1. Introduction
The Qinghai–Xizang Plateau (QXP), often known as the “roof of the world,” has been the subject of increasing attention due to unprecedented and rapid environmental change in recent decades
[1]. For example, the permafrost that covers about 40% of the QXP
[2] is experiencing rapid warming and thawing, along with areal shrinkage
[3],
[4]. The rapid change exhibited by the permafrost is clearly altering the ecological and hydrological environment
[3],
[4],
[5], with impacts on vegetation composition, community structure, and succession, as well as shrinking of alpine wetlands, meadows, and steppes
[3],
[6].
Beyond its potential environmental influence, permafrost degradation also threatens the safety of infrastructures positioned on the rugged permafrost terrain of the roof of the world
[7], impacting socioeconomic development and human activity in the remote regions of the QXP. Important infrastructures, such as highways, speedways, railways, power transmission lines, and pipelines, cross the permafrost regions in the QXP
[7],
[8],
[9]. Climate warming, coupling with engineering influence, is accelerating and amplifying the degradation of the permafrost under such engineering structures, increasing the infrastructures’ risk of instability
[7]. Moreover, an anomalous increase in the risk of natural hazards, such as thermokarst slope failure and thermal erosion, threatens these infrastructure
[10],
[11]. Prediction results show that the permafrost will continue to warm and thaw under the scenarios of representative concentration pathways (RCPs) and shared socio-economic pathways (SSPs) in the QXP
[12],
[13]. The risk of infrastructure damage will gradually increase with permafrost degradation, with 30%–50% of the infrastructure expected to be at high risk by 2050
[10]. These impacts pose a great challenge in further considering the influence of climate change on engineering design within permafrost regions
[14].
In this article, we consolidate information and present a comprehensive overview of the impacts of permafrost degradation on the Qinghai–Xizang Railway (QXR) and the adaptation of engineering measures to address these impacts. First, we outline critical problems associated with the QXR construction in permafrost regions; then, we discuss permafrost change and hazards associated with permafrost degradation along the QXR. Subsequently, we report on observed changes in the permafrost beneath the embankment and embankment deformation, as well as geotechnical measures for mitigating permafrost change. Finally, we discuss important issues concerning the effects of climate change on infrastructure in the future.
2. The QXR on permafrost
Construction of the QXR on the rugged permafrost terrain of the QXP presented many challenges, including climate warming, a fragile eco-environment, and the thermal influence of engineering endeavors. The coupled impact of these factors has accelerated and amplified the degradation of the permafrost, given the high spatial variability of permafrost distribution, temperature, and ground ice types
[15],
[16]. Thus, constructing the QXR on permafrost necessitated accurate exploration and the spatiotemporal prediction of permafrost properties with high spatial variability to guide both the design and the construction.
The QXR crosses about 550 km of the delicate and treacherous permafrost region, including 445 km of permafrost and 102 km of thawed area (i.e., subsurface thawed ground)
[17].
Figs. 1(a) and
(b) show proportion of permafrost type with ice content and warm and cold permafrost in the section of each geomorphic unit. Within this expanse, approximately 275 km of permafrost beneath the railway remains within −1 °C (i.e., warm permafrost in
Fig. 1(b)) and half of the permafrost beneath the railway is high ice content permafrost (i.e., with a volume ice content larger than 25%) (
Fig. 1(a))
[17],
[18]. This implies that 122 km of warm permafrost with a high ice content could be particularly susceptible to melting with an increase in temperature of just 1 °C, resulting in a remarkable increase in the thaw settlement of the railway embankment
[7],
[18]. Consequently, the embankment’s position on such warm permafrost with a high ice content may pose a significant threat to the safety of the QXR
[7].
Unstable warm permafrost with massive ground ice exerts a profound impact on the stability of the QXR in the face of climate warming and engineering influence. The possibility of unpredictable change in the thermal state of the permafrost and the thawing of various types and thicknesses of ground ice emphasizes the need for geotechnical engineering techniques that can minimize the associated risks and support the QXR as far as possible. In the initial stage of QXR construction, efforts were made to explore methods for keeping the permafrost cool. Experimental studies were conducted in the Qingshui River, Beilu River, Tuotuo River, and Amdo River sections of the QXR, taking into account the ground temperature and ground ice
[17],
[18]. Most of the methods investigated in these experimental sections to mitigate permafrost degradation ultimately yielded recommendations for several effective strategies
[17]. Among these approaches, the notion of keeping the permafrost under the embankment cool by regulating heat convection, solar radiation, and conduction notably emerged as a prominent design concept (
Table 1)
[18],
[19],
[20].
2.1. Regulating heat convection
A crushed rock structure embankment (CRSE) is an innovative approach that uses the “thermal diode” effect in crushed rocks to modify the thermal convection. This technique capitalizes on the superior heat convection during frigid plateau winters and reduces the efficiency of air convection during the relatively milder summers (
Table 1)
[21]. Variations of this technique include the crushed rock-based embankment (CRBE), the U-shaped crushed rock embankment (UCRE), the crushed rock revetments embankment (CRRE), and others. The utilization of this technique can be traced back to earlier instances of experimental application in railway engineering within permafrost regions in Siberia in Russia
[22] and in China
[23], as well as highway construction or rehabilitation in the permafrost region of interior and south–central Alaska during the 1990s
[24]. These methods gained prominence in the 2000s, as they were widely integrated into the design and construction of the QXR, spanning over 240.7 km in total mileage of each geomorphic unit (
Fig. 1(c))
[20],
[21].
A ventiduct embankment is a proactive method for regulating convection and facilitating active heat exchange in order to cool the underlying permafrost (
Table 1). This approach enables the extraction of heat from beneath the embankment through wind-driven convection, which is achieved by installing a ventilated duct within the embankment
[25],
[26]. The distinct disparity in air temperature between winter and summer, coupled with the high wind speeds in the winter, leads to a significantly greater release of heat from a ventiduct embankment compared with heat absorption, resulting in a net heat release over the course of the year
[27]. During summer, the ventilated duct is blocked to prevent warm air from entering, thereby enhancing the cooling effect by integrating a so-called “wind door”
[28],
[29]. Due to the potential hindrance posed by aeolian sand blocking the ventilated duct, the ventiduct embankment approach has not yet been employed for the QXR; instead, experiments have been conducted to assess its cooling efficacy
[26],
[27].
2.2. Regulating heat conductivity
Thermosyphons (
Table 1) are another “thermal diode” technique; they regulate thermal conductivity via two-phase convection involving gas–liquid interactions within sealed steel pipes. This technique has previously been applied in the construction and retrofitting of foundation soils for oil pipelines, highways, and various buildings within Arctic and subarctic regions
[25],
[30]. This approach has been extensively employed in the QXR, spanning over 43.6 km in total mileage of each geomorphic unit (
Fig. 1(c))
[31]. Furthermore, a combined approach involving both thermosyphons and insulation layers was developed and put into practice within the QXR
[32].
Passive measures—exemplified by elevating the embankment height and utilizing insulation materials (
Table 1)—can be used to delay permafrost degradation
[33]. However, widespread engineering practice has shown that this method cannot fundamentally improve the thermal state of the permafrost beneath the embankment under a warming climate
[33],
[34].
2.3. Regulating heat radiation
Sunshade measures can be broadly categorized into two types: sunshade boards (
Table 1), which reduce the penetration of solar radiation energy into the embankment slope; and awnings, which function similarly to limit solar radiation into the embankment
[35],
[36],
[37]. While sunshades can significantly enhance the thermal conditions beneath the QXR embankment by shading both the embankment itself and its slope, the materials used for such measures are susceptible to damage from the strong winds prevalent in the QXP. This issue, along with related safety concerns and construction costs, limits the practicality of the widespread use of sunshade measures to protect embankment slopes
[38]. Although experimentation with sunshades was undertaken in the early stage of construction for the QXR and the Qinghai–Xizang Highway
[36], this technique has not been extensively adopted in practice.
2.4. Synthesis techniques
The concept of “dry bridges,” which involves replacing the embankment with bridges in regions characterized by extremely warm permafrost and massive ground ice, has proved successful within the QXR, spanning over 59.1 km in total mileage of each geomorphic unit (
Fig. 1(c)). Dry bridges effectively mitigate the thermal disruption caused by embankment construction via the mechanisms of ventilation and solar radiation reflection (
Table 1).
During the early stage of QXR construction, passive techniques such as elevating the embankment height and incorporating an insulation layer within the embankment were primarily employed in permafrost regions. Simultaneously, proactive techniques for cooling the roadbed were developed and evaluated within experimental sections. However, the effect of climate warming prompted a shift in the QXR’s design approach. Considering the varying mean annual ground temperature (MAGT) and ice content of the permafrost, the QXR’s construction strategy was evolved to incorporate diverse techniques for cooling the underlying permafrost (
Fig. 1). These methods encompassed the use of CRSEs
[7] and thermosyphons, as well as replacing embankments with bridges.
3. Permafrost degradation
3.1. Permafrost warming and active layer thickening
The ice-rich permafrost supporting the QXR has undergone rapid thawing and warming, resulting in surface subsidence, thaw hazards, and infrastructure damage
[7],
[16],
[39]. Over the past five decades, the lower boundary of permafrost distribution on the QXP has been generally elevated, ranging from 40 and 200 m, with considerable spatial heterogeneity
[40]. The permafrost continues to warm up. During the 1970s–1990s, the temperature of discontinuous permafrost increased by approximately 0.3–0.5 °C, while that of continuous permafrost increased by about 0.1–0.3 °C
[41].
Following the 1990s, the permafrost warming trend persisted across the QXP. Notable disparities among the different thermal stabilities of the permafrost were observed during 1999–2015 (
Table 2)
[4],
[16],
[42],
[43],
[44],
[45],
[46],
[47]. Active layer thickness (ALT) deepening has been continuously observed (
Fig. 2(a),
Table 2). From 1980 to 2018, the ALT deepened at a rate of 19.5 cm per decade
[42],
[46]. Across all 28 observation sites along the QXR, a significant increase in the ALT (at a 95% confidence level (CI)) of 89% was observed, with an average increase rate of (46.74 ± 26.88) cm per decade. Of the observation sites, 32% showed an extremely significant increase (at a 99% CI), with an average increase rate of (69.7 ± 47.3) cm per decade (
Fig. 2(a),
Table 2). Across all 65 observation sites along the QXR, around 67% exhibited a significant increase in permafrost temperature (at a 95% CI), with an average warming rate at the depth of 6 m of ((0.18 ± 0.12) °C per decade), and at the depth of 15 m of (0.17 ± 0.11) °C per decade. Approximately 43% of the observation sites showed a notably substantial increase (at a 99% CI), with an average warming rate at the depth of 6 m of (0.29 ± 0.19) °C per decade (
Fig. 2(b))
[4],
[16],
[42],
[43],
[44],
[45],
[46], and at the depth of 15 m of (0.28 ± 0.09) °C per decade (
Table 2,
Fig. 2(c)).
3.2. Permafrost thaw hazards
Approximately 250 thermokarst lakes are scattered along the QXR, with a total area of about 1.39 × 10
6 m
2; the average thermokarst lake area is about 5580 m
2. These lakes are predominantly distributed in high plains and gully basins
[48]. Concurrent with permafrost degradation, there has been a notable increase in the occurrence of freezing and thawing hazards, including the formation of new thermokarst lakes, retrogressive thaw slumps, and active layer detachment slides. For example, 52 new retrogressive thaw slumps were identified from Wudaoliang to Fenghuoshan along the QXR
[49]. The thermokarst process occurring on the permafrost slopes is attributed to permafrost degradation, and their impact on engineering is generally minimal, due to their distance from the engineering structures. However, under climate change, engineering-induced phenomena such as retrogressive thaw slumps and active layer detachment slides have emerged, exemplified by incidents such as underground ice exposure resulting from soil extraction at K3035 of the Qinghai–Xizang Highway. These occurrences pose a certain level of threat to the safe operation of the highway
[50]. In the Fenghuoshan area, embankment construction resulted in creep and tension cracks at the rear edge of a gentle slope. Subsequently, abnormal precipitation led to an active layer detachment slide (
Fig. 3(a))
[51], causing the slope to extend nearly 9 m into a culvert. This posed a certain risk to the safe operation of the QXR
[51]. Similar hazards have also taken place in the Fenghuo Mountain along the QXR (
Fig. 3(b)), posing threats to the railway’s secure operation.
4. Permafrost changes under the embankment
The construction of an embankment significantly changes the surface albedo, roughness, and overall transportation coefficient. These changes consequently affect the surface radiation balance and energy balance, leading to substantial shifts in the thermal state of the underlying soil. At the same time, railway ballast paved on a gravel surface can mitigate the thermal state of the underlying soil, with the exception of dispersive pressure
[52],
[53]. Therefore, a gravel surface with ballast, functioning as an opening system, enhances the heat exchange between the surface and the atmosphere. Moreover, due to the wetting and evaporation effect, the infiltration of precipitation into the embankment is limited
[53]. Consequently, the influence of the highway and railway on the thermal state of the permafrost beneath the embankment varies significantly in permafrost regions
[54].
4.1. Permafrost changes beneath a general embankment
A general railway embankment reduces heat intake in summer and heat extraction in winter by increasing the thermal resistance within the embankment
[38], inevitably changing the thermal state of the permafrost beneath the embankment (
Fig. 4). Under the impact of climate change and the thermal disruption caused by engineering, the permafrost temperature beneath the embankment will experience a notable increase, leading to an upward shift of the permafrost table (
Fig. 4(a)). However, a newly formed permafrost table may be unstable. When the newly formed permafrost table begins to decrease and approach the original permafrost table under the natural surface, the original permafrost starts to thaw, leading to thaw settlement of the permafrost beneath the embankment. This process poses a potential risk of instability for the embankment.
4.1.1. Permafrost table change
During the initial operation stage of the QXR during 2005–2010, the engineering disturbances led to a notable rise in the permafrost table, which depended on the thermal stability of the permafrost beneath the embankment. For permafrost with an MAGT colder than −0.5 °C, the permafrost table experienced an average elevation of 2.35 m, with a maximum increase of 5.8 m. For permafrost with an MAGT warmer than −0.5 °C, the permafrost table exhibited an average decrease of 1.63 m, with a maximum of 8.5 m. It is noteworthy that, on the sunny shoulder of the embankment, the permafrost table exhibited a fluctuation of less than 1.0 m for an individual section
[55],
[56]. Under the impact of climate change and the prolonged operation of the QXR, the permafrost table under the general embankment has become unstable. This transformation nearly shifted from a rising trend prior to 2010 to a decreasing trend during 2010–2019 (
Fig. 5)
[56],
[57]. The timing and extent of these changes in the permafrost table are shaped by the interplay among climate change, engineering disturbances, and the thermal stability of the permafrost itself.
4.1.2. Permafrost temperature changes
Climate change and thermal disturbances due to engineering will lead to rapid degradation of the permafrost beneath the embankment, resulting in the permafrost table decreasing and the permafrost warming, which will in turn impact the embankment deformation
[58],
[59],
[60]. For cold permafrost with an MAGT ≤ −1.0 °C, an accumulation of the amount of heat released during the winter can be observed in the shallow permafrost layer beneath the embankment. Consequently, no apparent warming or cooling trend in the permafrost was observable during 2006–2010
[55],
[56]. It was only in 2015 that the cold permafrost shifted from a fluctuation state to a warming trend
[54]. In contrast, warm permafrost does not exhibit the same release of heat during the winter; thus, the shallow permafrost layer shows a distinct warming trend
[56],
[57],
[61]. As a consequence of climate change and thermal disturbances from engineering, the MAGT exhibits a clear rising trend. From 2005 to 2010, the average increase is 0.08 °C, with a maximum of 0.25 °C, surpassing the MAGT under the natural surface
[56].
4.1.3. Sunny-shade slope effect
Embankment construction in the QXP leads to a pronounced sun-shade slope effect
[62], largely due to the variation in the surface temperature and radiation of the embankment slopes. Research has demonstrated that the radiation and surface temperature of the embankment slopes are proportional to the embankment orientation
[62],
[63],
[64]. Due to the amount of heat coming from the railway embankment slopes
[65], the sun-shade effect of a railway embankment is stronger than that of a highway embankment with asphalt pavement
[65]. The difference in the permafrost table beneath the sunny versus the shaded slopes of the embankment ranges from 0.1 to 3.0 m, with an average of 1.13 m
[64]. Similarly, the variation in the permafrost temperature beneath the sunny versus shaded shoulder ranges from 0.23 to 1.58 °C, with an average of 0.86 °C
[64]. Considering the embankment orientation is of paramount importance for the future design of embankment engineering. Recognizing and accounting for the sun-shade slope effect and its influence on the permafrost can significantly impact the stability and performance of the embankment under the QXP’s unique environmental conditions.
4.2. Permafrost change beneath a CRSE
CRSEs—including the various structural forms such as CRBEs, UCREs, CRREs, and others—are designed to create a “thermal diode” effect that optimizes convection during the winter while minimizing its effectiveness during the summer (
Fig. 4(b))
[21],
[38]. Starting in the 2000s, this method was extensively applied in the design and construction of the QXR, covering over 240.7 km of the railway, and has undergone systematic study. Among the variations, the UCRE has proven to be superior to the CRBE and the CRRE in terms of its cooling effects on the permafrost beneath the embankment
[66],
[67]. The CRESs approach effectively leverages the thermal properties of the crushed rock structure to regulate the permafrost temperature and promote thermal stability.
4.2.1. Permafrost table change
During the initial stage of QXR construction, the implementation of CRSEs had a substantial impact on the permafrost beneath the embankment. Based on observation data of permafrost temperature under the CRSEs (
Fig. 6(a)), this effect was particularly evident in the significant increase in the permafrost table beneath the embankment, accompanied by a notable decrease in the temperature of shallow permafrost. It is important to note that the cooling amplitudes of warm and cold permafrost exhibit significant differences
[55],
[56],
[66]. Under the effects of climate change and engineering disturbances, the permafrost table beneath a CRSE dynamically changes during different operating stages, involving significant rise, stability, and decline in the permafrost table
[55],
[68]. Under a sunny slope, the permafrost table beneath a CRSE may continue to experience fluctuation, with some cases exhibiting rising trends while others showing a decrease (
Fig. 6(b))
[69]. However, the amplitude of these changes is relatively small. Conversely, under a shaded slope, the permafrost table beneath a CRSE tends to consistently rise (
Fig. 6(c)). Importantly, the present permafrost table is lower than the permafrost table that is newly formed during the early stage of CRSE implementation
[69]. These results suggest that the long-term effect of CRSE on the permafrost table is essentially stable, and the influence of climate warming remains relatively modest. This finding underscores the effectiveness of CRSEs in mitigating the impact of temperature changes on the thermal stability of the permafrost beneath the embankment
[56],
[68].
4.2.2. Permafrost temperature change
A rise in the permafrost MAGT has been demonstrated with CRSE construction; however, the magnitude of this rise is notably smaller than that of the permafrost MAGT under the natural surface
[56],
[57],
[70]. Extensive data has consistently shown that the cooling effect of CRSEs varies significantly between warm and cold permafrost regions
[56],
[69],
[70]. Cold permafrost beneath a CRSE undergoes rapid cooling during the early stage of embankment construction (3–4 years), followed by a gradual cooling trend as the operation progresses (10–15 years)
[56],
[67],
[69],
[70]. In contrast, warm permafrost exhibits a more gradual cooling process, with the permafrost beneath the CRSE cooling down slowly during the initial years of embankment construction (3–4 years) and exhibiting an increasingly evident cooling trend over time (10–15 years)
[66],
[69]. The depth at which the CRSE’s cooling effect impacts the permafrost varies depending on the thermal stability of the permafrost itself
[70]. For a UCRE, the impact depth of the cooling effect is up to 10 m in warm permafrost and up to 14 m in cold permafrost. For a CRBE, the impact depth is around 5 m in warm permafrost to 10 m in cold permafrost. Finally, for a CRRE, the impact depth is about 10 m in cold permafrost, while no specific data is available for warm permafrost
[69],
[70]. These variations in the impact depth emphasize the intricate interaction between the thermal properties of the CRSE and the characteristics of the underlying permafrost.
Under the impacts of climate change, the permafrost beneath both the natural surface and the general embankments is experiencing significant warming. In contrast, the implementation of CRSEs has demonstrated the CRSE’s ability to maintain the coolness of the permafrost table beneath it. Remarkably, even in cases of extremely unstable permafrost (MAGT > −0.5 °C), CRSEs have exhibited the capacity to cool shallow permafrost. These findings underscore the robust adaptability of the CRSE’s cooling effect under changing climate conditions. The cooling effect observed with CRSEs is consistent under varying degrees of permafrost thermal stability and is effective in counteracting the warming trend experienced by permafrost beneath the natural surface and general embankments. This adaptability of the CRSE to climate change highlights its potential significance in the context of permafrost preservation and infrastructure stability within warming permafrost regions.
4.2.3. Sunny-shade slope effect
The CRSE can modulated the sunny-shade slope effect of the embankment, significantly decreasing the difference in the permafrost table and permafrost temperature under the sunny slope and the shaded slope
[64]. During QXR construction, a CRRE structure with a different thickness of the crushed rock layer under the sunny-shade slope embankment was proposed to address the sunny-shade slope effect
[71]. Numerical simulation results showed that a crushed rock revetment with different thicknesses of 80 cm under the sunny embankment slope and 160 cm under the shaded embankment slope could eliminate the sunny-shade slope effect
[71]. In comparison with a general embankment, a CRSE can reduce the temperature difference between the sunny slope and the shaded shoulder at a depth of 0.5–3 m by 0.60 °C on average, as well as reducing the permafrost table difference by at least 0.34 m on average
[64]. However, it is worth noting that a CRSE can cause the mean temperature difference under the embankment between winter and summer to decrease by 1.6 °C during the winter but increase by 0.8 °C during the summer.
4.2.4. Numerical modeling of climate change impacts
Currently, it is difficult to use monitoring data of the permafrost temperature to determine the long-term cooling mechanics of CRSE. Numerical simulation has thus emerged as an important method to study the long-term cooling mechanics of CRSE
[72],
[73]. Studies using numerical simulation have revealed that, under a warming climate scenario of 2.0–2.6 °C over the next 50 years, the permafrost at a depth of 5 m beneath a CBRE can maintain a frozen state during the summer, even in permafrost regions with air temperatures ranging from −3.5 to 4.0 °C
[72],
[73],
[74],
[75]. However, it is important to note that the crushed rock layer can become obstructed by aeolian sand, potentially leading to a decrease in its long-term cooling effect
[76],
[77]. While the impact of aeolian sand on the cooling effect may weaken after around 20–25 years, it might strengthen again after 50 years
[76],
[77]. Overall, numerical simulations suggest that climate warming could have a slight diminishing effect on the cooling efficiency of CRSEs in permafrost regions with a mean annual air temperature (MAAT) lower than −4.0 °C. These findings are partially supported by monitoring data and provide valuable insights into the long-term behavior of CRSEs in response to a warming climate.
4.3. Cooling effect of thermosyphons on underlying permafrost
Since their development for the QXR, thermosyphons have been employed for 20 years in various applications in permafrost regions, including highways, power transmission lines, and pipelines in both the QXP and northeast China
[78],
[79],
[80],
[81]. This technique has been recognized as a crucial measure for adapting to climate change. Thermosyphons have been implemented in approximately 43.6 km of embankment in order to cool the underlying permafrost
[82] and have demonstrated the capability to rapidly cool the permafrost and elevate the permafrost table beneath the embankment. After an initial period of around five years, the permafrost temperature beneath the embankment tends to stabilize
[82]. Moreover, the effectiveness of thermosyphons is influenced by their proximity to the embankment. The closer the thermosyphon is to the embankment, the more pronounced the decrease in the permafrost temperature and the increase in the permafrost table elevation are
[83]. Even under the impact of climate warming, thermosyphons can continue to fulfill their role in cooling the permafrost beneath the embankment. In permafrost regions with a MAAT lower than −3.5 °C or a mean annual surface temperature (MAST) lower than −1.0 °C, thermosyphons have demonstrated the ability to adapt to a climate warming impact of 1.0 °C over 50 years, effectively ensuring that the permafrost beneath the embankment remains frozen
[32]. This demonstrates the valuable contribution of thermosyphons in maintaining the stability and integrity of the infrastructure in permafrost regions in the face of changing climate conditions. However, few analyses have been conducted on the long-term cooling effect of thermosyphons.
4.4. Reinforcement measures for permafrost thermal stability
To counteract the detrimental effects of climate change and engineering activities on permafrost degradation, a range of reinforcement measures have been implemented on the QXR that are specifically designed to address the challenges posed by permafrost degradation. These measures are intended to mitigate embankment deformation caused by permafrost thawing and warming in order to enhance the stability of the railway infrastructure. These specialized reinforcement measures go beyond traditional engineering approaches and consider the unique challenges presented by permafrost conditions and climate change. By utilizing techniques such as thermosyphons, crushed rock revetments
[57],
[84], and new structural enhancements, the QXR’s engineers aim to ensure the long-term stability and safety of the railway infrastructure in permafrost regions
[57],
[84]. New structural reinforcement measures include adding crushed rock berms beside CRBEs or CRREs
[68],
[85],
[86], and adding crushed rock berms beside crushed rock revetments paved with soil
[87].
The effectiveness of crushed rock revetments as a reinforcement measure for cooling the underlying permafrost appears to be a complex subject of debate. Monitoring data suggests that a crushed rock revetment can indeed mitigate the underlying permafrost degradation after the embankment is reinforced, but permafrost degradation may still occur 3–4 years later
[57]. However, various factors need to be considered, especially in situations involving permafrost with a high ice content. Studies indicate that, while a crushed rock revetment can offer evident benefits in terms of permafrost degradation after reinforcement, it might also potentially reduce the cooling effect on permafrost with high ice content. This highlights the importance of tailoring reinforcement strategies to the specific characteristics of the permafrost conditions in each area.
Furthermore, numerical simulations have provided insight into the long-term effectiveness of crushed rock revetments as a reinforcement measure. It appears that, while the use of a crushed rock revetment can stabilize the permafrost’s thermal regime for a certain period (e.g., 35−45 years after 10 years of operation), there are limitations to its long-term impact
[86]. The application of reinforcement measures such as thermosyphons or combined measures such as a crushed rock revetment with a thermosyphon can significantly cool the underlying permafrost but will significantly increase the engineering cost as well
[57],
[84]. Some new reinforcement measures might be challenging to implement on the QXR, although numerical simulations have verified their cooling effect on the underlying permafrost
[84],
[87],
[88]. Overall, the selection of reinforcement measures for cooling the permafrost must be carefully considered in the context of the specific permafrost conditions and the associated trade-offs. While crushed rock revetments can be effective, there is a need for a comprehensive analysis and for the recognition that certain reinforcement methods might be more suitable for specific situations. The ultimate goal is to strike a balance between effective reinforcement, long-term stability, and cost considerations in the challenging permafrost environments of the QXR.
5. Observed infrastructural damage
5.1. Engineering stability
Due to the influence of climate change and engineering activities, permafrost thawing and warming have triggered alternations in engineering stability. These alternations have consequently given rise to deformations—a structural issue in engineering—including thawing settlement, uneven deformation, and longitudinal cracking.
The deformation observed in embankment engineering in permafrost regions primarily stems from soil frost heave and permafrost thaw settlement. To illustrate, in the case of the Qinghai−Xizang Highway, soil frost heave deformation accounts for 15%, while thaw settlement contributes to approximately 85% of the deformation
[89]. For the QXR, frost heave deformation mainly occurs in the permafrost regions where the MAGT is less than −1.0 °C. This kind of deformation occurs during the initial formation of the permafrost subsequent to embankment construction
[60]. Thawing settlement deformation involves multiple factors, including thaw settlement of the permafrost, thawed soil compression deformation, and compression and creep of warm permafrost
[58],
[90]. The deformation caused by permafrost thawing due to climate change and engineering disturbances is the primary contributor, constituting approximately 80% of the total deformation
[91],
[92]. Embankment deformation is mainly related to the thermal stability of the permafrost beneath the embankment. For cold permafrost with an MAGT ≤ −1.0 °C, soil frost heave and permafrost thaw settlement will occur simultaneously
[60]. Generally, frost heave deformation ranges from 5 to 10 mm per year, while thaw settlement deformation remains at less than 10 mm per year, with a maximum rate of 25 mm per year
[91],
[92]. Conversely, for warm permafrost with an MAGT > −1.0 °C, thaw settlement deformation is the predominant form of deformation, with a deformation rate of 10–20 mm per year
[91]. Notably, differential thaw settlement deformation occurs between the sunny and shaded shoulder
[91].
The cooling measures of CRSEs can effectively reduce the thaw settlement deformation of the embankment, as well as the differential thaw settlement deformation between the sunny and shaded shoulder
[91],
[92]. Until 2018, among the 34 deformation monitoring sections, thaw settlement deformation of only one section exceeded the allowable deformation, and thaw settlement deformation greater than 100 mm basically occurred in warm permafrost regions with a high ice content
[93].
5.2. Engineering damage
Since the operation of the QXR began in 2006, several so-called “engineering damages” have occurred in permafrost regions, including: surface water at the slope foot due to supra-permafrost
[94]; longitudinal cracking of the embankment slope
[95]; ballast subsidence and slide collapse
[96]; and unstable embankments with a high slope
[96]. From 2010 to 2018, the engineering damage of the QXR were addressed along an integrated maintenance section about 55 km long; as a result, the length of railway affected by engineering damage was reduced from 7.60 km in 2010 to 0.48 km in 2018
[97]. Nonetheless, the QXR’s susceptibility to engineering damage remains notably high, which can be attributed to the ongoing degradation of the underlying permafrost. The sections considered to have a medium to high risk of experiencing such damage may encompass up to 77 km
[97].
In addition to embankment-related damages, there has been a notable occurrence of incidents within the transition sections connecting embankments to bridges. A survey conducted in 2009 encompassing 164 bridges and 656 embankment–bridge junctions revealed significant findings. In transition sections spanning approximately 220 km, embankment deformation surpassed 7 cm on average
[95]. Further investigations identified slow settlement phenomena within a 50–100 m distance from the bridges. More specifically, within about 161 transition sections with embankment–bridge connections, evident deformations were observed. These deformations ranged from 10 to 40 cm, with a maximum of more than 160 cm
[98], seriously threatening the safe operation of the QXR.
6. Projected impacts
6.1. Permafrost change
Projections of permafrost changes over the QXP under various emission scenarios (i.e., the Special Report on Emissions Scenarios (SERS), RCPs, and SSPs) have been the focus of research
[12],
[99],
[100],
[101]. It is important to note that, while these models provide valuable insights, they also come with significant uncertainty
[4]. The consensus from these projections is that the permafrost over the QXP is expected to clearly degrade, characterized by an increase in the ALT, permafrost warming, and a shrinkage of permafrost areas. However, limited research has been conducted regarding the projection of permafrost changes along the QXR. In the early stage of QXR construction, the trajectory of permafrost change was studied by considering potential future climate warming of 1.0 and 2.0 °C over the next 50 years
[102]. It was found that, under the impacts of climate change, with a warming of 2–4 °C per 100 years, the thermal stability of the permafrost along the QXR would be altered, leading to a notable increase in areas with unstable permafrost
[102]. Under the climate scenario of the coupled model inter-comparison project 5 (CMIP5), the projection indicated that the permafrost along the QXR will experience significant degradation, including permafrost warming and ALT increasing. By the year 2050, approximately one-third of the permafrost regions along the QXR are expected to have a high risk of thawing settlement, and about 50% of the permafrost regions will have a significant impact on the stability of the QXR
[103]. Using the coupled model inter-comparison project 6 (CMIP6) climate scenario, it was projected that 12%–20% of the permafrost along the QXR will experience evident degradation, the MAGT is anticipated to warm by 0.4–2.3 °C, and the ALT is predicted to increase by 0.4–7.3 m
[100].
6.2. Risk of permafrost thaw
Permafrost changes on the QXP have led to a significant increase in infrastructure-related damages
[104],
[105]. Various scenarios, including SERS, RCPs and SSPs, have been used to assess the risks associated with the thawing and warming of the permafrost, changes in the bear capacity of the permafrost, and freezing–thawing hazards at both the scale of the QXP and the regional scale
[13],
[99],
[100],
[101]. For example, using the historical SSP2-4.5 scenario, it was projected that approximately 40% of the infrastructure on the QXP will be situated in high-risk regions. This could lead to an additional economic cost of 27.5 billion CNY
[105]. By the end of the century, this percentage is expected to rise to nearly 80%, necessitating an additional maintenance cost of 43.5 billion CNY
[105]. The implications of these findings underscore the growing urgency of addressing permafrost-related challenges and their potential impact on infrastructure on the QXP.
Indeed, the vulnerability of thaw-induced hazards, such as thermokarst lakes and thermokarst slope failure, has been assessed in the Qinghai–Xizang engineering corridor
[11],
[106],
[107]. These assessments have shed light on the susceptibility of various regions to such hazards. In particular, in the Qinghai–Xizang engineering corridor, approximately 20% of thermokarst lakes exhibit a high to very high susceptibility to these events
[13]. From Wudaoliang to Fenghuo Mountain, around 23% of the slopes have been identified as having a high susceptibility to thaw-induced slope failure
[106]. Over the QXP, it has been estimated that approximately 34% of thaw-induced sink phenomena are located in high-risk regions. Notably, the area from Budongquan to Beilu River along the QXR is characterized by a very high risk of thaw-induced sink occurrence
[108]. However, it is important to highlight that these assessments predominantly consider the impact of thaw-induced hazards caused by current climate change and might not account for potential changes in susceptibility due to future climate change—which could further alter the risk landscape of these permafrost-related phenomena.
In the context of the QXR, assessments of infrastructure risks focus on engineering damages occurring in the transition section connecting embankments to bridges. However, these risk assessments often center on permafrost thawing and warming caused by engineering disturbances and typically do not consider the potential impacts of climate change. Studies have indicated that, among the 890 transition embankment–bridge sections, approximately 30% are located in high-risk regions, while 40% fall in middle-risk regions
[85]. Limited research has been conducted on assessing the risk of CRSE serviceability in the QXR under future climate change
[109]. Based on this research, UCREs and CRBEs could be used to adapt to the impacts of a climate warming of 2 °C per 50 years, with the failure risk of engineering serviceability remaining minimal until the end of the century
[109].
In summary, while there has been a focus on assessing risks related to infrastructure on the QXR, there is a need to consider both the impact of engineering disturbances and the potential consequences of climate change, particularly when evaluating the susceptibility of various sections and types of infrastructure.
7. Summary and future perspectives
Over the past two decades, the permafrost along the QXR has been experiencing rapid degradation, with the ALT deepening and the permafrost warming. This degradation has led to a noticeable rise in the risk of and susceptibility to thaw-induced disasters. Both climate change and engineering disturbances have played a role in the permafrost degradation process. The permafrost beneath general embankments of the QXR has undergone rapid degradation, which has been more pronounced beneath the embankment compared with under the natural surface. Various measures have been employed to address these challenges, such as CRSEs and thermosyphons. These measures have demonstrated their ability to mitigate the underlying permafrost degradation by maintaining a cooling effect over the long term. However, despite these measures, the coupled impacts of climate change and engineering still lead to thaw settlement deformation. While the over-deformation of the embankment remains within the allowable deformation, a notable concern arises regarding embankment–bridge transition sections. Here, the embankment deformation in 50% of these sections exceeds the allowable limits, highlighting the need for additional reinforcing measures to counteract the embankment deformation induced by the combined impacts of climate change and engineering activities. As permafrost in the QXP continues to undergo changes in response to climate change, permafrost degradation beneath the embankment may accelerate due to the coupled impact of climate change and engineering, actively increasing the risk of thermal thaw hazards. Thus, the safety of the QXR remains a concern. Effective reinforcement measures are imperative to mitigate the adverse effects of climate change and ensure the long-term stability of the railway infrastructure.
The engineering approach of cooling the underlying permafrost that was implemented during the construction of the QXR has been proposed as a means of mitigating the impacts of climate warming. This approach aims to counteract the effects of warming by maintaining a cooler environment for the permafrost. This strategy diverges from the methods proposed by Russian scientists (i.e., Principles I, II, and III) and from the “proactive and positive protection” principle advocated in North America
[14],
[38]. Different embankment structures have been explored to achieve this cooling effect and mitigate permafrost degradation. However, it can be challenging to achieve a sustained and long-term cooling effect though these measures. As a result, there is a growing need for a more effective method to adapt to the potential impacts of climate change in the future. While combination methods of cooling the underlying permafrost have been attempted—such as various hybrid-thermosyphon embankments and the incorporation of ventilation ducts in embankments, with features including crushed rock interlayers, hollow concrete bricks, and thermosyphons
[110]—these measures will obviously increase the engineering costs. As our understanding of permafrost dynamics and the effects of climate change continue to evolve, ongoing research and innovation will be essential in developing more efficient and cost-effective methods for maintaining the stability of infrastructure such as the QXR.
High-resolution climate data, including variables such as air temperature, surface temperature, and precipitation, is crucial for accurately predicting permafrost change and assessing the risk of engineering damages. However, there is currently a scarcity of weather stations in the permafrost regions along the QXR. Only three national weather stations have been established in these regions; moreover, while an additional nine weather stations were established during 2000–2010
[46], data from these stations might not be readily available. Therefore, there is a pressing need for more weather stations that can share their data to support the operation of the QXR. Fortunately, reanalysis data such as the China Meteorological Forcing Dataset (CMFD) (1979–2018) can be utilized as an alternative. These data, which have a spatial resolution of 10 km
[111], can serve as inputs for modeling permafrost change
[101]. For a more detailed examination of the impacts of climate change on the permafrost along the QXR, a high-resolution reanalysis dataset downscaled from the CMFD, with a spatial resolution of 1 km, can be employed
[112]. By utilizing these data, researchers can better understand the relationship between climate factors and permafrost changes, allowing for more accurate predictions and risk assessments related to engineering damages along the QXR.
Indeed, in order to ensure the safety of the QXR in permafrost regions, comprehensive monitoring networks are essential to monitor not only high-resolution climate data but also the permafrost temperature beneath the natural surface and embankments, as well as embankment deformation
[113]. However, current coverage is limited, with only about 43 monitoring sections and three weather stations established along the QXR. Moreover, critical parameters such as the soil moisture and heat flux within the active layer and the embankment have not been systematically measured. The lack of data on the hydrothermal exchange between the embankment and the atmosphere—especially the role of ballast in the freezing/thawing process—presents a challenge, as this deficiency in understanding hinders accurate prediction of permafrost change and embankment deformation. With a limited number of monitoring sections and weather stations, along with the absence of comprehensive hydrothermal process monitoring, the ability to anticipate permafrost change and associated infrastructure risks is compromised. To address this gap, more monitoring sites must be established. By expanding the monitoring network and collecting detailed data on the permafrost and hydrothermal dynamics, researchers and engineers will be better equipped to predict permafrost changes, assess potential deformation, and ensure the safety and stability of the QXR.
Developing more accurate numerical simulation methods is crucial for forecasting permafrost change and deformation in the context of the QXR, especially considering the constituents of the frozen soil, thermo–hydro–mechanical coupling, and permafrost parameters in permafrost geotechnical models. These methods, when built upon a foundation of extensive monitoring data on the thermal regime of the permafrost, can yield valuable insights and inform risk assessments regarding engineering safety. Combining monitoring data with numerical simulation analysis will allow for a comprehensive assessment of the potential associated with permafrost engineering under future climate change. Such an assessment would encompass a range of factors, including the impacts of different technical roadbed and engineering measures. In particular, it is essential to enhance the accuracy of projections related to climate change and subsequent permafrost change over the next 10–20 years. This timeframe is especially relevant for planning and ensuring the long-term stability of infrastructure. By refining and calibrating numerical simulations with monitoring data, researchers and engineers can create more reliable models for predicting permafrost behavior and associated engineering risks. This, in turn, will aid in making informed decisions and implementing necessary measures to safeguard the QXR and other infrastructure in permafrost regions.
During the early stage of the construction of the QXR, the concept of a “digital roadbed” for the QXR was proposed
[114]. This concept involved the establishment of a geographic information system for the QXR
[115], integrating data, modeling, and a simulation platform
[114]. In the future, we expect to build an intelligent QXR, including a digital roadbed, big data related to permafrost engineering geology, engineering design (especially in the maintenance of the QXR), and an artificial intelligence model for engineering risk assessment. This approach aligns with the broader trend of digitalization and intelligent infrastructure development. By leveraging advanced technologies and data-driven insights, it will be possible to optimize the performance of the QXR, enhance decision-making, and respond effectively to the challenges posed by permafrost dynamics and climate change. This forward-thinking approach holds the potential to revolutionize the way in which the QXR operates and serves as an example of how technological innovation can be harnessed for the greater good.
CRediT authorship contribution statement
Qingbai Wu: Writing – original draft, Writing – review & editing. Wei Ma: Writing – review & editing. Yuanming Lai: Writing – review & editing. Guodong Cheng: Writing – review & editing.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors would like to thank Dr. Hjort, Professor of Geography Research Unit, University of Oulu, Oulu, Finland, for providing helpful comments on the manuscript. This research was financially supported in part by the Second Tibetan Plateau Scientific Expedition and Research Program (STEP) (2021QZKK0205) and the National Natural Science Foundation of China (42230512).
PDF
(2841KB)
