Influence of Climate Warming on the Ground Surface Stability over Permafrost along the Qinghai–Tibet Engineering Corridor

Abstract

The warming climate has posed a serious threat on ground surface stability. In permafrost regions, ground surface instability may induce engineering and geological disasters, especially for the engineering corridor. It is difficult to evaluate ground surface stability over permafrost because the stability is influenced by various factors in permafrost regions. Many single index models cannot comprehensively evaluate the ground surface stability for permafrost. We, therefore, proposed an evaluation model considering different influential factors based on the trapezoidal fuzzy Analytical Hierarchy Process (AHP) method. And the ground surface stability was calculated and analyzed along the Qinghai–Tibet Engineering Corridor under three climate warming conditions (the slow climate warming, the medium climate warming and the rapid climate warming). The results show that the ground surface stability influential factors, including the mean annual ground temperature, the active layer thickness, and the volume ice content, will be greatly changed with the warming climate. By 2100, the percentage of high-temperature permafrost (−0.5 °C < T ≤ 0 °C) will increase about 29.45% with rapid climate warming. The active layer thickness will have an average thickening rate of about 0.030 m/year. Most of the high ice content permafrost will change to low ice content permafrost. The ground surface stability, therefore, will be greatly changed with the warming climate along the Qinghai–Tibet Engineering Corridor. Compared to the present, the stable area will decrease about 5.28% by 2050 under the slow climate warming. And that is approximately 7.91% and 21.78% under the medium and rapid climate warming, respectively. While in year 2100, the decrement is obviously increased. The stable area will decrease about 11.22% under the slow climate warming and about 17.3% under the medium climate warming. The proportion of stable area, however, has an increasing trend under the rapid climate warming. This phenomenon is mainly caused by the warming climate which can lead to the permafrost being degraded to melting soil. The unstable area is mainly distributed near the Chumaer River high plain, Tuotuohe–Yanshiping, Wudaoliang, Tangula Mountains, and other high-temperature permafrost areas. This paper provides a reference for geological hazard prevention and engineering construction along the Qinghai–Tibet Engineering Corridor.

1. Introduction

Permafrost is defined as soil or rock having a temperature at or below 0 °C for at least two consecutive years [1]. It is, therefore, sensitive to temperature variations. In recent years, the warming climate has posed a serious threat to the ground surface stability in permafrost regions. Various geological hazards, such as landslides, frost heaving, and thaw slump, have already occurred in permafrost areas [2]. Meanwhile, typical engineering diseases, including ground surface settlement, wave transformation, and fissures, have seriously affected the stable operation of engineering [3,4]. The ground surface stability over permafrost, therefore, is very important for engineering safe operation [5]. This will greatly influence the sustainable development in permafrost regions, especially for the Qinghai–Tibet Plateau (QTP).

The QTP, called the “Third pole of the Earth”, has an average altitude exceeding 4000 m. It is a driver and amplifier of the global climate changing [6]. The Intergovernmental Panel on Climate Change Fifth Assessment Report (IPCC AR5) shows that the global annual ground surface temperature will increase about 0.9 °C~4.02 °C by the end of the 21st century [7]. And the phenomenon will be more significant in the QTP. The Qinghai–Tibet Engineering Corridor (QTEC) is located in the central part of the QTP. It is a critical engineering and transportation corridor in the QTP. There are many major linear engineering projects that have been constructed in this corridor, such as the Qinghai–Tibet Highway (QTH), Qinghai–Tibet Railway (QTR), high-tension line, communication optical fiber cable, and a petroleum pipeline [8]. And there are still many important projects planned in this corridor. Meanwhile, it is the most important channel of contact between the inlands of China and the QTP. However, more than 550 km of the corridor is covered by permafrost. The QTEC, therefore, is the key climate-change indicator for the QTP because it is influenced by both the climate warming and by human activities.

Ground surface stability over permafrost is influenced by various influences, such as the active layer thickness (ALT), the mean annual ground temperature (MAGT), the volume ice content of permafrost (VIC), and so on. These influential factors have been significantly changed with the warming climate along the QTEC. Field monitoring shows that the ALT has been thinned with a rate of about 6.3 cm/year~7.8 cm/year in the past 10 years [8]. The MAGT also has been increased. The in situ test results illustrated that the increasing rate is greater than 0.1 °C/year~0.3 °C/year [9]. The volume ice content has constantly decreased. And the rate of the ice thawing is accelerating, especially for the ice close to the permafrost table because of the rapid climate warming [10]. The changing of the influential factors will pose a serious threat to the ground surface stability over permafrost along the QTEC. Various geological and engineering disasters, such as frost heaving and thawing settlement, have been happening with the sustainable warming climate along the QTEC in the past few decades [11]. Meanwhile, the ground surface instability is aggravated by the global sustainable warming climate accelerating.

There are three methods, including the empirical formula, the theoretical model, and the artificial intelligence training algorithm, used to evaluate the ground surface stability in permafrost regions. The empirical formula was established based on abundant in situ measurements [12]. The availability of the empirical formula, therefore, is questionable due to the lack of sufficient field results in permafrost regions because it is difficult to monitor ground surface deformation on a large scale for a long term under bad environmental conditions, especially for the QTP (the QTP has a low pressure, high altitude, and strong radiation). Meanwhile, the spatio-temporal monitoring method has many restrictions, such as a lack of feature points, in permafrost regions. The theoretical model was established based on different theory, such as the grey system theory (GST), the fuzzy mathematic theory, and so on [13,14]. The theoretical model, therefore, has little reliance on the field measurements. But it mostly belongs to qualitative analysis. And the artificial intelligence model is mainly based on training algorithms, including the artificial neural network (ANN) algorithm, the adaptive neuro-fuzzy inference system (ANFIS) algorithm, and so on [15,16]. The theoretical model is constructed based on the training algorithm to learn and analyze the field measurements. So, the more field measurements, the more precise the model. The accuracy of this method, therefore, is difficult to guarantee because there are not enough in situ measurements.

Many researchers, therefore, mainly proposed the theoretical model to evaluate the ground surface stability over permafrost because of the lack of field measurements. There is a freeze–thaw disease index model that was established to analyze the susceptibility of freeze–thaw diseases along the QTEC [12]. This study shows that the ground surface has an obvious settlement with the temperature increasing. But in that study, the author only considered the influence of ice content and active layer thickness on the ground surface stability in permafrost regions. The ground surface stability is influenced by various factors. So, the result of that study was incomprehensive. In addition, there is a stability evaluation model established based on the logistic regression theory. Many factors, including the ice content, the annual average ground temperature, the surface vegetation, and so on, were considered in this model [17]. However, in this model, the influential factors are treated as independent, and the field diseases are treated as the dependent variable. So, it needs enough field disease information to verify the accuracy of the evaluated results. Additionally, there is an index model proposed based on the Analytical Hierarchy Process (AHP) method [18]. This study also considered many factors, such as the MAGT, the altitude, and so on. And it is very useful to evaluate the vulnerability of the thermokarst lake from the Chumaer River High Plain to the Feng-volcano area along the QTEC. But it is mainly just for the thermokarst lake hazard. There are various hazards, such as the ground surface settlement and frost heave, induced by the ground surface instability along the QTEC. In conclusion, the previous evaluation model can be improved in the following aspects. Firstly, the factors that influence ground surface stability over permafrost should be fully considered in the model. Secondly, as few field measurements as possible should be used to establish the most accurate model. In addition, the evaluation process should be as simple as possible. So, we proposed an evaluation model, considering the above research, based on the trapezoidal fuzzy Analytical Hierarchy Process (AHP) theory [19]. In the model, the factors that influenced ground surface stability were comprehensively considered based on the results of other researchers and our previous study. In addition, the geographic information system (GIS) tool provides a valuable method for ground surface stability evaluation in permafrost regions [20]. It has a powerful computing capability. With the help of GIS, we can evaluate the ground surface stability in a simple and speedy way.

This research aims to evaluate the influence of the warming climate on ground surface stability over permafrost along the QTEC. In this study, we first proposed a ground surface stability model (I_s) to evaluate the ground surface stability. This model was based on the trapezoidal fuzzy Analytical Hierarchy Process (Tra-AHP) method combining with GIS technology. And in this model, we comprehensively considered three ground surface stability influential factors, including the mean annual ground temperature, the active layer thickness, and the volume ice content. Secondly, we calculated and analyzed the distribution of the three influential factors. The changing of these factors under three different climate warming conditions were analyzed. Thirdly, we calculated and evaluated the ground surface stability according to the established model. And lastly, we classified and mapped the ground surface stability along the QTEC with the help of GIS technology. The results show that it is necessary to take some reasonable prevention measures in poor stability areas along the QTEC with the climate warming in future so that the ground surface over permafrost will be in a stable state for a long time. This can keep the safe operation of engineering facilities along the corridor and promote the sustainable development of the economy in permafrost regions. Meanwhile, it will provide persistent guidance for the planning and construction of new projects in this region.

2. Study Area

The Qinghai–Tibet Engineering Corridor (QTEC) from Golmud (Qinghai Province) to Lhasa (Tibet Autonomous Region), is a north–south corridor in the central part of the Qinghai–Tibet Plateau (QTP) (Figure 1a). More than 550 km, from Xidatan area to Amdo area, traverses the permafrost regions. It is within the longitude 91° E–95° E and the latitude 32° N–36° N (Figure 1b). The altitude changes from 3700 m to 6200 m in this area. The soil is made of loam, silty clay, clay, and sand [21]. The mean month air temperature varies from −30 °C (the lowest temperature) in winter to 25 °C (the highest temperature) in summer [22]. The mean annual precipitation is very low, approximately 250 mm~400 mm. And more than 80% of the precipitation occurs from May to September in a year. While the mean annual evaporation is about 1500 mm, about four to six times more than the precipitation [17]. The ground surface is mainly covered by alpine meadow and the steppe which belongs to the typical plateau vegetation [23]. The long-term field measurements show that the mean annual air temperature has been significantly increased in this engineering corridor [22]. Meanwhile, the field investigations indicate that various geologic hazards and engineering diseases, such as ground surface frost heave and thaw settlement, induced by ground surface instability already took place along the QTEC [11,24]. The warming climate, therefore, has a significant influence on the ground surface stability of this engineering corridor.

3. Data Sets and Method

3.1. Data and Processing

3.1.1. Basic Data

In this research, the basic data mainly contains the influential factors and the climate warming data. The influential factors are in the evaluation model, including the active layer thickness (ALT), the mean annual ground temperature (MAGT), the volume ice content of permafrost (VIC), according to our previous research [25]. The ground surface slope degree (SD), the ground surface vegetation (NDVI), and the soil type of permafrost (ST) were needed to calculate the influential factors of the study area. The SD was extracted from digital elevation model (DEM) data (https://daac.gsfc.nasa.eu/ (accessed on 1 April 2022)). The ST was mapped according to the soil classification data (https://www.gscloud.cn/ (accessed on 20 April 2022)). The NDVI was calculated based on the Landsat data (https://nex.nasa.gov.nex/ (accessed on 3 April 2022)) by Formula (1). In the formula, the NIR and R represent the reflection value of the ground object near the infrared band and in the red band, respectively. The influential factors (the ALT, the MAGT, and the VIC) were computed using the empirical models proposed by previous researchers based on in situ observations with the help of GIS technology. The climate warming change was simulated and calculated based on the different Representative Concentration Pathway (RCP) scenarios reported in the IPCC AR5. For further study, all the basic data have been converted to uniform resolution by scaling down using GIS.

$$NDVI=\left(NIR-R\right)/\left(NIR+R\right)$$

(1)

3.1.2. Active Layer Thickness (ALT)

The active layer thickness (ALT) refers to the maximum thawing depth of permafrost. It is influenced by various factors, such as the ground surface temperature, the thermal properties and water content of soil and so on. There are some models that have been proposed to calculate the active layer thickness of permafrost [26,27]. In this study, we calculated the ALT along the QTEC based on the Stefan theory which has been widely used in the QTP (Formula (2)). In the formula, the λ is the thermal conductivity of the soil. The I_s represents the ground surface freeze and thaw index changing with the warming climate. It was calculated mainly based on ground surface temperature data (Formula (3)). The L is the latent heat of ice fusion. The γ is the bulk density of the soil. The W is the water content in thawed soil. The W_u represents the unfrozen water content in the soil. In the formula, $\overline{T_s}$ represents the mean annual temperature. The T_s is the mean annual temperature. T_c and T_w represent the mean monthly temperature in the coldest and warmest month, respectively. The W, N, and H represent the longitude, the latitude, and the elevation of the study area, respectively. The ΔT is the variation of temperature in the future. It equaled to zero when we calculated the current ALT. These parameters were determined based on the physical and thermal features of soil in permafrost (

Table 1. The parameters for the ALT calculation.

Soil Type	γ/(kg•m −3)	λ/(W•m −1 •°C−1)	W/(%)	Wu/(%)
Silty clay	1500	1.57	17	5
Clay	900	0.53	15	5
Sand	1800	1.42	6	5

).

$$ALT=\sqrt{\frac{2\lambda I_s}{L\gamma (W-W_u)}}$$

(2)

$$I_s=\overline{T_s}\times L_s$$

(3)

$$L_s=365(\beta /\pi )$$

$${\overline{T}}_s=T_s+\alpha (\sin \beta /\beta )$$

$$\alpha =(T_w-T_c)/2$$

$$\beta ={\cos }^{-1}(-T_s/\alpha )$$

$$T_s=62.434-0.152W-0.768N-0.00463H$$

$$T_c=56.086-0.042W-1.579N-0.00457H$$

$$T_w=69.443-0.371W-0.035N-0.00478H$$

3.1.3. Mean Annual Ground Temperature (MAGT)

The MAGT is generally treated as the temperature of 0.1 °C annual amplitude of the ground temperature [28]. There are some empirical models that have been proposed to calculate the MAGT in permafrost regions based on the in situ measurements [29,30,31]. In this study, the MAGT along the QTEC was calculated based on Formula (4) using GIS technology. In the formula, the N and H represent the latitude and elevation of the study area, respectively. It can be acquired from the DEM data.

$$MAGT=50.633-0.830N-0.005H$$

(4)

3.1.4. Volume Ice Content (VIC)

The volume ice content of permafrost is influenced by various factors, such as the topography, ground surface vegetation, soil type, and so on. There are some nonparameter experiential models, between the VIC and the factors, that were established based on the field investigations over permafrost along the QTEC [11,32]. In this study, the model proposed by Zhang et al. [11] (Formula (5)) was used to calculate the ice content of permafrost of the study area. In Formula (5), the ST, the NDVI, the SD, and the MAGT are the normalized values in Section 3.1.1 (Figure 2). The weight coefficient of the influential factor was calculated by GIS technology using the analysis hierarchy process method.

$$VIC=0.34\times ST+0.29\times NDVI+0.24\times SD+0.13\times MAGT$$

(5)

Figure 3 shows the calculated influential factors (ALT, MAGT, VIC) of the study area at present.

3.1.5. Climate Warming

Climate warming is the main reason inducing ground surface instability in permafrost regions. In this study, the climate change data were obtained from the Climate Change Agriculture and Food Security (CCAFS) data portal (https://www.ccafs-climate.org/ (accessed on 12 June 2022)). The current temperature data were used to predict the climate change in the future. The changing climate was simulated and calculated based on the BCC-CSM1-1 model [33]. Four Representative Concentration Pathway (RCP) scenarios, including the RCP2.6, RCP4.5, RCP6.0, and RCP8.5, in IPCC AR5 reports are proposed to represent the radiation force level under four greenhouse gas concentration trajectories. The RCP2.6 is the greenhouse gas strictly reducing emission scenario. The RCP4.5 and RCP6.0 belong to the greenhouse gas medium emission scenario. And the RCP8.5 is the greenhouse gas heavy emission scenario [34]. In this study, we predicted the climate changing in 2050 and 2100 under the RCP2.6 scenario (representing the slow climate warming condition), the RCP4.5 scenario (representing the medium climate warming condition), and the RCP8.5 scenario (representing the rapid climate warming condition), respectively. The results are shown in

Table 2. The climate changes in future periods of the study area.

RCPs	Period	Temperature (°C)
		Minimum	Maximum	Mean	Standard Deviation	Covariation
	Current	−21.16	24.48	−1.42	4.97	24.70
RCP2.6	2050	−21.22	25.67	−0.18	5.08	25.81
	2100	−21.41	25.62	0.02	5.03	25.30
RCP4.5	2050	−20.10	26.10	−0.64	5.02	25.20
	2100	−19.90	26.40	0.25	5.08	25.81
RCP8.5	2050	−20.41	26.24	1.02	5.04	25.40
	2100	−17.92	28.76	2.78	4.99	24.90

.

3.2. Method

There are many models that have been used to evaluate the ground surface stability in permafrost regions. The settlement model was mainly used to evaluate the thawing and settling of permafrost [35]. It, therefore, cannot be used to evaluate frost heave for permafrost. The logistic regression model has a higher accuracy [18]. It is, however, greatly dependent on in situ measurements. And the process is relatively complicated, too. The comprehensive model considers various influential factors [36]. Meanwhile, the results are more scientific and accurate. And the process is simple. We, therefore, used this model (Formula (6)) to evaluate the ground surface stability for our study area. In the formula, the I_s represents the ground surface stability index. The P is the various influential factors. And the W represents the weight coefficient of different influential factors.

$$I_{\mathit{s}}={\displaystyle \sum _{i=1}^n{P}_i}\times W_i$$

(6)

Ground surface stability is influenced by various factors in permafrost regions. Previous research and our study have indicated that the ALT, the MAGT, and the VIC are the three main factors [11,12,25,32]. We, therefore, chose these factors as the influential factors in Formula (6).

The weight coefficient of the three influential factors was calculated using the trapezoidal fuzzy Analytical Hierarchy Process (Tra-AHP) method [19]. The Tra-AHP method is an improvement of the AHP which is a comprehensive method based on multicritical indexes. It is used to analyze both the qualitative and quantitative questions [37]. In the AHP method, however, only a single weight is used to express relative importance for the factors. The Tra-AHP overcomes the crisp value shortcoming of the original AHP. It allowed us to make use of the intermediate values between two extremes: real numbers 0 and 1. And it also suggested that the membership function should be operated in the range of real numbers [0, 1]. This is based on the fuzzy logic theory introduced by Zadeh in 1965 [38]. The Tra-AHP, therefore, can arithmetically handle and interpret fuzzy numbers in a variable way.

In the Tra-AHP method, a trapezoidal fuzzy number (M) replaces the traditional single numerical. It can be used to construct a more realistic judgment matrix for the evaluation index model so that the calculated weight coefficient for the influential factor is more in line with the actual situation.

In this study, we assumed that M = (a, b, c, d) is the trapezoidal fuzzy number. And its subordinate function can be expressed by Formula (7).

$${\mathit{u}}_M\left(\mathit{x}\right)=\left\{\begin{array}{cc}0& \mathit{x}<\mathit{a},\mathit{x}>\mathit{d}\\ \frac{\mathit{x}-\mathit{a}}{\mathit{b}-\mathit{a}}& \mathit{a}\le \mathit{x}\le \mathit{b}\\ 1& \mathit{b}\le \mathit{x}\le \mathit{c}\\ \frac{\mathit{x}-\mathit{d}}{\mathit{c}-\mathit{d}}& \mathit{c}\le \mathit{x}\le \mathit{d}\end{array}\right.$$

(7)

where a ≤ b ≤ c ≤ d, a and d are the upper and lower bounds of M, respectively. And the median of M are c and d.

In Figure 4, the trapezoidal fuzzy number (M) will become a triangle fuzzy number when b equals to c. And that will be changed to an interval fuzzy number when b equals to c and c equals to d. It can be seen that the trapezoidal fuzzy number can deal with fuzzy numbers very well using a variable way. Meanwhile, the 1–9 evaluation scale expert consultation system was considered, and the specific value was determined as in Figure 5.

The weight coefficient of three influential factors was calculated using Formulas (8)–(10). In the formulas, the w(i) represents the trapezoidal fuzzy evaluation value of the influential factor. The w(i) is the evaluation expectation of the influential factor. And the $\overline{w}$(i) represents the normalized weight coefficient for influential factor.

$$\begin{gathered} \mathit{w}(\mathit{i})=({\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{a}}_{\mathit{i}\mathit{j}}},{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{b}}_{\mathit{i}\mathit{j}}},{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{c}}_{\mathit{i}\mathit{j}}},{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{d}}_{\mathit{i}\mathit{j}}})\otimes {({\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{a}}_{\mathit{i}\mathit{j}}},{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{b}}_{\mathit{i}\mathit{j}}},{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{c}}_{\mathit{i}\mathit{j}}},{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{d}}_{\mathit{i}\mathit{j}}})}^{-1} \\ =(\frac{{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{a}}_{\mathit{i}\mathit{j}}}}{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{\mathit{d}}_{\mathit{i}\mathit{j}}}},\frac{{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{b}_{\mathit{i}\mathit{j}}}}{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{c}_{\mathit{i}\mathit{j}}}},\frac{{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{c}_{\mathit{i}\mathit{j}}}}{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{b}_{\mathit{i}\mathit{j}}}},\frac{{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{d}_{\mathit{i}\mathit{j}}}}{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}}{\displaystyle \sum _{\mathit{j}=1}^{\mathit{n}}{a}_{\mathit{i}\mathit{j}}}}) \end{gathered}$$

(8)

$$\mathit{w}(\mathit{i})=\frac{{\mathit{a}}_{\mathit{i}}+{\mathit{b}}_{\mathit{i}}+{\mathit{c}}_{\mathit{i}}+{\mathit{d}}_{\mathit{i}}}{4}$$

(9)

$$\overline{\mathit{w}}(\mathit{i})=\frac{\mathit{w}\left(\mathit{i}\right)}{{\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}\mathit{w}\left(\mathit{i}\right)}}$$

(10)

The calculated weight coefficient of three factors in our study was shown in

Table 3. Weight coefficient for the three influential factors.

Influential Factors	ALT	MAGT	VIC
Weight ($\overline{w}$(i))	0.20	0.45	0.35

.

4. Results

4.1. Influential Factors Variation

We evaluated the ground surface stability along the QTEC using the evaluation model. In this model, the SD and ST belong to the geological data that will have a small change with time. The vegetation coverage over permafrost is relatively low along the QTEC (Figure 2c). We, therefore, assumed that the SD, ST, and NDVI will not change with climate warming in the future.

4.1.1. Active Layer Thickness (ALT) Variation

The variation and distribution of the ALT along the QTEC can be seen in Figure 6 and Figure 7. It can be seen that the ALT varies between 1.44 m and 5.72 m under the slow climate warming condition in 2050 (Figure 6a). It will be thickened by about 0.16 m~0.91 m more than at present. And the ALT changes from 1.59 m to 6.28 m under the medium climate warming condition (Figure 6b). It will be thickened by about 0.28 m~1.08 m more than the current thickness. While under the rapid climate warming condition, the ALT varies from 1.67 m to 6.61 m (Figure 6c). The ALT will be thickened by about 0.33 m~1.23 m compared to the current thickness. While by 2100, the ALT changes from 1.48 m to 5.84 m under the slow climate warming condition (Figure 7a). It will be thickened by about 0.21 m~1.01 m more than the current thickness. The ALT is between 1.71 m and 6.87 m under the medium climate warming condition (Figure 7b), thickened by about 0.35 m~1.14 m. And the ALT varies from 1.98 m to 7.62 m under the rapid climate warming condition (Figure 7c). Compared to currently, it will be thickened by about 0.51 m~1.29 m. So, it can be seen that the thickness of the active layer will increase as time goes by at the same warming climate condition. And the ALT will have an obvious thickening near Fenghuo Mountain, Tuotuo River, and Tangula Mountain. This is mainly caused by the high temperature and high ice content in these areas. The thickening of the ALT is significantly greater under the rapid climate warming condition than that under the slow and the medium climate warming conditions.

For further study, we compared the changing of the ALT in 2050 and 2100 under different climate warming conditions. We calculated the minimum ALT, the mean ALT, and the maximum ALT, respectively (Figure 8). The results show that the ALT thickened relatively slowly under the slow climate warming condition. The minimum thickness will increase by about 0.22 m in 2050. The maximum of ALT will be thickened 0.75 m. The average thickness is approximately 0.36 m, while the minimum thickness and the maximum thickness will increase by about 0.29 m and 0.91 m in 2100. And the average thickness is increased about 0.56 m. In general, the ALT will have an average thickening rate of about 0.009 m/year under the slow climate warming condition, while the active layer will have a significant variation under the rapid climate warming condition. The minimum thickness will increase by about 0.45 m in 2050. The maximum thickness will be thickened about 1.64 m. In 2100, the minimum thickness and the maximum thickness will increase by about 0.76 m and 2.65 m, respectively. The average thickness is approximately 0.72 m and 1.48 m. And the average thickening rate is about 0.021 m/year. It is significantly greater than the thickening rate under the slow climate warming condition.

4.1.2. Mean Annual Ground Temperature (MAGT) Variation

The variation of the mean annual ground temperature (MAGT) with the warming climate is shown in Figure 9 and Figure 10. We can see that most of the study area was in a low temperature condition. The MAGT is below −1 °C in 2050 and in 2100 under the slow climate warming condition (Figure 9a and a). While under the rapid climate warming condition, the proportion of low temperature (T ≤ −2 °C) is clearly decreased. Meanwhile, the area of high temperature (−0.5 °C~0 °C) is significantly increased (Figure 9c and Figure 10c). The statistical results show that the mean MAGT will increase by about 0.63 °C under the slow climate warming condition in 2050. And that it will increase approximately 0.94 °C and 1.57 °C under the medium and rapid climate warming condition, respectively. The mean increasing rate of MAGT is about 0.12 °C/10 years under the slow climate warming condition. And that increase is greater under the medium and rapid climate warming conditions at 0.17 °C/10 years and 0.22 °C/10 years, respectively. While the increase of the mean MAGT will reach 1.12 °C under the slow climate warming condition in 2100. And that will rise to 1.68 °C and 2.48 °C under the medium and rapid climate warming conditions, respectively. The mean increasing rate of MAGT is about 0.15 °C/10 years, 0.21°C/10 years, and 0.31°C/10 years under three different warming climate conditions, respectively. It also can be seen that there will be almost no low-temperature permafrost (T ≤ −2 °C) along the QTEC in 2100 and the study area will be mainly changed to high-temperature permafrost (−0.5 °C~0 °C). Meanwhile, most of the permafrost will be degraded to melting soil under the rapid climate warming condition. And the degeneration will mainly occur near the Chumaer River, the Tuotuo River, and the Tongtian River. In addition, the same phenomenon will happen along the Wudaoliang and Feng-volcanic areas. So, these areas will have poor ground surface stability. It is, therefore, necessary to take some reasonable precautions.

Figure 11 shows the percentage of MAGT in different intervals of the study area under different climate warming conditions. The results show that the low-temperature permafrost (T ≤ −2.0 °C) will decrease to 25.3% under the slow climate warming condition in 2050. And that it will decrease to 9.1% and 5.6% under the medium and rapid climate warming conditions, respectively. While in 2100, the low-temperature permafrost will have the same tendency. The proportion will decrease to 19.24% under the slow climate warming condition. And it will decrease to 8.5% and 2.7% under the medium and rapid climate warming conditions. This is obviously less than the proportion of low-temperature permafrost at present (about 51.5%). In contrast, the high-temperature permafrost (−0.5 °C < T ≤ 0 °C) will increase in the future. It is approximately 5.2% at present, while in 2050, it will increase to 16.7%, 27.2%, and 32.1% under the three different warming climate conditions. And the increasing trend will continue as time goes by. In 2100, it will increase to 20.8% and 31.2% under the slow and medium climate warming conditions, respectively. And under the rapid climate warming condition, it will increase to 26.2%. In addition, it also can be seen that the permafrost has a significant degradation in 2100. The percentage of melting soil (T > 0 °C) is about 11.3% and 23.6% under the slow and medium climate warming conditions, respectively. But it is as high as 56.3% under the rapid climate warming condition. That is so say, part of the low-temperature permafrost will be degraded into high-temperature permafrost with the warming climate. It will even degrade to melting soil.

4.1.3. Volume Ice Content (VIC) Variation

The variation and distribution of volume ice content (VIC) of the study area in the future is illustrated in Figure 12 and Figure 13. We can see that the changing of VIC is smaller under the slow climate warming condition (Figure 12a and Figure 13a). The decreasing of VIC mainly occurs in the areas where the MAGT has an obviously increasing trend. While the VIC of permafrost is significantly decreased with the warming climate under the rapid climate warming condition. The high ice content permafrost (I > 30%) is clearly decreased near the Chumaer River, the Tuotuo River, and the Tongtian River (Figure 12c and Figure 13c). And the decreasing trend is especially distinct near Fenghuo Mt. That is because the MAGT has been significantly increased in these areas under the rapid climate warming condition. From Figure 12 and Figure 13, it also can be seen that the VIC has less obvious decreasing trend in the low ice permafrost along the QTEC.

4.2. Ground Surface Stability

The variation and distribution of the ground surface stability along the QTEC was evaluated based on the established model (I_s) in Section 3.2. Then, we classified and mapped the ground surface of the study area into three regions, including the stable region, the sub-stable region and the unstable region, with the help of GIS technology.

Figure 14 shows the classification of the ground surface stability in the study area at the current time. It can be seen that most of the study area is stable. There is only a small unstable area with a larger ground surface stability index. And the unstable area is mainly distributed near the Tuotuo River and the Tongtian River. That is because these areas belong to the high temperature and the high ice content areas. In these areas, the MAGT varies from −1 °C to 0 °C and the VIC is higher than 30%. Meanwhile, the ALT is very thin, about 3 m in these areas.

As Figure 15 and Figure 16 show, the ground surface stability of the study area will have a great change in the future with the warming climate. The stable area will decrease and the sub-stable and unstable areas will increase. And the increased area will mainly be distributed near the Chumaer River and the Tongtian River, Fenghuo Mt., Tangula Mt., and the Wudaoliang area (Figure 15). These areas belong to the high temperature and the high ice content areas (Figure 3). For further study, the proportion of these three regions were calculated (

Table 4. Proportion of the ground surface stability level along the QTEC in the future with climate warming.

Stability Level	Proportion (%)
	Current	RCP2.6		RCP4.5		RCP8.5
		2050	2100	2050	2100	2050	2100
Stable	63.35	58.07	52.13	55.44	46.05	41.57	58.24
Sub-stable	33.58	34.02	36.11	35.42	35.09	37.81	24.87
Un-stable	3.07	7.91	11.76	9.14	18.86	20.62	15.89

). From Figure 15a and Table 4, it can be seen that the study area is mainly stable and sub-stable under the slow climate warming condition in 2050. The proportion is about 58.07% and 34.02%, respectively. There is about 7.91% of the area that is unstable. And the unstable area is mainly distributed in the high temperature permafrost areas between the Chumaer River plain and the Tuotuohe–Yanshiping area. From Figure 15b and Table 4, we can see that the stable area is decreased, while the proportion of sub-stable and unstable areas are increased in 2050 under the medium climate warming condition. But, in general, the study area is mainly dominated by stable area. The proportion of stable area is approximately 55.44%, which is greater than the sub-stable area and the unstable area of about 35.42% and 9.14%, respectively. From Figure 15a and Table 4, it can be seen that the ground surface stability of the study area also has a great change in 2050 under the rapid climate warming condition. The proportion of stable area will decrease to 41.57%. And the unstable area will increase to 20.62%. It is worth noting that part of the low-temperature permafrost areas will change to unstable areas. This phenomenon mainly occurs near Wudaoliang and the Tangula Mountains. It is closely related to the warming temperature of permafrost.

Figure 16 and Table 4 show the distribution of the ground surface stability in 2100 under different warming climate conditions. We can see that most of the study area is still stable under the slow climate warming condition (Figure 16a). The proportion is about 52.13%. The proportion of unstable area is less at 11.76% (Table 4). And the unstable area is mainly distributed near the Chumaer River High Plain, the Fenghuo mountain area, the Tuotuohe–Yanshiping area, and other high-temperature permafrost areas. While the stable areas will decrease to 46.05% under the medium climate warming condition (Table 4). Meanwhile, the unstable area will increase to 18.86%. And the distribution of the unstable area will be spread to Wudaoliang, the Tongtian River, and other areas (Figure 16b). However, the variation tendency of ground surface stability is a little different under the rapid climate warming condition. The sub-stable area and unstable area will decrease. The proportion is about 24.87% and 15.89%, respectively (Table 4). But the proportion of stable areas will increase to 58.24%. The study area, therefore, will be mostly stable. That is because most of the permafrost will be degraded to melting soil which has a temperature higher than 0 °C. The melting soil has a relatively good stability compared to permafrost. So, the occurrence of freezing–thawing disease is relatively small.

In general, the ground surface stability of the study area will be greatly changed with the warming climate in the future. The permafrost will be severely degraded. Most of the study area will be evolved into high-temperature and low volume ice content permafrost. More and more stable area will be degraded in sub-stable area, even into unstable area and melting soil. These phenomena will greatly threaten the stability of the ecological environment and engineering facilities in the study area. It is, therefore, necessary to pay more attention in these sub-stable and unstable areas. Some reasonable protections should be taken to prevent engineering disasters and geological hazards in these areas.

5. Discussion

The sustainable warming climate has a great influence on the ground surface stability in permafrost regions, especially for the Qinghai–Tibet Engineering Corridor (QTEC). However, ground surface stability is the premise for the sustainable development of the engineering corridor. Meanwhile, it is the key to keeping the projects in safe operation in the future along the corridor. In this study, we therefore focus on evaluating the influence of climate warming on ground surface stability over permafrost along the QTEC. An evaluation model considering three influential factors was proposed based on the trapezoidal fuzzy Analysis Hierarchy Process (Tra-AHP) method. It is expected to help keep the safe operation of engineering in permafrost regions. The results have found that the warming climate will seriously threaten the ground surface stability over the permafrost along the QTEC. And the influential factors, both the active layer thickness and the mean annual ground temperature, will be greatly changed with climate warming. The result has a good guiding significance for the prevention of freeze–thaw diseases along the Qinghai–Tibet Engineering Corridor in the future. However, the accuracy, the application in permafrost regions, and the long-term effectiveness of the established model are not elaborated upon in detail in this study because there is a lack of in situ measurements. For these issues, we only simply discuss and analyze.

Firstly, the accuracy of results is key to ensure the proposed model is reasonable and effective. Having only a few verification results will have great limitations on its accuracy. However, the collection and accumulation of large field measurements is a long-term work. We, therefore, will need more laboratory tests and field measurements to evaluate the accuracy of the newly established model. In the next step, we will construct the laboratory tests under different geological conditions and considering different influential factors. Meanwhile, periodical field observations should be carried out so that we can better evaluate the accuracy of the established model using more data.

Secondly, the application of the evaluation model is a concern in the engineering practice. The application of the model is determined by various factors, including the geological conditions and the properties of the permafrost. In this research, the study area was located in an extreme cold region (on the QTP). The ground surface stability is mainly induced by influential factors including the active layer thickness (ALT), the mean annual ground temperature (MAGT), and the volume ice content of permafrost (VIC), while the influential factors may change in other cold regions. So, it is necessary to verify the application of the evaluation model in other areas. However, it needs more laboratory test results and field measurements in different conditions to evaluate the application of the evaluation model. This will be our priority in the future.

Thirdly, it is important to evaluate the long-term effectiveness of this evaluation model. Besides needing more field measurements, it is also necessary to verify the results by the numerical simulation. The regular disease field analysis and remote sensing monitoring can be performed along the Qinghai–Tibet Engineering Corridor (QTEC) to monitor and evaluate the long-term effectiveness of this evaluation model. And the numerical simulation results can predict the ground surface deformation while evaluating the effectiveness of the model.

In general, the results of this paper have a great significance for sustainable development in permafrost regions.

6. Conclusions

In this study, we proposed a model to evaluate the influence of the warming climate on the ground surface stability over permafrost along the Qinghai–Tibet Engineering Corridor (QTEC). This model was established based on the trapezoidal fuzzy Analytical Hierarchy Process (AHP) method. Three main influential factors, including the active layer thickness, the mean annual ground temperature, and the ice content of permafrost, were considered in this model. We calculated and discussed the variation and distribution of the ground surface stability along the QTEC with simulations of the warming climate in the future. The following conclusions can be drawn:

(1)

The ground surface stability influential factors will be greatly changed with the warming climate in the future. The active layer will have an average thickening rate of about 0.021 m/year under rapid climate warming conditions. It is obviously greater than that under slow climate warming conditions. The proportion of low-temperature permafrost will be greatly decreased. And the high-temperature permafrost will be distinctly increased. Some permafrost, even, will decrease to melting soil. And the high ice content permafrost will decrease with warming climate under rapid climate warming conditions.

(2)

The study area will mainly be stable under medium climate warming conditions in 2050. The proportion of unstable area is less than 10%. And it is mainly distributed in the high-temperature permafrost areas between the Chumaer River high plain and Tuotuohe–Yanshiping, while most of the study area will be sub-stable and unstable under rapid climate warming conditions. And the distribution of unstable areas will be spread to Wudaoliang and the Tangula Mountains.

(3)

The ground surface stability will see a great change along the QTEC with the warming climate in 2100. The stable area will be greatly decreased under medium climate warming conditions. Meanwhile, the unstable area will increase. And the amplitude is greater than 10%. There will be an opposite variation trend under rapid climate warming conditions. The stable area will increase. And the unstable area will decrease. This is mainly because some permafrost will be degraded to melting soil.