Glacier Surface Velocity Variations in the West Kunlun Mts. with Sentinel-1A Image Feature-Tracking (2014–2023)

Abstract

Glacier velocity is a crucial parameter in understanding glacier dynamics and mass balance, especially in response to climate change. Despite numerous studies on glaciers in the West Kunlun Mts., there is still insufficient knowledge about the details of inter- and intra-annual velocity changes under global warming. This study analyzed the glacier velocity changes in the West Kunlun Mts. using Sentinel-1A satellite data. Our results revealed that: (1) The velocity of glaciers across the region shows an increasing trend from 2014 to 2023. (2) Five glaciers were found to have been surged during the study period, among which two of them were not reported before. (3) The surges in the study region were potentially controlled through a combination of hydrological and thermal mechanisms. (4) The glacier N2, Duofeng Glacier, and b2 of Kunlun Glacier exhibit higher annual velocities (32.82 m a⁻¹) compared to surging glaciers in quiescent phases (13.22 m a⁻¹), and were speculated as advancing or fast-flowing glaciers.

1. Introduction

High Mountain Asia (HMA) contains the highest concentration of glaciers outside the poles, is the main source of Asia’s rivers and lakes, and is often called ‘Asia’s water tower’ [1]. As the climate continues to warm, most of the glaciers in the HMA are thinning and retreating, which affects the hydrological processes of downstream regions [2]. Prior research has demonstrated a persistent trend of thinning and deceleration in most HMA glaciers from 2000 to 2017 [3,4,5]. However, glaciers in the eastern Pamir, Karakoram, and West Kunlun Mts. (WKL) are near equilibrium or slightly gaining mass, with certain glaciers undergoing acceleration or terminal advancements, commonly called the ‘Karakoram anomaly’ [6,7]. Many glacier hazards have emerged in the HMA, including glacier surges, glacial lake outburst floods (GLOFs), and ice avalanches, most of which are closely related to rapid glacier movement [8]. As an important parameter of glacier dynamics, monitoring glacier surface velocity can not only capture signals of rapid glacier movement or surges but also provides information about variations in strain rates and mass balance [9].

In situ observation has always been the most accurate method for glacier flow velocity monitoring. However, most of the glaciers in the HMA have been developed in areas with harsh environments, which makes it difficult for researchers to carry out the measurements [10,11]. Large-scale and repeated monitoring of glacier flow velocity has become possible with the rapid development of remote sensing technology. Optical image cross-correlation is a quick and effective solution to quantify glacier flow velocity, serves as the major method for glacier velocity estimation, and is reinforced by the burst of global optical image archives [12]. Nevertheless, optical remote sensing for glacier flow velocity assessment commonly faces substantial constraints due to protracted winters and persistent cloud cover over glacierized regions. Synthetic Aperture Radar (SAR) has the unique ability to operate independently from the availability of sunlight and is rarely affected by cloud cover, largely overcoming the shortcomings of optical satellites. There are two fundamental methods for estimating glacier flow velocity using SAR images: D-InSAR and offset tracking. D-InSAR often has limited applicability when measuring fast glacial movements or long repetitive cycles, which lead to decorrelation [13,14]. By contrast, offset tracking of the intensity of SAR image pairs to estimate the displacements can partially overcome the shortage of image interference. Hence, this technique is widely used for glacier velocity estimation [15].

The successive launches of the Sentinel-1A and Sentinel-1B SAR satellites with 6–12 repeat cycles provided an abundant data source for global glaciers surface velocity assessments, making it possible to monitor short-term variations in glacier flow velocity [16,17]. Milczarek et al. [18] used Sentinel-1 to determine the annual velocity changes of tidewater glaciers in the Hornsund Fjord, and analyze the seasonal and transient variations. Yang et al. [14] used the time series of Sentinel-1 data to determine the 69 surface velocity fields on the Neche Peninsula, Alaska, and analyze the differences in flow velocity variations among tidewater, lake, and land-terminating glaciers. Yang et al. [14] also investigated the effects of glacial lake outburst flooding (GLOF) and other environmental factors on ice flow. Furthermore, Sentinel-1 offers a wide range of applications for monitoring glaciers in the HMA region. Its high revisit period and weather-independent characteristics make it particularly useful for observing rapid velocity variations of surge-type glaciers during the active phase [10,19]. In addition, its Interferometric Wide mode can monitor glacier velocity variations on a regional scale.

Previous studies on the glacier dynamics of the WKL mainly focused on glacier mass balance, area changes, and their responses to climate change compared with other regions in the HMA, and also on updating the inventory of surging glaciers. However, studies on the inter- and intra-annual changes in glacier velocity in this region remained limited. Yasuda et al. [20] obtained glacier velocity information from 2003 to 2011 using PALSAR and ASAR data to investigate the seasonal variations and surging events in glacier flow velocity. They successfully identified four surge-type glaciers and investigated the possible mechanisms. However, determining the seasonal variations for certain glaciers is still challenging due to data limitations. In a recent study, Guan et al. [21] obtained glacier surface flow rates in the WKL using 86 Sentinel-1A pairs (2017–2019) combined with 78 Landsat images (1972–2020), and published an elevation dataset to update the inventory of surge-type glaciers and investigate their influences on glacier change. Therefore, it is essential to comprehensively assess the detailed variations in glacier dynamics to improve our understanding of glacier changes over the WKL.

This study investigated the glacier surface velocity changes of the West Kunlun Mts. between October 2014 and September 2023 via Sentinel-1A images, detected ongoing surge events, and analyzed the possible surge mechanisms.

2. Study Area

The West Kunlun Mts. (WKL) is one of the most glacierized regions in China, situated on the northwestern edge of the Tibetan Plateau (Figure 1). The main ridge of the WKL is more than 6000 m a.s.l., with Kunlun Peak (Kunlun Goddess Mt, ~7167 m a.s.l.) as the highest peak [22]. Hence, numerous large glaciers developed on the WKL, with less than 3% of the area covered by debris [23]. Morphologically, WKL glaciers include valley glaciers, cirque glaciers, and ice caps. Valley glaciers in northern and southern slopes typically consist of multiple tributaries, whereas ice caps are mainly located on the lower-elevation southern slopes [20]. The terrain on the south slope is relatively gentle (mean slope < 11°), and no glaciers develop below 5200 m. In contrast, some glaciers on the steep northern slope are below 5000 m [23].

The WKL is mainly affected by the mid-latitude westerlies, leading to a cold and semi-arid climate. Therefore, the WKL glaciers are considered to be a polar type (quasi-polar or extra-continental) [23]. According to the field survey, the annual average temperature and precipitation around the ELA (5930 m) were approximately −13.9 °C and 300 mm, respectively, with most precipitation occurring between May and September [24].

Figure 1. Study area and Sentinel-1A (paths 56 and 158) data coverage in this study. Background: ESRI standard map. The blue line is the glacier outline from RGI 6.0 [25]. Those glaciers with N or S indicate unnamed glaciers in the northern and southern slopes, respectively. Glaciers with centerlines are selected for a maximum length greater than 10 km. The lilac areas represent surging glaciers during the study period.

Figure 1. Study area and Sentinel-1A (paths 56 and 158) data coverage in this study. Background: ESRI standard map. The blue line is the glacier outline from RGI 6.0 [25]. Those glaciers with N or S indicate unnamed glaciers in the northern and southern slopes, respectively. Glaciers with centerlines are selected for a maximum length greater than 10 km. The lilac areas represent surging glaciers during the study period.

3. Data and Methods

3.1. Data

3.1.1. Sentinel-1A

Sentinel-1A, launched by the European Space Agency (ESA) in April 2014, carries a C-band sensor with a 12-day revisit period and acquires data in four imaging modes: Interferometric Wide (IW), Wave (WV), Strip Map (SM), and Extra Wide (EW). The IW mode has a swath length of 250 km, and a spatial resolution of 5 m in the ground range and 20 m in the azimuth [26]. Moreover, the revisit period was halved after the launch of Sentinel-1B in 2016, which formed a twin satellite constellation. With its global coverage, short revisit time, and detailed spatial information, Sentinel-1 has become an important data source for studying glacier motions [27]. Single-Look Complex (SLC) SAR images acquired in IW mode on two paths (Path 158/Frame 113 and Path 56/Frame 112, with 227 and 222 images, respectively) from October 2014 to September 2023 were used in this study to estimate glacier surface velocity in the WKL.

3.1.2. Copernicus DEM GLO-30-DGED

The foundational data of COP30 DEM originates from the TanDEM-X mission, which generates a highly accurate DSM based on SAR differential interferometry techniques that cover the global terrestrial surface, including Antarctica and the Arctic region. The COP30 DEM boasts remarkable accuracy, has an absolute vertical precision of less than 4 m (90% linear error), and a relative vertical accuracy of less than 2 m for slopes up to 20% [28]. Previous research has proved that the COP30 DEM is the most dependable open-source DEM [29].

3.2. Methods

3.2.1. Intensity Tracking

We used intensity tracking provided by GAMMA^® software to generate an offset field based on Sentinel-1A SAR image pairs through assuming parallel flow at the surface. The main steps include:

• Co-registration: The earlier image in image pairs was chosen as the master image, and another as the slave image. The precise orbit information was used to correct the orbit of the image to eliminate the initial offset error between the image pairs supplemented with COP DEM, using image matching techniques and spectral diversity methods. The final alignment accuracy of Sentinel-1 IW SLC data reaches 0.001 pixel.
• Offset tracking: The azimuthal phase slopes are resampled with an oversampling factor of two, the search window is set to 256 × 128, and the step size is set to 40 × 10. Offsets with a correlation coefficient lower than 0.05 were regarded as unreliable and removed. The outliers in the distance and azimuth directions were filtered out using a median filter with a 5 × 5 kernel [30].
• Geocoding: offset maps were geocoded using COP DEM and SAR image geometry, resampled to a 30 × 30 m grid, and projected to UTM coordinates based on the WGS 84 ellipsoid.

3.2.2. Uncertainty Assessment

Errors in the velocity maps may be induced through image registration, intensity tracking algorithms, ionospheric offset, and terrain-associated offsets [30]. Directly evaluating the results of the offset tracking is challenging, as no in situ measurements of glacier flow velocities are available in the WKL. Through considering the flat (slope < 10°) ice-free areas as stable [31], we computed the mean offsets of all image pairs in those regions and regarded them as the uncertainties of each image pair (Figure 2). The two paths’ mean uncertainties in glacier surface velocity are 0.011 m d⁻¹ and 0.012 m d⁻¹, respectively, and indicate the reliability of the velocity data in this study.

4. Results

4.1. Spatial Distribution of Glacier Velocity

Figure 3 illustrates the changes in average velocity in the WKL between October 2014 and December 2018, and from January 2019 to September 2023. The subplots depict the velocity data collected from flat ice-free areas. The annual average biases for these ice-free flat areas during the two periods are 3.00 ± 0.64 m and 4.26 ± 0.92 m, respectively.

The annual average glacier velocities of the WKL ranges from 0 to 440 m a⁻¹, with notable variations observed in glacier N1, Xikunlun East, Alakesayi, and Zhongfeng Glaciers. Before 2019, the velocities of Xikunlun West Glacier, glacier N1, and Alakesayi Glacier were relatively high, reaching 201 m a⁻¹, then significantly decreasing after 2019. On the contrary, the velocities of Xikunlun East Glacier and b4 of Zhongfeng Glacier were only ~47 m a⁻¹ before 2019. However, their annual average velocity increased significantly after 2019, with certain areas reaching up to 220 m a⁻¹. Previous research has indicated that glacier N1, Xikunlun East, and Alakesayi Glaciers have experienced surges before 2019 [21,32]. A comparison of the annual average velocity variations indicates that b4 of Zhongfeng Glacier and Xikunlun West Glacier may undergo new surges.

Other glaciers haven’t shown a clear difference in the annual average velocity in the two time periods. Notably, Duofeng Glacier and b2 of Kunlun Glacier consistently maintain high velocities throughout their tongues, with mean velocities of 30~50 m a⁻¹. But other larger glaciers, such as the Yulong, Chongce, and Xiyulong Glaciers, only exhibited lower velocities (<15 m a⁻¹) on their tongues.

4.2. Inter-Annual Velocity Changes

Inter-annual velocity maps during 2014–2023 were generated for the WKL along the glacier centerlines. Figure 4 presents the inter-annual velocity of glaciers with notable velocity fluctuations (see Section 4.1). The Alakesayi Glacier has consistently maintained a high velocity since 2014, gradually accelerating until 2016 with a peak velocity of around 602 m a⁻¹. The acceleration movements caused the upstream high-velocity zone to shift downstream, substantially increasing downstream velocity. Subsequently, the velocity of the Alakesayi Glacier decreased, reaching ~72 m a⁻¹ by 2018 and returning to ~30 m a⁻¹ by 2020 (Figure 4a).

The b4 (branch4) of Zhongfeng Glacier showed significant variations in inter-annual velocity, especially in the central region. The b4 entered a fast-moving phase in 2020, reaching its peak velocity in 2022 (Figure 4b–d), which caused the subsequent accelerations of b2 and b3, and then decelerated in 2023. Inter-annual velocity changes of Xikunlun East Glacier and glacier N1 have a similar temporal pattern, which both show gradual velocity declines all through the studied period (Figure 4e,f). The two branches of Xiunlun West Glacier exhibit comparable patterns of acceleration and deceleration. However, the velocity changes of b2 happen on the entire glacier, while for b1 this only occurs on its terminus (Figure 4g,h).

The inter-annual velocity of the other glaciers appears to be relatively stable, as shown in Figure 5. For example, the average velocity of the tongues of b1 of Kunlun, Chongce, and Xiyulong Glaciers was ~13.22 m a⁻¹ during the studied period (Figure 5a–c). In comparison to these glaciers, glacier N2, Duofeng Glacier, and b2 of Kunlun Glacier exhibit higher inter-annual velocities with a multi-year average of ~32.82 m a⁻¹, and glacier N2 showing a gradual acceleration since 2022 (Figure 5d–f).

4.3. Intra-Annual and Seasonal Changes of Glacier Velocity

Figure 6 and Figure 7 illustrate the intra-annual and seasonal changes along the glacier centerlines. The Alakesayi Glacier experienced high velocities as early as October 2014, reaching a maximum of ~1.5 m d⁻¹, and the region of high velocities shifted downstream over time, reaching its peak in October 2015 (~3.8 m d⁻¹; Figure 6a). Following multiple years of gradual acceleration, b2 of the Xikunlun West Glacier underwent a significant increase starting in the summer of 2020, reaching its peak in the winter of the same year. This was followed by a substantial downturn in flow velocity, with a minor resurgence in the summer of 2022 (Figure 6b). A gradual acceleration during summer 2021 can be observed on b4 of Zhongfeng Glacier, becoming apparent in February 2022, and reaching its maximum velocity in August 2022 at ~8.46 m d⁻¹. The higher velocity remained for several months, and then gradually decreased, with the most significant drop occurring in the surge front, while the higher velocity of the middle to upper regions was maintained until 2023 (Figure 6c). Both glacier N1 and Xikunlun East Glacier showed a similar deceleration trend from October 2014 (Figure 6d,e), of which glacier N1’s deceleration ended earlier (around the end of 2018; Figure 4f). Glacier N6 exhibits a different velocity change pattern than the other glaciers. The daily average velocity of around 1.5–3.5 km from the terminal varied greatly, and was consistently higher during winters (up to 1.9 m d⁻¹) and lower in the summers (Figure 6f).

In contrast to the previously mentioned glaciers, other glaciers demonstrate negligible intra-annual velocity variations (Figure 7). These glaciers can be classified into two categories based on their seasonal variations. The first group shows a predominant acceleration during the summer, as evidenced by glaciers like Chongce and Duofeng Glaciers and b1 of Zhongfeng Glacier (Figure 7a–c). The second group, such as Xiyulong Glacier, b1 of Kunlun Glacier, and b1 of Gongxing Glacier, displayed fewer seasonal variations (Figure 7d–f).

5. Discussion

5.1. Regional Glacier Movement Characteristics

Figure 8 reveals the overall trend of inter-annual glacier velocities in the WKL from 2014 to 2023. Glaciers with rapid annual average velocity changes shown in Figure 3 are not included in this statistic, as these glaciers may have surged during the study period. Generally, the glacier velocity in the WKL has accelerated from 2014 to 2023, consistent with Dehecq et al. [3]. The increasing rate of glacier velocity in this study is faster than previous studies on the WKL, which may be related to the larger rate of mass gain [33]. Li et al. [34] have estimated the changes of glacier thickness and mass balance in the WKL using ICESat-2, TanDEM-X 90m DEM, and SRTM DEM. They found that the glaciers have generally been in positive balance over the past 20 years. However, the mass accumulation rate during 2013–2019 (0.228 ± 0.055 m w.e. a⁻¹) was higher than that in 2000–2013 (0.173 ± 0.014 m w.e. a⁻¹). The same trend of glacier velocity changes has also been observed in the Karakoram. Dehecq et al. [3] found that the glacier speed in the Karakoram region has slightly accelerated from 2000 to 2017. In addition, Huang et al. [35] found the same trend in Eastern Pamir, where the glacier velocity has slightly increased since 2000 but suddenly increased rapidly after 2013. Furthermore, we compared the inter-annual velocity of glaciers in the WKL and Eastern Pamir and found that glacier velocity in the WKL is faster than in Eastern Pamir. As a comparison, glaciers in other regions beside Karakoram, WKL, and East Pamir were experiencing various degrees of deceleration, with the glaciers in the Nyainqêntanglha (−37.2 ± 1.1% decade⁻¹) and Spiti Lahaul (−34.3 ± 4.5% decade⁻¹) experiencing the largest slowdown [3].

5.2. Surge Event during the Study Period

Previous studies show that there are many surge-type glaciers in the WKL including glacier N1, b1 of Zhongfeng Glacier, b1 of Kunlun Glacier, Xikunlun East, Chongce, Xiyulong, Alakesayi, and Yulong Glaciers [21,32]. Some of these surge-type glaciers had low velocities, which implies that these glaciers might be in their quiescent phase during the studied period. Other glaciers, however, showed significant changes in inter- and intra-annual velocity, suggesting that these glaciers may be in their active phase. Using multisource remote sensing data, Fu et al. [36] revealed the surging behavior of the Alakesayi Glacier starting in 2013. This study also found that the glacier had higher velocities as early as October 2014 (Figure 6a).

Previous studies have shown that the surge of glacier N1 and Xikunlun East Glacier began in 2008 [20,32]. Our results show that both glaciers have a long active phase of >10 years, and the termination of N1 glacier’s surge is earlier than Xikunlun East Glacier. The inter-annual velocity changes of glacier N1 and Xikunlun East Glacier show similar trends, with both maintaining a gradual decline over the study period. The average annual velocity of the Xikunlun East Glacier gradually decreased from 166 m a⁻¹ in 2014 to 29 m a⁻¹ until 2021, indicating that its surge terminated in 2021 (Figure 4e). The inter-annual velocity of N1 glacier was 70 m a⁻¹ in 2014 and slowed down to 40 m a⁻¹, with its surge terminating in 2018 (Figure 4f).

Based on the inter- and intra-annual velocity changes, the velocity changes of b2 of Xikunlun West Glacier and b4 of Zhongfeng Glacier illustrated that they surged during the studied period (Figure 6b,c), which was not reported before. The b2 of Xikunlun West Glacier began to slowly accelerate from 2015 to 2020, and started to accelerate significantly in September 2020, peaking in the winter of the same year. The higher velocity continued until March 2022, then decelerated significantly, with a slight rebound in the summer of 2022 that lasted only a few months. Meanwhile, the glacier terminus also significantly advanced during the active phase until it converged with Xikunlun East Glacier (Figure S1), which provided additional evidence on its surges. Although no advances were found on b4 of Zhongfeng Glacier, its velocity changes show a clear surging pattern, which started in 2021, reached peak velocity in August 2022, and terminated in 2023.

5.3. Possible Control Mechanisms

Two mechanisms regarding glacier surges have been proposed in early studies, i.e., hydrological and thermal [37,38]. Thermal control mechanisms are mainly linked to the temperature at the glacier bed, wherein the meltwater generated after the glacier bed reaches the melting point under a higher overhead pressure leads to bed sliding, and thus surges. This type of surge is mainly characterized by progressive accelerations and abrupt terminations, as well as long active and quiescent phases [38]. In comparison, hydrologically regulated surges typically commence during winter and terminate in summer, with relatively shorter active periods and faster surging velocities, which are controlled mainly by developments of the sub-glacial hydrology system [37,39]. Previous research on HMA surge glaciers have indicated that the control mechanism of surges is primarily determined by their particular topographic, hydrological, and thermal conditions, however, it’s challenging to articulate a single control mechanism for HMA surges [10,40,41].

This study shows that the peak glacier velocities during their surges in the WKL are mostly below 5 m d⁻¹ (except b4 of Zhongfeng Glacier). This is considerably lower than the peak velocities of the hydrologically controlled surges [42] but comparable to the peak velocities observed in the surges controlled by a thermal mechanism [43]. The extended phases of acceleration and deceleration, as well as longer recurrence intervals, suggest that the glacier surges in the WKL are most likely controlled by thermal factors. However, the surges display acceleration in winter and deceleration in summer, indicating possible hydrological controls, corroborating the role of surface meltwater in the surge process. Therefore, it is suggested that the controlling mechanism behind the surge in the WKL may be comparable to that of the Karakoram and the Pamirs [40,41], which means both hydrological and thermal factors may have contributed to the surge process.

Furthermore, the study reveals that several glaciers in the WKL have recently experienced surges. While some of these glaciers may have been in a deceleration phase during our study period, the frequency of glacier surges seems to increase in comparison to the 20th century [32]. The Karakoram also shows a similar trend with an increased frequency of glacier surges in 21st century [44]. The thickening of glacial accumulation areas in the HMA may be linked to higher annual precipitation [45]. Although the influences of climate change on glacier surges are still in debate, scientific evidence indicates that changes in glacier mass balance are undoubtedly climate-related [44,45]. In conclusion, the trigger mechanisms for glacier surges are complex. Further climate research, subglacial thermal conditions, hydrological systems, and basal conditions are needed to fully understand the surge mechanisms in the WKL.

5.4. Characteristics of the Inter-Annual Velocity Changes

In this study, the glaciers that were previously proved to be surge-type glaciers (b2 of Kunlun Glacier, Chongce, and Xiyulong Glaciers; Figure 5a–c) exhibit relatively low inter-annual velocities in their quiescent phases, suggesting that they are essentially frozen to their beds. However, certain glaciers previously identified as possible surge-type glaciers (b2 of Kunlun Glacier, glacier N1, and Duofeng Glacier; Figure 5d,e) consistently maintain higher inter-annual velocities. When considering the surface velocities data (2003–2011) provided by Yasuda et al. [20], it is evident that these glaciers have maintained high velocities over the past two decades, much more significantly than surface velocities of confirmed surge-type glaciers during the quiescent period. A comparable phenomenon was observed in the study by Lv et al. [46] which reported the changes of two adjacent glaciers (one surging and one advancing) in eastern Karakoram over 30 years. It was observed that the advancing glaciers maintained a perennial velocity of 30–60 m a⁻¹, while the surge-type glaciers were virtually stagnant after its surge. Therefore, it is suggested that b2 of Kunlun Glacier, glacier N1, and Duofeng Glacier are more likely to be advancing or even so-called fast-flowing glaciers rather than previously identified surge-type glaciers. It is essential to acknowledge that this assessment may be influenced by our relatively short observation period and the long recurrence periods of surge-type glaciers in the WKL. Nonetheless, it introduces a new perspective on the identification of surge-type glaciers.

The glacier velocity data obtained through Sentinel-1 SAR intensity tracking show numerous gaps in the accumulation regions of the WKL glaciers, caused by rapid fresh snow and their rapid freezing and melting, thereby altering surface backscattering characteristics. These phenomena lead to larger biases in the velocity data of the accumulation regions in this study. Therefore, we mainly focused on the dynamics of ablation zones rather than accumulation regions in our analyses. Notably, this phenomenon is less sensitive to optical images [47]. Consequently, we propose merging velocity data collected from optical images with those derived from SAR (Sentinel-1) images to fill the gaps in future studies. This will provide a comprehensive understanding of seasonal variations in glacier surface velocity.

6. Conclusions

In this study, we utilized Sentinel-1A imagery pairs to generate glacier surface velocities in the WKL from 2014 to 2023, based on the intensity tracking method. We conducted a detailed investigation into the spatio-temporal changes of the surface velocity of glaciers in this region. Our major conclusions include:

(1)

Glaciers in the WKL show an accelerating trend between 2014 and 2023, and the acceleration is greater than before, which may be induced by enhanced mass gain in this region.

(2)

Five glaciers experienced surges during the study period, including b2 of Xikunlun West Glacier, Alakesayi Glacier, glacier N1, Xikunlun East Glacier, and b4 of Zhongfeng Glacier, among which the surges of b2 of Xikunlun West Glacier and b4 of Zhongfeng Glacier were not reported before.

(3)

Based on the characteristics of the observed glacier surges, it is suggested that both hydrological and thermal mechanisms may have contributed to the surge in WKL.

(4)

The glacier N2, b2 of Kunlun Glacier, and Duofeng Glacier have relatively higher inter-annual velocities than surging glaciers during their quiescent phase. It is speculated that these glaciers are more likely advancing or fast-flowing glaciers.