Dynamic variations in thermal regime and surface deformation along the drainage channel for an expanding lake on the Tibetan Plateau

ABSTRACT

The outburst of Zonag Lake in 2011 triggered a series of floods in the continuous permafrost region of the hinterland of the Qinghai-Tibet Plateau. This re-distributed the surface water in the basin and caused rapid expansion of the tail lake (Salt Lake). To avoid potential overflow of the expanding Salt Lake, a channel was excavated to drain the lake water into a downstream river. In this study, to investigate the permafrost thermal regime and the surface deformation around the expanding Salt Lake and the channel, in-situ monitoring sections were settled from Salt Lake to the downstream of the channel to obtain the permafrost temperature. Additionally, using small baseline subset interferometric synthetic aperture radar (SBAS-InSAR), the surface deformation around Salt Lake and the channel was measured. The data showed that the ground temperature at the channel was 0.6°C higher than the natural field and the mean subsidence rate around the channel was 1.5 mm/yr higher than that at Salt Lake. These results show that the permafrost temperature in the study area changed considerably with variations in the distance from the lake/channel, and the deformation in the study area was dominated by subsidence.

KEYWORDS:

• Permafrost environment
• ground temperature
• SBAS-InSAR
• climate change

1. Introduction

Rock or soil with a temperature below 0°C and containing ice is called frozen soil, and is referred to as permafrost if the frozen state maintains for at least two years (Mu et al. 2020). In the Northern Hemisphere, permafrost is mainly distributed in the Arctic region and the Tibetan Plateau (TP). Compared with permafrost in high-latitude regions, permafrost on the TP has higher temperatures, is thinner, and responds more drastically to climate change and anthropogenic activities (Luo et al. 2019; Wang and French 1994). With profound climate warming in recent decades, the increasing rate of air temperature on the TP has been approximately 0.35/10a, which is twice the global average rate (DChen et al. 2015). Additionally, the occurrence of a series of extreme warming events has accelerated permafrost degradation, which has impacted engineering construction, ecology, and hydrological environments, posing a new challenge for the sustainable development of the TP (Chen et al. 2014; Mu et al. 2014; Zhu et al. 2022).

Known as the “Asia Water Tower” and “Third Pole,” the TP is the source of several large rivers, and lakes on the TP account for 51.4% of the total lake area in China (Ma et al. 2011). In recent years, lakes on the TP have experienced an expansion process along with rapid climate change, which has accelerated permafrost degradation and induced several natural hazards, threatening the safety of human lives and property (Wang et al. 2015). Researchers have revealed variations in the permafrost in the context of climate warming and wetting (Liu et al. 2022). However, most studies have focused on railways, highways, and energy transmission lines, while studies on hydraulic engineering and surface water changes have been limited to permafrost regions (Ma, Niu, and Mu 2012; Mu et al. 2016, 2014; Zhang et al. 2021).

To quantify variations in permafrost, in-situ monitoring and remote sensing techniques have been widely used on the TP. Based on temperature data obtained from various boreholes from the lake center to the natural field, Niu et al. (2011) indicated that thermokarst lakes have a substantial impact on the permafrost configuration beneath and around it. Luo et al. (2012) analyzed the data of boreholes located in different locations and found that the ground temperature of the lake shore increased faster than that of the natural field. Although in-situ monitoring is one of the most direct methods for investigating permafrost, the high altitude and harsh environment on the TP limit the application of this method. Consequently, remote sensing techniques have become increasingly popular for investigating the permafrost state with the accumulation of satellite data, and their accuracy has continuously improved (Zhao et al. 2017, 2011; Zhou et al. 2022; Zhou et al. 2022). Interferometric synthetic aperture radar (InSAR) is an effective method for monitoring ground deformation in cold regions (Li, Xu, and Li 2022; Li et al. 2004; Rykhus and Lu 2008; Wang et al. 2021). In recent years, with the development of InSAR techniques, multitemporal InSAR (MT-InSAR) techniques have played an important role in monitoring surface deformation in permafrost regions (Lu et al. 2020; Lu et al. 2023). Chou, Zhen, and Xinwu (2009) applied the persistent scatterer InSAR (PS-InSAR) technique to extract surface deformation information around the Beilu River and indicated that the InSAR results were reliable when compared with in-situ monitoring data. In 2020, the Stanford method for persistent scatterers InSAR (StaMPS-InSAR) was used in the Wudaoliang Basin to measure the seasonal deformation of permafrost (Lu et al. 2020). Using Envisat and Sentinel-1A images, Lu et al. (2020b) extracted the surface deformation rate before and after the expansion of a Salt Lake in Hoh Xil and concluded that the expansion of the lake accelerated the surrounding permafrost degradation.

In this study, the permafrost thermal regime and surface deformation around a rapidly expanding lake and drainage channel were analyzed in the hinterland of the TP. The outburst of Zonag Lake triggered a series of floods in an endorheic lake basin, resulting in the expansion of Salt Lake since 2011 (Ding et al. 2022; Liu et al. 2019; Yao et al. 2012). To avoid the potential overflow of the Salt Lake, a drainage channel was built east of the lake (Wang et al. 2022a). By setting ground temperature sensors at different positions and using the small baseline subset InSAR (SBAS-InSAR) technique, the thermal regime and surface deformation was integrated to analyze the permafrost changes around the Salt Lake and the channel.

2. Study area, data and methods

2.1. Study area

The study area is located in the hinterland of the TP. Several lakes are distributed near the study area, with the Chumar River in the south and the Kunlun Mountains in the north (Figure 1a). Influenced by climate and Quaternary glacial conditions, the soil in the study area is stony and sandy, and small lakes and ponds are widely distributed. Before 2011, the Salt Lake was mainly recharged by seasonal rivers, and the area of the lake remained relatively stable (Figure 1b). After the outburst of upstream Zonag Lake in 2011, a cascade of water overflowed into Kusai Lake and Haidingnor Lake and eventually reached Salt Lake, resulting in the continuous and rapid expansion of the lake area in the following years (Figure 1c). In 2019, the construction of the drainage channel established a hydraulic connection between Salt Lake and the downstream Qingshui River, effectively integrating the lake into the Qingshui basin (Shanlong Lu et al. 2021) (Figure 1d).

Figure 1. The overview of study area as observed by Landsat images; (a) the topography around Salt Lake (2010) and the coverage of the radar satellite data; (b) Salt Lake status before the outburst (2010); (c) Salt Lake status after the outburst (2018); (d) Salt Lake status and channel (2021).

2.2. Data

Sentinel-1A images were used to extract surface deformation surrounding Salt Lake and the channel. Launched in 2014, the Sentinel-1 series is composed of Sentinel-1A and Sentinel-1B satellites. Utilizing C-band (4–8 GHz) synthetic aperture radar and a revisit period of six days, satellites are capable of delivering high-resolution images with exceptional spatial details. Sentinel-1 satellites generate multi-polarization products (HH, VV, HH+HV, VV+HV) for a variety of applications, including sea ice mapping, geophysical changes, earthquakes, and landslides (Choe et al. 2021; Nagler et al. 2015; Shimizu, Ota, and Mizoue 2019). In this study, level 1 single look complex (SLC) images were downloaded from the Alaska Satellite Facility (ASF). Precise orbit data corresponding to the SLC images were downloaded from the European Space Agency (ESA) data center to correct systematic errors caused by orbital errors. In addition, Shuttle Radar Topographic Mission (SRTM) files with a resolution of 30 m were downloaded from the USGS website and used to remove topographic phases and unwrap the differential interferograms. To present an overview of the study area, four Landsat images were merged and downloaded using the GEE platform with raw scenes provided by the USGS.

2.3. Methods

In this study, the SBAS-InSAR technique and in-situ monitoring were used to measure the deformation and obtain ground temperature of the study area, respectively. The SBAS-InSAR technique extracts surface deformation, including the deformation velocity, cumulative deformation, and time series of deformation. In-situ monitoring provided the properties of the permafrost state, including permafrost temperature, active layer thickness, and time series of ground temperature. Based on the data obtained by the SBAS-InSAR technique and in-situ monitoring, the state of surface deformation and permafrost and the association between the permafrost state and surface deformation can be presented spatially and temporally. The analytical procedure is illustrated in Figure 2.

Figure 2. Flowcharts for the permafrost state, surface deformation, and their association.

2.3.1. SBAS-InSAR technique

The InSAR technique was initially used to obtain surface elevation information. The D-InSAR technique, which was developed from the InSAR Technique, can measure surface deformation using phase information (Massonnet et al. 1993). Furthermore, the development of the MT-InSAR technique, including PS-InSAR and SBAS-InSAR, has improved measurement accuracy and time series. The PS-InSAR technique extracts deformation information by analyzing the phase changes of permanent scatterers on the ground, which is suitable for regions with dense buildings and many permanent scatterers, such as cities. Compared with PS-InSAR, which relies only on a single SAR image as the master product, SBAS-InSAR significantly reduces the need for a multitude of images. This technique enables an arbitrary combination of SAR images that satisfies the small baseline requirement and is more suitable for the analysis of spatially dispersed targets. Thus, since it was first proposed in 2002, SBAS-InSAR has performed better in regions with fewer permanent scatters, particularly permafrost regions (Wang et al. 2021). The main SBAS-InSAR procedure used in this study is summarized below.

1. Data read-in and reduction: DEM data, 45 Sentinel-1A images, and matching precise orbit data were imported into the software and converted into readable data. The settings for the mapping resolution, slant range, and azimuth were 20 m, 5, and 1, respectively. Using the vector data of the study area, the converted SAR images were cropped to reduce the processing time for subsequent steps.

2. Baseline estimation: Theoretically, for N-images of SAR, a maximum of (N*(N-1))/2 differential interferograms s would be generated. However, considering that interferogram formation is mainly affected by the thresholds of spatiotemporal baselines (Chen et al. 2013), 250 differential interferograms were generated with a spatial baseline of less than 2% of the critical baseline and a temporal baseline of less than 90 days (Figure 3).

   Figure 3. Time-position plot in the procedure of SBAS-InSAR.

   

3. Interferogram Process: In step, interferogram generation and flattening, adaptive filtering, coherence generation, and phase unwrapping were performed. Using the Goldstein adaptive filter and minimum cost flow (MCF) methods, a series of unwrapped phase diagrams was generated by 0.35 of unwrapping coherence thresholds.

4. Refinement and re-flattening: By properly selecting the Ground Control Points (GCPs) (no residual terrain stripes, no deformation stripes, no phase leap, and within half a pixel of error), the unwrapped phase was refined, and the residual phase was removed.

5. Inversion of SBAS: This process can be realized through two main operations. The first operation estimated the deformation rate and residual terrain and performs a secondary unwrapping to optimize the input interferograms. The second operation calculated the displacement of the time series and performed atmospheric filtering to remove the atmospheric phase based on the deformation rate obtained in the first operation. The coherence threshold was set to 0.35 for both operations.

6. Geocoding and raster-to-vector: To display the results properly, the images were geocoded, and the surface deformation was projected in the proper direction. The raster images were converted into vector data for subsequent processing.

2.3.2. In-situ monitoring

From 2019 to 2020, several ground temperature monitoring sections were strategically deployed along the shorelines of Salt Lake and the channel (Figure 4a). Each monitoring section contained five to seven boreholes. The depth of every borehole was 20 m and thermistor temperature sensors (TTSs) were placed at 0.5 m intervals from 0 to 10 m depth and at 1 m intervals from 10 to 20 m depth (Figure 4b). The temperature sensors used in this study have a working range of − 40–40°C, with a temperature resolution of 0.01–0.005°C in negative temperature conditions and of 0.01–0.03°C in positive temperature conditions. The accuracy of temperature measurement is better than 0.05°C. Temperature data were automatically collected using data loggers at intervals of 6 h.

Figure 4. In-situ ground temperature monitoring (a) arrangement of monitoring sections coded from 1# to 7# and the elevation (b) placement of temperature sensors at different depths.

3. Results

3.1. Changes in permafrost temperatures

3.1.1. The permafrost temperature around Salt Lake

As a tail lake in the basin, the area of the Salt Lake has expanded rapidly since the outburst of Zonag Lake (Ding et al. 2022). Following the outburst, the original shoreline of Salt Lake was gradually submerged by rising lake water, resulting in a significant alteration in the thermal conditions of the permafrost beneath and around the lake. To quantify the detailed configuration and spatial and temporal variations of the permafrost, seven boreholes, designated S1 to S7, were drilled from the lake to the natural field in Section 1#. Ground temperatures were collected at a depth of 20 m (Figure 5a).

Figure 5. The ground temperature monitoring around Salt Lake: a) the distribution of seven boreholes; the ground temperature of b) Salt Lake; c) the shoreline; and d) the nature field.

The ground temperature data collected from borehole S2 showed that the depth of the ALT in Salt Lake was 4 m and the permafrost temperature (PT, at depth of 10 to 15 m) was approximately − 0.5°C (Figure 5b), and reached its lowest point at a depth of 15 m. Situated on the shoreline of Salt Lake, the data collected from S4 indicated that the ALT was 3 m, which is 0.5 m deeper than that of S2, yet the permafrost temperature was same as that of S2 (Figure 5c). Drilled in the natural field, the ground temperature in S7 revealed that the ALT of the natural field was approximately 2.4 m, which was thinner than those in S2 and S4. The permafrost temperature recorded from S7 was also the lowest among the boreholes, reaching − 0.7°C (Figure 5d). Comparing the data from 2020 to 2022, the increase in ground temperature in S2 revealed a thickening tendency of the ALT beneath the Salt Lake, while the temperatures in S4 and S7 were relatively stable.

3.1.2. The ground temperature of the channel

As an excavation engineering project, the thermal condition of the permafrost beneath the drainage channel was significantly impacted by the movement of the surface soil and the direct flow of water from Salt Lake over the exposed soil. Four sections were constructed in the thermokarst and normal regions of the channel to investigate the permafrost state around the channel. Various engineering parts were monitored for ground temperature data, and thermal conditions were recorded.

Five boreholes were drilled from the lake to the natural field, and section 4# was located on the south bank of a thermokarst lake (Figure 6a). The data collected from borehole Fr1, which was situated in the thermokarst lake, indicated an ALT of approximately 2.8 m and a PT of − 0.4°C (Figure 6b). Similarly, Fr2 in the shoreline of the lake presented that the ALT was about 3.4 m and the PT was − 0.4°C (Figure 6c). The ground temperature data of Fr5 revealed that the ALT in the natural field was approximately 5.8 m, which was thicker than those of Fr1 and Fr2. The PT was approximately − 0.5°C at 20 m and − 1.0°C at 10 m, lower than other boreholes in this section (Figure 6d).

Figure 6. The ground temperature around the left bank of thermokarst lake: a) the overview of section 4#; the ground temperature of b) thermokarst lake; c) the shoreline; d) the nature field.

Section 5# is located on the right bank of a thermokarst lake. Five boreholes, numbered Se1 to Se5, were situated in lakes, shorelines, original shorelines, and natural fields (Figure 7a). Borehole Se1 in the lake showed the temperature of the soil down the lake were higher than 0.6°C, indicating that the thermokarst were formed sufficiently long and the underlying permafrost has degraded completely. However, the ground temperature data for Se1 also indicated that the soil in the lake cooled over time. On 5 October 2020, the temperature at 15 m depth was 3.36°C. However, in 5 February 2022, the temperature at the same depth dropped to 2.16°C, a reduction of 1.2°C (Figure 7b). Based on the data collected from borehole Se2 (Figure 7c), the ALT was 7.0 m and the PT was − 0.3°C in the shoreline of the thermokarst lake, much lower than the temperature of Se1. In the natural field, the borehole Se5 presented the thinnest ALT in this section with the depth of 3.5 m and the PT of − 0.3°C (Figure 7d).

Figure 7. The ground temperature around the right bank of thermokarst lake: a) the overview of section 5#; the ground temperature of b) thermokarst lake; c) the shoreline; d) the nature field.

Section 6# is situated on the right bank of the drainage channel, with five boreholes located from the central channel to the disposal platform (Figure 8a). In the slope foot of the channel (Th2), the ALT was 2.5 m and the PT was − 0.3°C (Figure 8b). In the natural field, data from Th4 indicate that the ALT in this area is thicker than that of Th2, reaching a depth of 3.8 m (Figure 8c). Borehole Th5 in the disposal platform had the thickest ALT at a depth of 4.2 m. The PT at the disposal platform was the same as the other boreholes, still keep − 0.3°C (Figure 8d). When comparing the temperature data from different boreholes, it was observed that PT remained constant across all boreholes. However, ALT varied significantly because the height of the boreholes increased from Th1 to Th5.

Figure 8. The ground temperature around the left bank of the channel: a) the overview of section 6#; the ground temperature of b) slope foot; c) the nature field; d) disposal platform.

Situated on the right bank of the drainage, section 7# contained five boreholes arranged from the central channel to the disposal platform (Figure 9a, where DP means disposal platform and TL means thermokarst lake). The data from the slope foot in the channel (Fo2) indicated that the ALT was 2.8 m and the PT was − 0.2°C (Figure 9b). As the height and distance increased, the ALT of Fo3 became thicker than Fo2, reaching a depth of 4.0 m and the PT reduced by 0.2°C, reaching − 0.4°C (Figure 9c). In the disposal platform, borehole Fo5 showed a similar ALT as that of Fo4, with a depth of 3.5 m and a PT of − 0.4°C (Figure 9d).

Figure 9. The ground temperature around the right bank of the channel: a) the overview of section 7#; the ground temperature of b) slope foot; c) the nature field; d). disposal platform.

3.2. The deformation results

3.2.1. The surface deformation around Salt Lake

Based on the SBAS-InSAR technique and Sentinel-1A satellite data, the cumulative deformation around Salt Lake was analyzed starting 8 September 2020. The results of cumulative uplift and subsidence are presented in Figure 10, where warm/cool colors represent surface deformation away from/toward the sensor line-of-sight (LOS). The results indicated that the surface deformation around the Salt Lake was dominated by subsidence and had a cumulative effect over time. Furthermore, subsidence increased with decreasing distance from the Salt Lake. Table 1 provides statistical information on the cumulative deformation since 8 September 2020. During the first period from Sep. 08 to 7 November 2020, the dominated deformation was uplifted with a mean deformation of 0.8 mm in the region. However, in the following period, it became apparent that the deformation around the Salt Lake was dominated by subsidence. After undergoing several freeze-thaw cycles, the cumulative subsidence deformation averaged more than 5 mm on 14 March 2022 (the detailed results around Salt Lake are presented in the supplementary data, Table SI1 and Figure SI1).

Figure 10. The cumulative deformation around Salt Lake, with reference to Sep. 08, 2020.

Table 1. The statistics of cumulative deformation (reference to 8 September 2020).

Date	Max (mm)	Min (mm)	Mean (mm)
Nov.07 2020	26	−36	0.8
Mar.07 2021	62	−37	−0.6
Jun.11 2021	99	−75	−2.2
Nov.02 2021	100	−86	−4.0
Mar.14 2022	118	−96	−5.6

*The symbol ‘-’ in the table indicated that the ground motions away from the satellite along line-of-sight (subsidence).

3.2.2. The deformation velocity of drainage channel

The deformation velocity along each side of the channel was obtained using the SBAS-InSAR technique, as shown in Figure 11. The warm color in the figure represents subsidence velocity, and the cool color refers to uplift velocity. The results indicated that subsidence was mainly concentrated in the upstream part of the channel, with the right bank subsiding more than the left bank. However, in the downstream part of the channel, the deformation was dominated by uplift, particularly on the left bank of the channel. The statistics provided in Figure 11 show that the maximum subsidence reached 39 mm/year and the maximum uplift reached 24 mm/year, whereas the average statistic indicated that the deformation trend around the channel was due to subsidence at a rate of 6.1 mm/year.

Figure 11. The deformation velocity around the channel.

3.3. The time series of deformation and ground/water temperature

Figure 12 shows the time series of the surface deformation and ground temperature in section 1# (Figure 12a) and 5# (Figure 12b). The water temperature (WT) in 1#/5# was measured 0.2 m above the bottom of the thermokarst lake/channel, while the ground temperature (GT) in 1#/5# was measured 0.5 m beneath the bottom of the thermokarst lake/channel. Based on the time-series data, the ground temperature at a depth of 0.5 m changed to a negative temperature in late November, which occurred later than the water temperature. During the warm seasons, both the water and ground temperatures changed to positive temperatures almost simultaneously. However, when the two variables were compared, the water temperature data exhibited more significant fluctuations.

Figure 12. The time series of deformation and the water/ground temperature: a) 1# section (Salt Lake); b) 5# section (the channel).

The deformation data revealed a trend opposite to that of the temperature data. Both section 1# and 5# experienced uplift, followed by violent subsidence, culminating in final subsidence. In section 1#, which is located near Salt Lake, subsidence reached − 17 mm in March 2022. However, the subsidence of section 5# of the channel was − 6.5 mm, much less than that of 1#. The temperature and deformation data indicate that the uplift of the surface reaches its maximum when the ground temperature reaches its minimum, whereas subsidence reaches its maximum at the same time when the ground temperature reaches its maximum. In addition, a comparison of the deformation/temperature data between section 1# and 5# demonstrated that although both sections contained similar water and ground temperatures, the deformation varied significantly in different locations.

4. Discussion

4.1. Permafrost degradation caused by climate change

Compared to the past 2000 years, climate warming over the past 50 years has accelerated (IPCC 2021). As the air temperature and precipitation continue to increase, the frequency of extreme events, including heavy precipitation and extreme heat events, has risen in recent years on TP (Lin Li et al. 2010; O’gorman 2015; Zhou and Qian 2021). In addition, as a region highly susceptible to climate warming, the processes of climate warming and wetting on the TP have intensified, leading to the degradation of the permafrost layer and the expansion of taliks (Cheng and Wu 2007; Wang et al. 2000; Zheng et al. 2020). Yin et al. (2021) indicated that the area of permafrost has been undergoing extensive shrinkage and predicted that more than half of the permafrost might disappear by 2100 under some scenarios. Based on the ground temperature from 152 climate stations, Ran et al. (2018) demonstrated that the total area of thermally degraded permafrost corresponded to 88% of the permafrost area in the 1960s and approximately 49.4 of very warm permafrost had degraded to seasonally frozen ground on the TP. Using a process-based model driven by satellite data, Zheng et al. (2020) estimated that the spatially averaged ALT was approximately 1.85 m and showed an overall increasing trend at an average rate of 3.17 cm/a.

The increasing impacts of climate change, including permafrost degradation, have intensified the risks to infrastructure in the TP. To mitigate hazards and plan engineering activities, Niu et al. (2014) divided a 10-km wide and 72-km long transect from Wudaoliang to the Fenghuo Mountain Pass along the QTEC into four classes based on slope failure susceptibility: 1) unlikely, 2) low, 3) moderate, and 4) high. By dividing the Qinghai-Tibet Railway (QTR) into 15 parts, Yin et al. (Yin et al. 2014) explained the impacts of excavation engineering activities on permafrost, demonstrating that excavation engineering triggered thaw slump, formation of surface water ponds and thermokarst lakes, and subsequent variation in the permafrost table and ground temperature. In a typical excavation engineering project, the construction of a drainage channel significantly influences the thermal regimes of the permafrost beneath and around the main infrastructure. Based on numerical simulations, Yu et al. (Yu et al. 2021) predicted the thermal regimes of a channel subgrade in both warm and cold seasons over 50 years, indicating that the permafrost beneath the channel mainly experienced downward degradation in the first 30 years and that the lateral thermal erosion of the flowing water would lead to permafrost degradation beneath the slopes. Additional monitoring sections and long-term observations are required to estimate the influence of engineering activities on permafrost degradation.

4.2. The chain impactions caused by the lake outburst

Deeply influenced by climate warming and wetting, lake outburst events have become more frequent on the TP and the number of potentially hazardous glacial lakes has increased since the 1930s (Liu, Cheng, and Su 2014; Richardson and Reynolds 2000). Using remote-sensing images from the 1970s to 2009, Wang et al. (Weicai Wang, Yao, and Yang 2011) indicated that the area of glacial lakes increased by 18.6%, while glaciers shrunk by 12.7% in the Boshula Mountain range, southeast Tibet. These changes increased the probability of glacial lake outbursts. In contrast to the outbursts of glacial lakes on the TP, the outburst of Zonag Lake in 2011 occurred in a continuous permafrost region. This extreme event triggered a series of outbursts downstream of Zonag Lake, including Kusai Lake and Haidingnor Lake, which continuously expanded the area of Salt Lake and changed the surface water distribution in the basin (Figure 13) (Ding et al. 2022; Liu et al. 2019). From 2000 to 2020, the area of Zonag Lake has decreased by approximately 43%, whereas that of Salt Lake has increased by approximately 400% (Lu et al. 2021). Additionally, the inflow of the Salt Lake increased from 0.246 km³/year in 1999 and 2010 to 0.758 km³/year during 2012 and 2019 after the outburst (Wang et al. 2022b).

Figure 13. The redistribution of the surface water caused by outburst: a) before the outburst in 2009; b) after the outburst in 2019.

Surface water changes in the basin after an outburst have impacted the permafrost environment around lakes in various ways (Xie et al. 2020; Zhang et al. 2022; Zhang et al. 2017). The rapid shrinkage of the lake area after the outburst directly exposed the original lake bottom to air, leading to a decrease in the ground temperature around Zonag Lake and resulting in the development of permafrost. In contrast, as the natural field was covered by the expansion of Salt Lake, the ground temperature in this area increased, indicating that permafrost degraded around Salt Lake after the outburst (Liu et al. 2019). The expanding area and rising water levels of Salt Lake put the downstream infrastructure at risk of flooding. Thus, a channel was built to drain water from Salt Lake into the Qingshui River (Ding et al. 2022; Lu et al. 2021). During construction of the excavation engineering project, the permafrost beneath and around the channel was influenced by water overflow. Additionally, the channel created the Zonag Lake, the headwater lake of the basin, which became the northern source of the Yangtze River. This influences various aspects of the headwater area of the Yangtze River, including the thermal regimes of the underlying permafrost, the water ionic composition, and infrastructure safety.

4.3. Permafrost environment and deformation around excavation infrastructure

The majority of infrastructure damage on the TP is caused by surface deformation, which has periodic variation due to freeze-thaw cycles. Because of the harsh environment of the TP, remote sensing techniques, in particular the InSAR technique, have been widely used to measure deformation along the infrastructure. Using the Envisat SAR dataset, Chang et al (Chang and Hanssen 2015). observed seasonal deformation over a range of 15 mm during 2007–2009 along the QTR track using the MT-InSAR technique. Based on the MTS-InSAR technique and 585 scenes of Sentinel-1 from 2017 to 2020, Wang et al. (2021b) proposed a new mapping method to classify the permafrost, seasonally frozen ground and degraded permafrost along the QTR. By monitoring thermokarst lakes using GEE and SBAS-InSAR techniques, Qin et al. (2023) concluded that the impact of the QTR on the acceleration of permafrost degradation may be overestimated. In this study, deformation around Salt Lake was measured using the SBAS-InSAR technique with Sentinel-1 data. The result indicated that the mean cumulative deformation around Salt Lake from 2020 to 2022 was approximately − 5.6 mm with the deformation velocity of − 4.6 mm/year. In particular, the deformation rate along the channel was presented as − 6.1 mm/year, indicating that subsidence around the channel was higher in Salt Lake.

Additionally, the ground temperature data demonstrates the spatial thermal regime of the permafrost around the channel (Figure 14). The ground temperature at both Salt Lake and the channel exhibited a decrease in permafrost with increasing distance from the lake (channel). At section 1#, the ground temperature at natural field, which is far from the lake, was 0.1°C cooler than the temperature at Salt Lake in the same depth at 5 m. As the depth increased, the difference between the natural field (far) and Salt Lake initially increased before decreasing and reaching 0.13°C at 7 m, 0.06°C at 15 m, and 0.04°C at 20 m. In comparison to section 1# at Salt Lake, the temperature in section 2# showed more intense changes with variations in the distance from the channel. The difference value between the natural field (far) and the channel was 0.02°C at 5 m, 0.6°C at 10 m, 0.2°C at 15 m, and 0.02°C at 20 m. As the first drainage project in the continuous permafrost region on the TP, the violent changes in ground temperature around the infrastructure revealed that the permafrost environment was influenced by excavation engineering within a certain range along the main infrastructure.

Figure 14. Permafrost temperature at a) Salt Lake, b) channel.

5. Conclusion

With continuous climate change and lake expansion, it is expected that drainage events of lakes on the TP will become more frequent. In this study, the surface deformation and permafrost thermal regime around a rapidly expanding lake and channel in the hinterland of the TP were investigated using in-situ ground temperature monitoring and SBAS-InSAR. The results of this study can provide insights for addressing such events in the future, and the main conclusions are summarized as follows:

1. The lake outburst in 2011 re-distributed the surface water in the basin, which resulted in the shrinkage of the headwater lake (Zonag Lake) and expansion of the tailwater lake (Salt Lake). Following the outburst, the area of Zonag Lake decreased by 43%, whereas that of Salt Lake increased by approximately 400%. This variation in the two lakes led to the development of a permafrost environment around the Zonag Lake and degradation around the Salt Lake. The ground temperature indicated that the active layer at Salt Lake was 1.1 m thicker than the active layer in the natural field, and the permafrost temperature decreased as the distance from the lake increased.

2. The ground temperature data collected around the Salt Lake and the drainage channel revealed a degradation trend in both areas. At Salt Lake, the temperature difference between the natural field and the lake was less than 0.2°C. However, the temperature difference reached 0.6°C between the natural field and the channel, highlighting the significant impact of the channel on the permafrost regime. In addition, the ground temperature in the channel exhibited more intense variation with depth, indicating that the surrounding permafrost environment was affected by the channel.

3. The deformation around Salt Lake was dominated by subsidence, with a mean subsidence of approximately 5.6 mm around Salt Lake from Sep. 08, 2020 to March 14, 2022. However, the subsidence rate decreased as the distance from the lake increased. The deformation around the channel was also dominate by the subsidence with a mean rate of 6.1 mm/year, which was 1.5 mm/year higher than the mean subsidence rate around Salt Lake. The deformation results demonstrate that excavation activities enhanced permafrost subsidence.

4. The time series of water temperature, ground temperature, and deformation exhibited periodic variations along with the freeze-thaw cycle. The ground temperature typically peaks between June and September, coinciding with surface deformation reaching its lowest point. Conversely, the peak of deformation usually occurs between December and May, whereas the temperature reaches its lowest point during the same period. The ground temperature time series for Salt Lake ranged from − 12. to 9.8°C, which was narrower than the water temperature ranging from − 14.1 to 13.9°C. The section 5# presented the ground temperature ranging from − 10.5 to 9.5°C and the water temperature ranging from − 16.8 to 14.8°C at the channel. In addition, the surface deformation at Salt Lake and the channel ranged from − 19 to 7 mm and from − 7 to 5 mm, respectively.

Conventionally, engineering activities on the TP have focused mainly on the filling and burial infrastructure. Due to the absence of large excavation engineering projects in continuous permafrost regions, research on the interaction between the permafrost environment and these activities remains limited. Although the thermal regime of the permafrost and the deformation around the channel were revealed by in situ monitoring and the InSAR technique in this study, there are numerous topics that require further research, including the variation of the thermokarst lake passing through the channel and the stability of the channel slope. To address these issues, comprehensive investigative methods, including in situ monitoring, remote sensing, and simulation, are necessary.

Acknowledgments

We are grateful to Alaska Satellite Facility (ASF), European Space Agency (ESA) data center and USGS for providing the datasets used in this study.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The raw data used in this study are publicly available from the sources provided in following links. The processed data generated from this study are available upon request to the corresponding author.

• Sentinel-1A data: https://search.asf.alaska.edu/

• Precious orbit data: https://scihub.copernicus.eu/

• Landsat data and SRTM data: https://earthexplorer.usgs.gov/

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15481603.2023.2266661

Additional information

Funding

The work was funded by the Second Tibetan Plateau Scientific Expedition and Research (STEP) program [2019QZKK0905]; the Science and Technology Project of Gansu Province [22ZD6FA004]; the Strategic Priority Research Program of the Chinese Academy of Sciences [XDA19070504].