerate climate warming through reduced albedo, which increases the amount of solar radiation absorbed by the land surface (Chapin et al. 2000; Pearson et al. 2013). On the other hand, the increased vegetation activity can attenuate climate warming by enhancing evapotranspiration, acting as a cooling process (Shen et al. 2015). The generally weak negative regression coefficients revealed in Fig. 10 between the soil temperature trends and the NDVI suggest that the cooling effect of vegetation conditions during the growing season may be slightly more important than its warming effect on the soil temperature changes on the TP. Overall, the net impact of the vegetation conditions on the soil temperature changes is negative but weak on the TP.

In high-latitude cold regions, snow conditions have an important effect on soil temperature changes (e.g., Subin et al. 2013; Zhang 2005a). On one hand, snow surfaces have a high albedo that can lead to a reduction in absorbed solar radiation, thereby decreasing soil temperature. On the other hand, snow has an extremely low thermal conductivity and thus acts as an excellent insulator between the atmosphere and the ground, thereby protecting soil from heat loss in cold season. In addition, more snowfall usually corresponds to higher evapotranspiration and hence more soil evaporative cooling, which would retard the soil warming in response to the increasing greenhouse forcing. The spatial distribution of climatological snowfall and the trends of snowfall during the cold season (January–April and October–December) of 1983–2013 at 85 stations on the TP are shown in Fig. 8e, f, respectively. The snowfall ranges from less than 1 mm at Chayu (56434, 2327.6 m ASL) located in the southeastern TP to 95.2 mm at Jiali (56202, 4488.8 m ASL) located in the alpine region. There are 76 out of the 85 stations where the snowfall is less than 50 mm (Fig. 8e). Due to the dominance of moisture depleted westerlies in the cold season on the TP, the snowfall on the TP is generally smaller than that in the high-latitude cold regions (e.g., Zhang 2001; Zhang et al. 2001; Zhao et al. 2004). The long-term trends of snowfall are negative at 53 out of the 85 stations, including 7 stations where the negative snowfall trends are statistically significant (Fig. 8f). The other 32 stations show a positive snowfall trend, but only 2 out of the 32 stations show a significant (p < 0.05) trend.

Regression coefficients and scatter plots of the soil temperature trends at the depths of 0, 5, 10, 15, 20, and 40 cm and the climatological snowfall at the corresponding stations during the cold season of 1983–2013 are shown in Fig. 11. The positive regression coefficients between the soil temperature trends at 0–20 cm depths and the climatological snowfall during the cold season indicate that the shallow soil temperature trends generally increase with the increase in the climatological snowfall across the TP. However, only surface layer displays a statistically significant positive regression coefficient and the regression coefficient of the climatological snowfall with the soil temperature trends at 40-cm depth is slightly negative, suggesting a rather limited impact of the snowfall on the soil temperature change across the TP. The snowfall trends exhibit negative regression relationships with the soil warming rates at 0–20 cm depths and a positive relationship with those at 40-cm depth. But none of these regression coefficients are statistically significant (not shown). Compared with the effect of snow on the soil temperature changes in the snowy northern high latitudes (Qian et al. 2011; Zhang et al. 2001), the impact of snow on the soil temperature trends on the TP appears to be weaker. The net effect of snow on the soil temperature change is determined collectively by the timing, duration, depth, density, structure, accumulation and melting processes of snow, and its interactions with micrometeorological conditions, local micro-relief, vegetation, and the geographical locations (e.g., Jafarov et al. 2014; Lawrence and Slater 2010; Zhang 2005b). As described in Zhang (2001), the difference in the timing may be the major cause of the difference in the snow effect on the soil temperature changes between the TP and the northern high latitudes.

Fig. 11

As in Fig. 9, but for the relationship between the soil temperature trends at the six depths and the climatological annual snowfall at individual stations on the TP during 1983–2013

Previous studies have debated the possible dependence of surface air temperature warming on the elevation on the TP. Some studies (e.g., Liu and Chen 2000; Qin et al. 2009; Wei and Fang 2013) reported a clear positive relation between the surface air temperature trends and station elevations, and this relation hence affects the spatial distribution of the surface air temperature warming. Whereas, other studies (e.g., Cuo et al. 2013; You et al. 2010) argued that there is insignificant or no clear relationship between the warming trends and the station elevation. The discrepancy might be due to different datasets and time periods that are analyzed in different studies (Pepin et al. 2015). As far as we know, no result has been reported on the relationship between the soil temperature changes and the station elevations on the TP. To fill the gap, we investigate the relationship between the annual mean soil temperature trends at depths of 0, 5, 10, 15, 20, and 40 cm and the elevations at individual stations. Figure 12 shows that except for a statistically significant positive linear regression coefficient between the 15-cm-depth soil temperature trends and the station elevation, the positive regression coefficients between the soil temperature trends at other depths and the station elevation are not statistically significant (p < 0.05). In summary, vegetation, snowfall and elevation have limited effect on the spatial distribution of the soil temperature warming on the TP. In contrast, precipitation has a significant effect on the spatial distribution of the soil temperature warming on the TP.

Fig. 12

As in Fig. 9, but for the relationship between the soil temperature trends at the six depths and the elevation at the corresponding stations on the TP during 1983–2013

4 Conclusions

We examined the spatiotemporal variations of annual mean soil temperature at the depth of 0, 5, 10, 15, 20, and 40 cm during 1983–2013 based on station observations on the TP. The results of our study are as follows:

There exists a largely similar spatial pattern of the climatological annual mean soil temperatures at each depth to that of surface air temperature, with generally higher (lower) temperature in lower (higher) elevations and latitudes on the TP.

Like surface air temperature on the TP and SST on the majority of ocean surfaces, most stations at all six depths on the TP have experienced significant warming trends in the annual mean soil temperature during 1983–2013 due to radiative forcing induced by increased greenhouse gases.

A southeast-northwest dipole of the interannual-decadal soil temperature variability exists for all soil layers on the TP. The dipole could be related to the warm season precipitation and atmospheric circulations that control the TP based on the analysis of the geopotential heights.

The spatial difference of the soil warming rates across stations on the TP is associated primarily with the spatial distribution of climatological precipitation (mainly rainfall). And it is also, to a lesser extent, affected by growing season NDVI, cold season snowfall and elevation on the TP.

Notes

Acknowledgements

This study is supported by the National Natural Science Foundation of China (Grant 41571067) and the International Partnership Program of Chinese Academy of Sciences (Grant 131C11KYSB20160061), and National Basic Research Program (Grant 2013CB956004).

Supplementary material

382_2017_4008_MOESM1_ESM.docx (34 kb)
Supplementary material 1 (DOCX 34 KB)

References

  1. Barbosa SM, Andersen OB (2009) Trend patterns in global sea surface temperature. Int J Climatol 29:2049–2055.  https://doi.org/10.1002/joc.1855 CrossRefGoogle Scholar
  2. Beltrami H (2001) On the relationship between ground temperature histories and meteorological records: a report on the Pomquet station. Glob Planet Change 29:327–348.  https://doi.org/10.1016/s0921-8181(01)00098-4 CrossRefGoogle Scholar
  3. Beltrami H, Kellman L (2003) An examination of short- and long-term air-ground temperature coupling. Glob Planet Change 38:291–303.  https://doi.org/10.1016/s0921-8181(03)00112-7 CrossRefGoogle Scholar
  4. Brown PJ, DeGaetano AT (2011) A paradox of cooling winter soil surface temperatures in a warming northeastern United States. Agric For Meteorol 151:947–956.  https://doi.org/10.1016/j.agrformet.2011.02.014 CrossRefGoogle Scholar
  5. Chapin FS, Eugster W, McFadden JP, Lynch AH, Walker DA (2000) Summer differences among Arctic ecosystems in regional climate forcing. J Clim 13:2002–2010. https://doi.org/10.1175/1520-0442(2000)013<2002:sdaaei>2.0.co;2CrossRefGoogle Scholar
  6. Chapin FS et al (2005) Role of land-surface changes in Arctic summer warming. Science 310:657–660.  https://doi.org/10.1126/science.1117368 CrossRefGoogle Scholar
  7. Chen B, Chao WC, Liu X (2003) Enhanced climatic warming in the Tibetan Plateau due to doubling CO2: a model study. Clim Dyn 20:401–413.  https://doi.org/10.1007/s00382-002-0282-4 CrossRefGoogle Scholar
  8. Cheng G, Wu T (2007) Responses of permafrost to climate change and their environmental significance, Qinghai-Tibet Plateau. J Geophys Res Earth Surf.  https://doi.org/10.1029/2006jf000631 Google Scholar
  9. Cheon J-Y, Ham B-S, Lee J-Y, Park Y, Lee K-K (2014) Soil temperatures in four metropolitan cities of Korea from 1960 to 2010: implications for climate change and urban heat. Environ Earth Sci 71:5215–5230.  https://doi.org/10.1007/s12665-013-2924-8 CrossRefGoogle Scholar
  10. Cuo L, Zhang YX (2017) Spatial patterns of wet season precipitation vertical gradients on the Tibetan Plateau and the surroundings. Sci Rep.  https://doi.org/10.1038/s41598-017-05345-6 Google Scholar
  11. Cuo L, Zhang Y, Wang Q, Zhang L, Zhou B, Hao Z, Su F (2013) Climate change on the Northern Tibetan Plateau during 1957–2009: spatial patterns and possible mechanisms. J Clim 26:85–109.  https://doi.org/10.1175/jcli-d-11-00738.1 CrossRefGoogle Scholar
  12. Cuo L, Zhang Y, Zhu F, Liang L (2014) Characteristics and changes of streamflow on the Tibetan Plateau: a review. J Hydrol Reg Stud 2:49–68CrossRefGoogle Scholar
  13. Cuo L, Zhang Y, Bohn TJ, Zhao L, Li J, Liu Q, Zhou B (2015) Frozen soil degradation and its effects on surface hydrology in the northern Tibetan Plateau. J Geophys Res Atmos 120:8276–8298.  https://doi.org/10.1002/2015jd023193 CrossRefGoogle Scholar
  14. Davidson EA, Janssens IA (2006) Temperature sensitivity of soil carbon decomposition and feedbacks to climate change. Nature 440:165–173.  https://doi.org/10.1038/nature04514 CrossRefGoogle Scholar
  15. Easterling DR, Wehner MF (2009) Is the climate warming or cooling? Geophys Res Lett.  https://doi.org/10.1029/2009gl037810 Google Scholar
  16. Gao S, Wu Q (2015) Period analysis and trend forecast for soil temperature in the Qinghai-Xizang highway by wavelet transformation. Environ Earth Sci 74:2883–2891.  https://doi.org/10.1007/s12665-015-4313-y CrossRefGoogle Scholar
  17. Gao Z, Horton R, Wang L, Liu H, Wen J (2008) An improved force-restore method for soil temperature prediction. Eur J Soil Sci 59:972–981.  https://doi.org/10.1111/j.1365-2389.2008.01060.x CrossRefGoogle Scholar
  18. García-Suárez AM, Butler CJ (2006) Soil temperatures at Armagh observatory, Northern Ireland, from 1904 to 2002. Int J Climatol 26:1075–1089.  https://doi.org/10.1002/joc.1294 CrossRefGoogle Scholar
  19. Hanks RJ (2012) Applied soil physics: soil water and temperature applications, vol 8. Springer, New York, NYGoogle Scholar
  20. Hao G, Zhuang Q, Pan J, Jin Z, Zhu X, Liu S (2014) Soil thermal dynamics of terrestrial ecosystems of the conterminous United States from 1948 to 2008: an analysis with a process-based soil physical model and AmeriFlux data. Clim Change 126:135–150.  https://doi.org/10.1007/s10584-014-1196-y CrossRefGoogle Scholar
  21. Helama S, Tuomenvirta H, Venalainen A (2011) Boreal and subarctic soils under climatic change. Glob Planet Change 79:37–47.  https://doi.org/10.1016/j.gloplacha.2011.08.001 CrossRefGoogle Scholar
  22. Hillel D (2013) Fundamentals of soil physics. Academic press, New York, NYGoogle Scholar
  23. Hu Q, Feng S (2003) A daily soil temperature dataset and soil temperature climatology of the contiguous United States. J Appl Meteorol 42:1139–1156. https://doi.org/10.1175/1520-0450(2003)042<1139:adstda>2.0.co;2CrossRefGoogle Scholar
  24. Hu Q, Feng S (2004a) GROUNDWORK: US soil temperature and its variation: a new dataset. Bull Am Meteor Soc 85:29–31.  https://doi.org/10.1175/bams-85-1-29 CrossRefGoogle Scholar
  25. Hu Q, Feng S (2004b) A Role of the Soil Enthalpy in Land Memory*. J Clim 17:3633–3643CrossRefGoogle Scholar
  26. Hu Q, Feng S (2005) How have soil temperatures been affected by the surface temperature and precipitation in the Eurasian continent? Geophys Res Lett.  https://doi.org/10.1029/2005gl023469 Google Scholar
  27. Hu Q, Feng S, Schaefer G (2002) Quality control for USDA NRCS SM-ST network soil temperatures: a method and a dataset. J Appl Meteorol 41:607–619. https://doi.org/10.1175/1520-0450(2002)041<0607:qcfuns>2.0.co;2CrossRefGoogle Scholar
  28. Hu H et al (2009) Influences of alpine ecosystem degradation on soil temperature in the freezing-thawing process on Qinghai-Tibet Plateau. Environ Geol 57:1391–1397.  https://doi.org/10.1007/s00254-008-1417-7 CrossRefGoogle Scholar
  29. Jacobs AFG, Heusinkveld BG, Holtslag AAM (2011) Long-term record and analysis of soil temperatures and soil heat fluxes in a grassland area, The Netherlands. Agric For Meteorol 151:774–780.  https://doi.org/10.1016/j.agrformet.2011.01.002 CrossRefGoogle Scholar
  30. Jafarov EE, Nicolsky DJ, Romanovsky VE, Walsh JE, Panda SK, Serreze MC (2014) The effect of snow: how to better model ground surface temperatures. Cold Reg Sci Technol 102:63–77.  https://doi.org/10.1016/j.coldregions.2014.02.007 CrossRefGoogle Scholar
  31. Jungqvist G, Oni SK, Teutschbein C, Futter MN (2014) Effect of climate change on soil temperature in Swedish boreal forests. Plos One.  https://doi.org/10.1371/journal.pone.0093957 Google Scholar
  32. Kalnay E et al. (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77:437–471. https://doi.org/10.1175/1520-0477(1996)077<0437:tnyrp>2.0.co;2CrossRefGoogle Scholar
  33. Kosaka Y, Xie SP (2013) Recent global-warming hiatus tied to equatorial Pacific surface cooling. Nature 501:403.  https://doi.org/10.1038/nature12534 CrossRefGoogle Scholar
  34. Lal R, Shukla MK (2004) Principles of soil physics. Marcel Dekker Inc., New York, NYGoogle Scholar
  35. Lawrence DM, Slater AG (2010) The contribution of snow condition trends to future ground climate. Clim Dyn 34:969–981.  https://doi.org/10.1007/s00382-009-0537-4 CrossRefGoogle Scholar
  36. Li X, Jin R, Pan X, Zhang T, Guo J (2012) Changes in the near-surface soil freeze–thaw cycle on the Qinghai-Tibetan Plateau. Int J Appl Earth Obs Geoinf 17:33–42CrossRefGoogle Scholar
  37. Ling F, Zhang TJ (2006) Sensitivity of ground thermal regime and surface energy fluxes to tundra snow density in northern Alaska. Cold Reg Sci Technol 44:121–130.  https://doi.org/10.1016/j.coldregions.2005.09.002 CrossRefGoogle Scholar
  38. Liu XD, Chen BD (2000) Climatic warming in the Tibetan Plateau during recent decades. Int J Clim 20:1729–1742. https://doi.org/10.1002/1097-0088(20001130)20:14<1729::aid-joc556>3.0.co;2-yCrossRefGoogle Scholar
  39. Lorenz EN (1956) Empirical orthogonal functions and statistical weather prediction. Science report 1, Statistical forecasting project, Department of Meteorology, MIT (NTIS AD 110268), p. 49Google Scholar
  40. Luo JJ, Yamagata T (2002) Four decadal ocean-atmosphere modes in the North Pacific revealed by various analysis methods. J Oceanogr 58:861–876.  https://doi.org/10.1023/a:1022831431602 CrossRefGoogle Scholar
  41. Luo JJ, Behera SK, Masumoto Y, Yamagata T (2011) Impact of global ocean surface warming on seasonal-to-interannual climate prediction. J Clim 24:1626–1646.  https://doi.org/10.1175/2010jcli3645.1 CrossRefGoogle Scholar
  42. Luo JJ, Sasaki W, Masumoto Y (2012) Indian ocean warming modulates Pacific climate change. Proc Natl Acad Sci USA 109:18701–18706  https://doi.org/10.1073/pnas.1210239109 Google Scholar
  43. McGuire AD et al (2016) Variability in the sensitivity among model simulations of permafrost and carbon dynamics in the permafrost region between 1960 and 2009. Glob Biogeochem Cycles 30:1015–1037.  https://doi.org/10.1002/2016gb005405 CrossRefGoogle Scholar
  44. Melillo JM et al (2002) Soil warming and carbon-cycle feedbacks to the climate system. Science 298:2173–2176.  https://doi.org/10.1126/science.1074153 CrossRefGoogle Scholar
  45. Mihalakakou G, Santamouris M, Lewis JO, Asimakopoulos DN (1997) On the application of the energy balance equation to predict ground temperature profiles. Sol Energy 60:181–190.  https://doi.org/10.1016/s0038-092x(97)00012-1 CrossRefGoogle Scholar
  46. Oelke C, Zhang TJ, Serreze MC (2004) Modeling evidence for recent warming of the Arctic soil thermal regime. Geophys Res Lett.  https://doi.org/10.1029/2003gl019300 Google Scholar
  47. Park HJ, Ahn JB (2016) Combined effect of the Arctic oscillation and the Western Pacific pattern on East Asia winter temperature. Clim Dyn 46:3205–3221.  https://doi.org/10.1007/s00382-015-2763-2 CrossRefGoogle Scholar
  48. Park H, Sherstiukov AB, Fedorov AN, Polyakov IV, Walsh JE (2014) An observation-based assessment of the influences of air temperature and snow depth on soil temperature in Russia. Environ Res Lett.  https://doi.org/10.1088/1748-9326/9/6/064026 Google Scholar
  49. Pearson RG, Phillips SJ, Loranty MM, Beck PSA, Damoulas T, Knight SJ, Goetz SJ (2013) Shifts in Arctic vegetation and associated feedbacks under climate change. Nat Clim Change 3:673–677.  https://doi.org/10.1038/nclimate1858 CrossRefGoogle Scholar
  50. Peng XQ et al (2017) Response of seasonal soil freeze depth to climate change across China. Cryosphere 11:1059–1073.  https://doi.org/10.5194/tc-11-1059-2017 CrossRefGoogle Scholar
  51. Pepin N et al (2015) Elevation-dependent warming in mountain regions of the world. Nat Clim Change 5:424–430.  https://doi.org/10.1038/nclimate2563 CrossRefGoogle Scholar
  52. Portmann RW, Solomon S, Hegerl GC (2009) Spatial and seasonal patterns in climate change, temperatures, and precipitation across the United States. Proc Natl Acad Sci USA 106:7324–7329.  https://doi.org/10.1073/pnas.0808533106 CrossRefGoogle Scholar
  53. Qi J, Li S, Li Q, Xing Z, Bourque CP-A, Meng F-R (2016) A new soil-temperature module for SWAT application in regions with seasonal snow cover. J Hydrol 538:863–877CrossRefGoogle Scholar
  54. Qian B, Gregorich EG, Gameda S, Hopkins DW, Wang XL (2011) Observed soil temperature trends associated with climate change in Canada. J Geophys Res.  https://doi.org/10.1029/2010jd015012 Google Scholar
  55. Qin J, Yang K, Liang S, Guo X (2009) The altitudinal dependence of recent rapid warming over the Tibetan Plateau. Clim Change 97:321–327.  https://doi.org/10.1007/s10584-009-9733-9 CrossRefGoogle Scholar
  56. Rankinen K, Karvonen T, Butterfield D (2004) A simple model for predicting soil temperature in snow-covered and seasonally frozen soil: model description and testing. Hydrol Earth Syst Sci 8:706–716CrossRefGoogle Scholar
  57. Ruiz-Barradas A, Kalnay E, Pena M, BozorgMagham AE, Motesharrei S (2017) Finding the driver of local ocean-atmosphere coupling in reanalyses and CMIP5 climate models. Clim Dyn 48:2153–2172.  https://doi.org/10.1007/s00382-016-3197-1
  58. Schlichtholz P (2016) Empirical relationships between summertime oceanic heat anomalies in the Nordic seas and large-scale atmospheric circulation in the following winter. Clim Dyn 47:1735–1753.  https://doi.org/10.1007/s00382-015-2930-5 CrossRefGoogle Scholar
  59. Sen PK (1968) Estimates of the regression coefficient based on Kendall’s tau. J Am Stat Assoc 63:1379–1389CrossRefGoogle Scholar
  60. Shen MG, Zhang GX, Cong N, Wang SP, Kong WD, Piao SL (2014) Increasing altitudinal gradient of spring vegetation phenology during the last decade on the Qinghai-Tibetan Plateau. Agric For Meteorol 189:71–80.  https://doi.org/10.1016/j.agrformet.2014.01.003 CrossRefGoogle Scholar
  61. Shen M et al (2015) Evaporative cooling over the Tibetan Plateau induced by vegetation growth. Proc Natl Acad Sci USA 112:9299–9304.  https://doi.org/10.1073/pnas.1504418112 CrossRefGoogle Scholar
  62. Sinha T, Cherkauer KA (2008) Time series analysis of soil freeze and thaw processes in Indiana. J Hydrometeorol 9:936–950.  https://doi.org/10.1175/2008jhm934.1 CrossRefGoogle Scholar
  63. Solomon S (2007) IPCC climate change 2007: the physical science basis. Contribution of working group I to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge and New York, NYGoogle Scholar
  64. Subin ZM, Koven CD, Riley WJ, Torn MS, Lawrence DM, Swenson SC (2013) Effects of soil moisture on the responses of soil temperatures to climate change in cold regions. J Clim 26:3139–3158.  https://doi.org/10.1175/jcli-d-12-00305.1 CrossRefGoogle Scholar
  65. Sun L, Shen BZ, Sui B, Huang BH (2017) The influences of East Asian Monsoon on summer precipitation in Northeast China. Clim Dyn 48:1647–1659.  https://doi.org/10.1007/s00382-016-3165-9 CrossRefGoogle Scholar
  66. Tesař M, Šír M, Krejča M, Váchal J (2008) Influence of vegetation cover on air and soil temperatures in the Šumava Mts. (Czech Republic). IOP conference series: earth and environmental science, 4, 012029.  https://doi.org/10.1088/1755-1307/4/1/012029
  67. Trenberth KE, Fasullo JT (2013) An apparent hiatus in global warming? Earths Future 1:19–32.  https://doi.org/10.1002/2013ef000165 CrossRefGoogle Scholar
  68. Vickers D, Mahrt L (1997) Quality control and flux sampling problems for tower and aircraft data. J Atmos Oceanic Technol 14:512–526. https://doi.org/10.1175/1520-0426(1997)014<0512:qcafsp>2.0.co;2CrossRefGoogle Scholar
  69. Wei Y, Fang Y (2013) Spatio-temporal characteristics of global warming in the Tibetan Plateau during the last 50 years based on a generalised temperature zone-elevation model. PloS One 8:e60044CrossRefGoogle Scholar
  70. Woodbury AD, Bhuiyan AKMH., Hanesiak J, Akinremi OO (2009) Observations of northern latitude ground-surface and surface-air temperatures. Geophys Res Lett.  https://doi.org/10.1029/2009gl037400 Google Scholar
  71. Yang K, Wu H, Qin J, Lin C, Tang W, Chen Y (2014) Recent climate changes over the Tibetan Plateau and their impacts on energy and water cycle: a review. Glob Planet Change 112:79–91.  https://doi.org/10.1016/j.gloplacha.2013.12.001 CrossRefGoogle Scholar
  72. Yesilirmak E (2014) Soil temperature trends in Buyuk Menderes Basin, Turkey. Meteorol Appl 21:859–866.  https://doi.org/10.1002/met.1421 CrossRefGoogle Scholar
  73. You Q, Kang S, Pepin N, Flügel W-A, Yan Y, Behrawan H, Huang J (2010) Relationship between temperature trend magnitude, elevation and mean temperature in the Tibetan Plateau from homogenized surface stations and reanalysis data. Glob Planet Change 71:124–133.  https://doi.org/10.1016/j.gloplacha.2010.01.020 CrossRefGoogle Scholar
  74. Zhan W, Zhou J, Ju W, Li M, Sandholt I, Voogt J, Yu C (2014) Remotely sensed soil temperatures beneath snow-free skin-surface using thermal observations from tandem polar-orbiting satellites: an analytical three-time-scale model. Remote Sens Environ 143:1–14.  https://doi.org/10.1016/j.rse.2013.12.004 CrossRefGoogle Scholar
  75. Zhang T (2001) Book review of “Geocryology in China” by Zhou et al. Permafr Periglac Process 12:315–322CrossRefGoogle Scholar
  76. Zhang T (2005a) Influence of the seasonal snow cover on the ground thermal regime: an overview. Rev Geophys.  https://doi.org/10.1029/2004rg000157 Google Scholar
  77. Zhang Y (2005b) Soil temperature in Canada during the twentieth century: complex responses to atmospheric climate change. J Geophys Res.  https://doi.org/10.1029/2004jd004910 Google Scholar
  78. Zhang T, Barry RG, Gilichinsky D, Bykhovets SS, Sorokovikov VA, Ye JP (2001) An amplified signal of climatic change in soil temperatures during the last century at Irkutsk, Russia. Clim Change 49:41–76.  https://doi.org/10.1023/a:1010790203146 CrossRefGoogle Scholar
  79. Zhang W, Li S, Wu T, Pang Q (2007) Changes and spatial patterns of the differences between ground and air temperature over the Qinghai-Xizang plateau. J Geog Sci 17:20–32.  https://doi.org/10.1007/s11442-007-0020-2 CrossRefGoogle Scholar
  80. Zhao L, Ping CL, Yang DQ, Cheng GD, Ding YJ, Liu SY (2004) Changes of climate and seasonally frozen ground over the past 30 years in Qinghai-Xizang (Tibetan) Plateau, China. Glob Planet Change 43:19–31.  https://doi.org/10.1016/j.gloplacha.2004.02.003 CrossRefGoogle Scholar
  81. Zhou LM, Dai A, Dai YJ, Vose R, Zou CZ, Tian YH, Chen HS (2009) Spatial dependence of diurnal temperature range trends on precipitation from 1950 to 2004. Clim Dyn 32:429–440.  https://doi.org/10.1007/s00382-008-0387-5 CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2017

Authors and Affiliations

  • Fuxin Zhu
    • 1
    • 2
  • Lan Cuo
    • 1
    • 2
    • 3
  • Yongxin Zhang
    • 4
  • Jing-Jia Luo
    • 5
  • Dennis P. Lettenmaier
    • 6
  • Yumei Lin
    • 2
    • 7
  • Zhe Liu
    • 1
    • 2
  1. 1.Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau ResearchChinese Academy of SciencesBeijingChina
  2. 2.University of Chinese Academy of SciencesBeijingChina
  3. 3.Center for Excellence in Tibetan Plateau Earth SciencesChinese Academy of SciencesBeijingChina
  4. 4.National Center for Atmospheric ResearchBoulderUSA
  5. 5.Australian Bureau of MeteorologyMelbourneAustralia
  6. 6.Department of GeographyUniversity of CaliforniaLos AngelesUSA
  7. 7.Key Laboratory of Resource Use and Environmental Remediation, Institute of Geographic Sciences and Natural Resources ResearchChinese Academy of SciencesBeijingChina

Personalised recommendations