*4.2. Comparison and Discussion on Deformation Law of Road Section*

#### 4.2.1. Comparison of Different Deformation of Roadbed Frozen Soil

In order to compare the different time series deformation laws of WB and TZ deformation sections and understand their differences in climate response, we obtain the surface time series deformation information from the deformation risk area, as shown in Figure 8. At the same time, in order to more accurately judge the response of time series deformation and climate change, we take the actual distance from each risk area to the meteorological station as the weight factor to obtain the weighted average temperature and precipitation.

In addition to the thawing and collapse of frozen soil in WB section, at the west side of the railway passing through Kunlun Mountain, the central latitude and longitude are 35.67944◦ N and 94.04924◦ E. The roadbed has been raised due to frost heaving of some frozen soil, with cumulative deformation of 50.507 mm and annual average deformation of 20.09 mm/a. From January to early March, the ground surface is constantly lifted due to the frost heaving of frozen soil. In summer, the frost heaving slows down due to the increase in temperature and precipitation, which is basically in a constant trend. In autumn and winter, when the temperature decreases, the surface temperature also decreases. The water in the active layer of frozen soil under the surface solidifies, and the frozen soil heaves. The maximum value of frozen soil heave is rising, as shown in Figure 8c. This shows that the active layer thickens, and because it is located on the shady slope, the frozen soil layer has good development conditions, which belongs to the developing permafrost.

The settlement area of TZ section is rapidly affected by the climate. From the end of April, the precipitation and temperature began to rise, the frozen soil active layer gradually began to melt, and the surface settlement is serious. Its trend line is shown in Figure 8d. The annual average time series deformation rate is 38.47 mm/a, the cumulative deformation variable is 86.39 mm, and the latitude and longitude of the deformation center are 32.8847◦ N and 91.5283◦ E. The seasonality of the climate is seriously affected by the

southeast monsoon. Most of the precipitation occurs from June to August in the form of rainstorm, which leads to flash floods and extensive surface erosion. At the same time, it also leads to the intensification of the melting of frozen soil active layer. The remaining precipitation occurs in the form of snow or hail. Sometimes snow will be generated during the ice period (usually 7–8 months, from September to April of the next year), which is consistent with the deformation. From September to April of the next year, the deformation will obviously slow down, while from June to August, the frozen soil will continue to melt under the influence of temperature and precipitation, resulting in intensified surface settlement. As the settlement area is located in the roadbed section of Qinghai–Tibet railway, it poses a threat to railway operation.

**Figure 7.** (**a**–**c**), respectively show the annual average deformation, topographic and temporal deformation of railway roadbed. The red arrows in (**a**,**b**) show the runoff formed by the thermal melting of frozen soil and the boundary of different deformation mechanisms at the north and south ends. The red box indicates the roadbed range of the railway section. The abscissa in (**c**) represents the distance of p3–p1 point from north to south, and the ordinate is the deformation.

**Figure 8.** (**a**,**b**) depict the distribution of roadbed deformation in WB and TZ sections, respectively, (**c**,**d**) depict the time series deformation of railway roadbed in WB and TZ sections under the influence of climate factors, respectively. The green box in the optical diagram is the selection point of time series deformation, and the red dotted line in the broken line diagram is the deformation subsection fitting line.

#### 4.2.2. Comparison of the Latest Available Results with Previous Results

Because Beilu River and Tuotuo River are located in the basin area, the river system has a great impact on the frozen soil, and the frozen soil section often becomes a research hotspot. This paper compares the deformation of the reach from Wudaoliang to Tuotuo with the study of Zhang et al. from 2009 to 2018, as shown in Figure 9. The three places C1, C2 and C3 pass through Beiluhe basin, Fenghuoshan area and Tuotuohe basin, respectively. C1 and C3 are located in the valley basin, rich in water resources and geographical environment, and there is no high mountain shelter. The frozen soil is greatly affected by thermal melting and forms a large number of thermal melting lakes, as shown in Figure 10a,b. Therefore, these two areas are scattered point deformation areas in InSAR detection results. There is no obvious frozen soil collapse at the railway roadbed. However, under the global warming environment and the expansion of inland rivers and thermal melting lakes, the frozen soil under the railway roadbed will also be affected, and there is still a certain risk of thawing collapse. The deformation at C2 is small, and it has been significantly improved compared with the results in 2018. This area not only takes heat dissipation measures for frozen soil, as shown in Figure 10c, but is also located in high mountain and valley area (Figure 10d), with an altitude of 5000–5200 m and a height drop of 400–600 m with the railway section. The risk of frozen soil collapse is further reduced. In short, although there is no serious deformation of railway roadbed at C1, C2 and C3, considering that there are still many thermal melting lakes near C1 and C3, it is still necessary to monitor the frozen soil section.

**Figure 9.** (**a**) shows the annual average time series deformation of the railway corridor from Beilu River to Tuotuo River in 2020, and (**b**) shows the deformation detection results of Zhang et al. in 2018. The three deformations detected by Zhang are marked with circles in the Figure, namely C1, C2 and C3. The deformation area obviously decreases, especially at C2. No obvious deformation is detected at C1 and C3 railway roadbed.

#### *4.3. Uncertainty Analysis of Results*

Due to the destructive effects of atmospheric precipitation (especially snowfall), the formal application of InSAR technique to monitor structures generating scattering may provide incorrect results. The corrupting impact of atmospheric precipitation on the phase of reference targets has been mentioned in many studies. In winter, the main source of alteration for the propagating signal properties is the growth in snow depth between the SAR observations [67].

As shown in Figure 10, part of our study area is covered by snow because it is located in the Qinghai–Tibet plateau. Snow can affect radar interference and skew results. Therefore, in addition to removing atmospheric phase, terrain phase and noise, we also need to remove the influence of snow removal. As the influence of snow in the main deformation area is small, this study does not deal with the influence of snow temporarily. However, the impact of snow removal will be the focus of our next study.

**Figure 10.** In (**a**,**b**), there are widely distributed hot melt lakes formed by ground collapse after frozen soil melting. (**c**,**d**) are the topographic distribution and solutions of Fenghuo Mountain section, respectively. It can be seen that hot rods are arranged on the left and right sides of the road to dissipate heat and reduce the impact of frozen soil hot melt and frost heaving on the roadbed.

### **5. Conclusions**

In this study, four Sentinel-1 satellite maps are selected to fully cover the study area, with a total of 122 images. The MT-InSAR technology is carried out for the 610 km-long Qinghai–Tibet railway section (from Naij Tal railway station to Anduo railway station) in the permafrost area so as to obtain the time series deformation information of the surface along the Qinghai–Tibet railway and compare its deformation with climate factors. The main conclusions are as follows:

(1) The areas with serious deformation of the Qinghai–Tibet corridor are mainly distributed in the railway section from WangKun station to Budongquan station and the section from Tanggula station to Za'gya Zangbo station, and there are many areas of railway roadbed subsidence and mountain collapse.

(2) The influence of the frozen soil section from WangKun station to Budongquan station on the railway roadbed is high in the middle and low at both ends. The influence of human activities of railway operation on the frozen soil deformation is smaller than that of topography and hydrothermal. At the same time, the geological strata and fault zone of this section also have a certain impact on the roadbed deformation. D2S, OS, and S3Q frozen soil layers are more stable than Qp2gl and Qp3gl strata.

(3) Between the Tanggula and Za'gya Zangbo station, there was a 620 m-long railway roadbed with uneven deformation on the east and west sides, with an average annual difference of 60.68 mm/a. At the same time, uneven deformation also exists in the railway roadbed on the north and south sides.

(4) Through comparison, it is determined that the roadbed deformation does not exist in the area from Beilu River to Tuotuo River, and the permafrost has no great impact on the railway roadbed.

(5) Under the situation of global warming, the frozen soil will continue to undergo thermal thawing and frost heaving. At the same time, the permafrost will continue to decrease and the active layer will continue to thicken. In order to ensure the stable operation of the Qinghai–Tibet railway in the permafrost section, it is necessary to regularly monitor the deformation of the permafrost area up to 610 km by MT-InSAR.

At the same time, our research has some shortcomings. The impact of snow cover on InSAR interference in the study area needs further treatment. This is the focus of our next research work.

**Author Contributions:** H.L. and S.H. performed experiments, analyzed the data and prepared the manuscript. C.X. and B.T. provided crucial guidance and support through the research. M.C. significantly contributed to the validation work and data interpretation. Z.C. provided valuable suggestions for this study. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Fujian Provincial Science and Technology Project (Science and Technology Service Network Initiative, CAS) (2020T3011), jointly funded by the Outstanding Youth Science and Technology Program of Guizhou Province of China ([2021]5615).

**Data Availability Statement:** The data supporting the findings of this study are available from the first author (H.L.) upon reasonable request.

**Acknowledgments:** The authors would like to thank ESA (European Space Agency) and Alaska satellite facility (https://asf.alaska.edu/, accessed on 30 July 2021) for providing the Sentinel-1 datasets of the Copernicus mission.

**Conflicts of Interest:** The authors declare no conflict of interest.
