Introduction

Ice-rock avalanches are a unique type of long-runout landslides that occur in high mountainous regions caused by ice avalanches or rock collapses, incorporating ice in the moving mass either during initiation or transportation process (Schneider et al. 2011; Deline et al. 2021). They are often characterized by large volumes, strong destructiveness, and high energy. Ice-rock avalanches are prevalent in high mountain regions worldwide, such as the Rocky Mountains in North America, the Andes in South America, the Alps in Europe, the Caucasus Mountains at the Asia-Europe border, and the Himalayas in Asia (Fig. 1a). These regions, characterized by steep terrain and glaciers, offer ideal conditions for the occurrence of ice-rock avalanches, making them highly susceptible to such disasters.

Fig. 1
figure 1

Global distribution of glaciers and typical ice-rock avalanche: a global glacier distribution; b deposition areas of the 2002 Kolka-Karmado ice-rock avalanche (photo: I. Galushkin on September 22, 2002); c boulder transported by the 1970 ice-rock avalanche at Huascaran, Peru, deposited near the Old Yungay cemetery (Klimeš et al. 2009); d the detachment area of the 2020 ice-rock avalanche in Chamoli, India, characterized by the 550-m rupture line (source: Twitter account @sgascoin); e damaged Langtang village, and the famous “Chess boulder” displaced by the destructive ice-rock avalanche (Zhuang et al. 2023)

Full size image

Ice-rock avalanches are often defined by their high speed and ability to travel long distances, ranging from several to tens of kilometers. These avalanches can cause severe damage to downstream areas (Fig. 1b–e). The 1962 and 1970 Huascarán large ice-rock avalanches, which originated as rock/ice falls from Nevado Huascarán, transformed into higher-volume high-velocity debris flows by incorporating snow from the glacier surface. Among them, the 1970 event caused by a strong earthquake is the most disastrous historical ice-rock avalanche event. The debris flow rapidly traveled 16 km at an average speed exceeding \(50\text{ m}/\text{s}\), destroying the town of Yungay, and resulting in an unfortunate loss of over 7000 lives (Plafker and Ericksen 1978; Körner 1983; Evans et al. 2009). In September 2002, the collapsed rock/ice from the northern face of Dzhimarai-Khokh in Russia’s Caucasus Mountains struck the Kolka Glacier, dislodging the glacier and releasing an approximately \(100\times {10}^{6} {\text{m}}^{3}\) mixture of ice, snow, and rock debris. This avalanche mass traveled 20 km and completely demolished villages along its path, causing about 120 fatalities (Haeberli et al. 2004; Huggel et al. 2005; Huggel 2009). On April 25, 2015, the Gorkha earthquake in central Nepal triggered a disastrous two-stage ice-rock avalanche in the Langtang valley with an approximate volume of \(14.38\times {10}^{6} {\text{m}}^{3}\), leading to the destruction of a village and over 350 fatalities (Gnyawali et al. 2020; Zhuang et al. 2023). On February 7, 2021, a massive ice-rock avalanche in the Chamoli district of India induced a catastrophic flood resulting in about 200 people dead or missing, and the destruction of two hydropower stations downstream (Shugar et al. 2021; Zhou et al. 2021). The combined influence of global climate change and glacier retreat has resulted in more frequent extreme rainfall events and accelerated snowmelt on high mountains; consequently, there is an increase in the scale and occurrence of ice-rock avalanches (Davies et al. 2001; Geertsema et al. 2006; Huggel et al. 2010; Coe et al. 2018; Smith et al. 2023; Allstadt et al. 2024), often with recurring and overlapping events from the same source areas (Geertsema and Bevington 2021).

The Qinghai-Tibet Plateau (QTP), the highest plateau in the world, has 73.11% of its area above 4000 m, making it the region with the largest glacier reserves outside the Arctic and Antarctic. In the past 20 years, the cold high mountain areas of the QTP have experienced accelerated ice loss and glacier retreat, with the rate of regional climate warming exceeding the global average (Mu et al. 2020). Furthermore, the QTP is renowned for its widespread high mountains, deep valleys, and modern glaciers, together with frequent extreme climate events that result in major ice-rock avalanche disasters characterized by distinct repeated occurrences. Notable examples include the ice-rock avalanche events in the Yigong Range in 1900 and 2000; Aru Lake Range on July 17 and September 21, 2019; Amney Machen Range in 2004, 2007, 2016, and 2019; and Sedongpu Basin in 1974, 2014, 2017, 2018, 2021, and 2024 (Yin and Xing 2012; Kääb et al. 2018; Zhuang et al. 2020; Fan et al. 2019; Hu et al. 2019; Paul 2019; Wang et al. 2020; Li et al. 2022; Zhang et al. 2023) (pink stars in Fig. 2b). Therefore, investigating the triggering mechanisms and the spatio-temporal evolution of ice-rock avalanches is crucial for the effective prevention and management of such catastrophes and their associated disaster chains in high mountain regions.

Fig. 2
figure 2

Distribution of glaciers and ice-rock avalanches in the QTP: a distribution of glaciers on the QTP (based on the second Chinese glacier inventory) and b distribution of ice-rock avalanches in the QTP

Full size image

This study focuses on the ice-rock avalanches in the Xiaoma Gully, Amney Machen Range, north-eastern QTP. Four ice-rock avalanches successively occurred in February 2004, October 2007, October 2016, and July 2019 in Xiaoma Gully. We utilized field investigations and high-resolution remote sensing data to verify the geological environment and disaster distribution in the Amney Machen Range. Coupled with glacier in situ observations and meteorological data from nearby stations, the changing trend of the geological and climatic environment was analyzed, revealing the triggering mechanisms of ice-rock avalanches. Based on offset-tracking, support vector machine (SVM) classification, and digital elevation model (DEM) differencing, the surface velocity and volume changes of Amney Machen glacier were monitored, and the spatial–temporal evolution characteristics were analyzed. The results provide some basic evidence for the triggering mechanisms and spatial–temporal evolution of recurring ice-rock avalanche chain disasters in high mountains, as well as some insights into hazard assessment and deformation monitoring in glacial areas.

Amney Machen ice-rock avalanches

Distribution of glaciers and ice-rock avalanches in the QTP

China has 48,571 glaciers, with an area of about \(5.18{\times 10}^{4} {\text{km}}^{2}\) and a volume of \(4.3\sim 4.7\times {10}^{3} {\text{km}}^{3}\) (Shi et al. 2009; Liu et al. 2012; Guo et al. 2015), that are mostly concentrated in the extensive cold mountain region centered on the QTP of western China (Fig. 2a). The glaciers are sources for major rivers such as the Yarlung Zangbo, Nujiang, and Yalong, and their meltwater provides an important recharge for surface runoff. Additionally, these rivers provide the conditions for the formation of chain disasters (e.g., dammed lakes and floods). The QTP is highly sensitive to climate change, with a temperature increase of about 1.1° in the last 50 years, which is twice the average global warming rate over the same period (Mu et al. 2020; Peng et al. 2020; Cui et al. 2022). In the QTP, regional warming has caused an overall glacier retreat. By 2000, the glacial area in the QTP decreased by about 20%, with the most significant glacier retreat and mass loss occurring in the Himalayas and southeast Tibet, and this trend is expected to grow gradually (CAS 2015).

Increasing regional temperature, frequent extreme precipitation, glacier recession, and upward shifting of the snow line have altered the hydrogeological conditions and disaster incubation environments in the QTP. Instability in high mountain environments has significantly increased, leading to frequent occurrences of catastrophic ice-rock avalanches. The mass movement processes tend to recur within a single progressively expanding source area, exhibiting the characteristics of a range of spatial and temporal dimensions (Allen et al. 2022). These notable ice-rock avalanches are not only massive in scale, with obvious cascading effects, but also exhibited a twofold dimension, continuing temporally and expanding spatially, with preceding disasters providing material and dynamic foundations for subsequent ones.

Amney Machen ice-rock avalanches

The Amney Machen Mountains, located in Maqin County, Golok Tibetan Autonomous Prefecture, Qinghai Province (north-eastern QTP), are 120 km long and 40 km wide, extending from northwest to southeast with coordinates E99°27′ and N34°49′ (Fig. 3a). They are composed of 13 peaks, with most of them ranging from 4500 to 5500 m in elevation. The highest peak is Maqengangri at an elevation of 6282 m, with a maximum height difference of about  from the deposition area at the Qinglong Gully. There are 85 modern glaciers in the region, covering an area of and serving as a significant water supply for the Yellow River. The source area of the repeated ice-rock avalanches is located on a steep slope at an altitude of 5800 m on the western side of the Maqengangri Glacier. The collapsed ice-rock masses (with 90% ice in volume) descended  and struck the Xiaoma Gully, fragmenting and entraining the rock debris along its path, forming an ice-rock mixed flows (Paul 2019; Kääb et al. 2021; Zhang et al. 2023). The induced mixed flows traveled through Xiaoma Gully and reached Qinglong Gully at an elevation of 4300 m, completely blocking Qinglong Gully, with the depositional area in a fan-shaped pattern

Fig. 3
figure 3

Geological feature and repeated ice-rock avalanches of the Amney Machen Mountains area: a geological feature and b repeated ice-rock avalanches

Full size image

The distribution range of the ice-rock avalanches at Amney Machen Range is shown in Fig. 3b (Paul 2019; Zhang et al. 2023). Field investigations and data analysis categorize the Amney Machen repeated ice-rock avalanche area into four distinct sub-areas: source area, transportation area, deposition area, and avalanche-prone area. Among them, the avalanche-prone area refers to the area surrounding the source area with unstable ice-rock masses, which may potentially become a new source area through the occurrence of another ice-rock avalanche. All of these avalanches originate from the southwest slope of the Maqengangri Glacier above the Xiaoma Gully. The profile has steep upper (40°) and gentle lower Sects. (15°), with an overall gradient of about 28° and a drop of 620 m. After the four ice-rock avalanche events in 2004, 2007, 2016, and 2019, the glacier distribution above the Xiaoma Gully shrank substantially, with many deep crevasses on the surface of the remaining glacier surface, indicating the possibility of another ice-rock avalanche.

In February 2004, a significant ice-rock avalanche occurred on Maqengangri’s main peak. The source material was primarily glacial ice, along with a small amount of bedrock from the ice-rock interface. Based on the observation of the source area with exposed bedrock and relevant literature, the volume proportion of ice and rock was 90% and 10%, respectively. The avalanche started at an elevation of 5800 m, with a substantial volume of glacier ice detaching from the source area and incorporating some rocks below it. After falling \(1150\text{ m}\) and colliding with the Xiaoma Gully, the avalanche material disintegrated and transformed into an ice-rock mixture, entraining ice-rock debris along the bottom of the gully along the way. This resulted in the formation of an erosion gully with a maximum depth of 100 m, as depicted in Fig. 4a. The ice-rock debris extended into Qinglong Gully in a fan-shaped form, covering an area of \(2.4 {\text{km}}^{2}\) with an average thickness of 10 m and a volume of \(21.5\times {10}^{6} {\text{m}}^{3}\) (Fig. 4b). The debris material consisted of a mixture of ice and rock debris, with ice accounting for about 70 to 80% of the entire volume. Soft carbonaceous shale and muddy slate make up the majority of the rock debris, with sizes less than 5 cm accounting for around 80% and those larger than 40 cm accounting for about 5%, with a maximum size of up to 3.5 m. The deposit blocked Qinglong Gully, forming a landslide dam 1530 m long, 200 m wide, and 30 m high, as well as a dammed lake of \(5.38\times {10}^{6} {\text{m}}^{3}\), which breached on July 4 with a peak discharge of 506 \({\text{m}}^{3}/\text{s}\) (recorded at 8:00 on 5 July 2004 by the Damitan hydroelectric power station in Xinghai County, and data from Zhou Bao of the Science and Technology Department of Qinghai Province) (Fig. 4c, d). This ice-rock avalanche was the largest among the four at Amney Machen Range, destroying about 3000 acres of pasture downstream, and preventing 175 herding households from accessing the summer pastures. The remaining three events were relatively small and similar and are summarized in Table 1.

Fig. 4
figure 4

Field photos of the 2004 ice-rock-snow avalanche in the Amney Machen Range (photos taken in April 2004): a source area and erosion area of the 2004 ice-rock avalanche; b debris deposition area and dammed lake at Qianlong Gully; c dammed lake formed by the blockage of the Qinglong Gully; d landslide dam and ice cracks in Qinglong Gully. Photos from Zhou Bao of the Science and Technology Department of Qinghai Province

Full size image
Table 1 Summary of the multiple ice-rock avalanches in Amney Machen Range (Paul 2019; Zhang et al. 2022; Zou et al. 2023)
Full size table

Method and data source

In addition to field survey data, meteorological data (Maqin station), and seismic data (USGS: https://earthquake.usgs.gov/earthquakes/search/), this paper also draws on high-resolution remote sensing imagery, which is now very mature and widely used, as supplementary research data to investigate the evolutionary trend of glaciers and the formation mechanism of glacial disasters. Different data sources and synthetic aperture radar (SAR) processing methods are used for interpreting remote sensing imagery (Table 2).

Table 2 Remote sensing image data used in this study
Full size table

The basic glacier data is from the second glacier inventory dataset of China by the National Cryosphere Desert Data Center (Liu et al. 2012; Guo et al. 2015). The basic glacier data was visualized using a Geographic Information System (GIS), summarizing the elevation and slope characteristics of the glaciers in the Amney Machen Mountains. Using the Sentinel-1 SLC images, the differential interferometry SAR (D-InSAR) method was used to analyze the glacier deformation rate in the source area during rainy and dry seasons. At the same time, the Small Baseline Subset InSAR (SBAS-InSAR) was used to conduct a time series analysis of deformation in the avalanche-prone area from 2021 to 2023, and hazard assessment of avalanche-prone mass was carried out. D-InSAR is a microwave remote sensing technique that can measure surface deformation phenomena with centimeter to millimeter precision and a large spatial coverage capacity (Gabriel and Goldstein 1988; Gabriel et al. 1989; Massonnet et al. 1993, 1995; Hu et al. 2014). By calculating the interferogram between two SAR images of the same survey area at different times, ground deformation measurements can be obtained. SBAS-InSAR forms differential interferometric pairs using a large amount of SAR images acquired within spatial and temporal baseline thresholds (Berardino et al. 2002; Lanari et al. 2004; Gong et al. 2016). It selects coherent target points and utilizes a linear phase change model for modeling and solving. Spatio-temporal filtering is applied to remove atmospheric delays while mitigating decorrelation effects and elevation/atmospheric errors in the D-InSAR processing to obtain the ground deformation time series.

To clarify the dynamic evolution characteristics of glaciers, six Sentinel-1 GRD images were selected, and the offset-tracking method suitable for glacier analysis was used to extract the distribution of glacier flow velocities in the Amney Machen Mountains between 2014 and 2023, and also for the 2016 and 2019 events. Under conditions of rapid and incoherent glacier flow, as well as large time intervals between acquisitions, D-InSAR is often limited by coherence loss and struggles to obtain effective information. In such cases, offset-tracking using SAR imagery serves as an alternative to D-InSAR for estimating glacier motion (Strozzi et al. 2002; Hu et al. 2014; Gomez et al. 2019).

Finally, to reveal the changes in glacier mass under the conditions of global warming and glacier retreat, Landsat 8 OLI data was used to quantify the changes in glacier area in the Amney Machen Mountains from 2014 to 2021 using the SVM classification method (Gunn 1998; Alifu et al. 2020; Cervantes et al. 2020). By manually delineating a small number of samples of glaciers, snow, and bedrock, and using SVM classification for sample training and automatic classification, the glacier area for each year was determined by combining the interpretation results with field investigations and visual inspection. At the same time, the DEM differencing was used to determine the changes in glacier thickness from 2007 to 2021 (Paul and Haeberli 2008; Gardelle et al. 2012; James et al. 2012). The 2007 glacier DEM in the Amney Machen Mountains was derived from ALOS PALSAR data, while the 2021 glacier DEM was extracted from ASTER stereo image data. By differencing the glacier DEMs from 2007 and 2021, the changes in glacier thickness in the Amney Machen Mountains were obtained. Regarding the detailed information about the D-InSAR, SBAS-InSAR, offset-tracking, SVM classification, and DEM differencing methods, the paper listed above (e.g., Gabriel and Goldstein 1988; Berardino et al. 2002; Strozzi et al. 2002; Hu et al. 2014) can be referred to, so it will not be elaborated further here.

Triggering mechanisms and spatio-temporal evolution

Climate change and meteorological conditions

Ice-rock avalanches involve water produced by ice-water phase transition and melting, so climate and meteorology have an important influence on their formation and occurrence. We analyzed the potential correlation between Amney Machen repeated ice-rock avalanche events, long-term climate change, and short-term meteorological conditions (Fig. 5), using temperature and precipitation data from 1961 to 2022 at Maqin station, the closest meteorological station to the avalanche source area. The mean average annual temperature and precipitation both show an increasing trend, with an increment of 0.30 °C/10a for the average annual temperature and 23.3 mm/10a for the annual precipitation from 1961 to 2022, which is consistent with the overall climate warming across the QTP (Fig. 5a).

Fig. 5
figure 5

Trends in temperature and precipitation in the Amney Machen Range: a mean annual temperature and precipitation changes from 1961 to 2022 and b temperature and precipitation in the Amney Machen Range from 2020 to 2022

Full size image

Climate warming causes glacier retreat and permafrost degradation, providing a favorable environment for ice-rock avalanche formation (Huggel et al. 2012; Hock et al. 2019; Yang et al. 2019; Deline et al. 2021). Steep glaciers experience long-term stability decline in response to stress redistribution following de-buttressing (the loss of basal support) by glacier retreat (Haeberli et al. 2017; Sæmundsson et al. 2020; Deline et al. 2021). Rainfall and ice meltwater act as lubricants to reduce friction at the glacier-bedrock interface, improving glacier mobility and further compromising its stability. Meanwhile, the glacial ice ablation produces near-surface ice crevasses parallel to the slope (due to stress release). Furthermore, the permafrost degradation caused by climate warming reduces the stability of glacier slopes and increase their sensitivity to landslides, and the control effect of permafrost degradation on ice-rock mass may be reflected in the following aspects (Noetzli and Gruber 2009; Huggel 2009; Gruber 2012; Krautblatter et al. 2013; Scherler et al. 2013; Coe et al. 2018; Deline et al. 2021): the melting of ice in the upper permafrost layer and the increase in active layer thickness lead to a decrease in the mechanical slope support for ice-rock mass, resulting in reduced stability; thermal anomalies in the surface rock in the permafrost zone due to climate warming, heat convection by melting of interstitial ice, and latent heat from refreezing lead to a decrease in rock strength as the rock-permafrost slopes warm; and the thawing of the upper permafrost layer, the disappearance of localized aquicludes, and the effects of meltwater and rainfall make the rock debris more prone to sliding or entrainment. Therefore, permafrost degradation may increase the instability of the ice-rock system, controlling the timing and size of ice-rock avalanches. The mean annual temperatures in the ice-rock avalanche years 2004, 2007, 2016, and 2019 were 0 °C, 0.7 °C, 0.9 °C, and 1.0 °C, and the annual precipitation was 545 mm, 555 mm, 663 mm, and 763 mm, respectively, all relatively high historically. In addition, since 2001, the mean annual temperature has been above 0 °C and continues to rise steadily, and the precipitation has also remained high and increased since 2014, with the mean precipitation being 704 mm. This indicates increasingly severe climate warming in the Amney Machen Mountains, as well as a very high probability of more ice-rock avalanches in the same area.

The East Asian monsoon influences the Amney Machen Mountains, resulting in concurrent rainy and hot seasons. For example, from 2020 to 2022, the rainy season is from May to October, with mean monthly temperatures above 0 °C, a maximum temperature of 22.1 °C, and monthly precipitations greater than 50 mm. Total precipitation in the rainy season is 753 mm, which is about 86% of the annual precipitation, and the rainfall and the meltwater converge rapidly. The dry season is from November to April of the following year, with mean monthly temperatures below 0 °C, a minimum temperature of − 26.7 °C, mean monthly precipitations less than 50 mm, and a total dry season precipitation of only 123 mm (Fig. 5b). The above results are based on the 3-year average values from 2020 to 2022 to exclude the influence of accidental weather conditions and better reflect the climatic characteristics of the Amney Machen Range.

During the rainy season, high temperatures and rainfall accelerate the melting of the surface glacier ice. This meltwater and rainwater infiltrate through ice crevasses formed by stress relief, reaching the glacier bed and contacting the underlying bedrock (Paul 2019; Kääb et al. 2021; Zhang et al. 2023). This influx of meltwater and rainfall input large amounts of liquid water into the ice-rock system, lubricating the ice-rock interface and softening the substrate material (Iverson et al. 1998; Evans et al. 2009; Gilbert et al. 2018). At the same time, infiltration of meltwater and rainfall increases pore water pressure, and seepage further enlarges fissures and provides an external driving force for glacier flow, which accelerates glacier movement (Eberhardt et al. 2004; Broughton 2018). Frost heave is significant in the dry season when the liquid water in the crevasses (rainfall and meltwater during high temperatures) re-freezes in the cold environment. The freezing force generated by the water–ice phase transition leads to the expansion of crevasses, resulting in reduced stability of the ice-rock mass (Blikra and Christiansen 2014; Deline et al. 2021). Additionally, there is a significant difference between the maximum and minimum monthly temperatures in the Amney Machen Range, with a mean monthly temperature difference of 27 °C from 2020 to 2022 and varying above and below the freezing point. This results in an intense freezing and thawing effect almost year-round, which is dominated by the thawing effect in the rainy season and the freezing effect in the winter dry season. The intense freezing and melting effects lead to crevasse expansion and strength degradation in the ice-rock mass, while also subjecting the glacier to an environment of significant temperature fluctuations. Under these circumstances, low-altitude glaciers would generate latent heat due to refreezing of meltwater penetrated through crevasses, as well as frictional heat from glacier flow and friction against bedrock (Huggel et al. 2005; Huggel 2009; Khamis et al. 2015; Benn et al. 2019; Deline et al. 2021). The latent heat and frictional heat cause more meltwater at the ice-rock interface, and the freeze–thaw action along with basal debris entrainment creates loose material, thereby forming a potential low-friction sliding surface that is further lubricated and softened by the effects of rainfall and meltwater. Freeze–thaw action, rainfall, and meltwater all reduce the stability of glaciers to varying degrees and increase the hazard of ice-rock avalanches.

Statistical data shows that since 2000, there have been a total of 25 typical ice-rock avalanche events in the QTP, with 19 occurrences (76%) happening during the rainy season. This suggests that rainfall, thawing damage, and meltwater in the rainy season are the dominant meteorological factors of ice-rock avalanches’ formation (Fig. 6a). This conclusion is also corroborated by the results of D-InSAR interpretation based on 2020 remote sensing images of Xiaoma Gully (Fig. 6b). The deformation values of the glacier in the ice-rock avalanches are higher in the rainy season than in the dry season, with the contrast being particularly pronounced in the avalanche-prone area. Furthermore, since 2000, there has been an increase in the number of major ice-rock avalanches, which aligns with the climate warming trend.

Fig. 6
figure 6

Major ice-rock avalanche events in QTP and D-InSAR deformation interpretation in rainy and dry seasons: a major ice-rock avalanche events in the Qinghai-Tibet Plateau since 2000 and b deformation interpretation in the avalanche-prone area during rainy and dry seasons in 2020

Full size image

The first ice-rock avalanche (February 2004) occurred during the cold dry season, while the latter three events (October 2007, October 2016, and July 2019) happened in different months of the warm rainy season. Together with the pre-failure temperature data from Maqin station (Fig. 7), the daily temperature difference is also very significant in the Amney Machen Range, reaching 13 °C before all four events. The large monthly and daily temperature ranges lead to frequent and intense freeze–thaw effects in the Amney Machen Range (Wang et al. 2022; Zhang et al. 2023), which is partly supported by the numerous dense and deep crevasses around the source area due to glacial melting and retreat.

Fig. 7
figure 7

Daily temperature and precipitation before the Amney Machen ice-rock avalanches

Full size image

According to the meteorological data obtained from the Maqin station (34.47°N, 100.24°E; altitude, 3723 m), located 82 km from Xiaoma Gully, as shown in Figs. 5 and 7, the freeze-thaw action prior to the four events was different. The maximum temperature before the 2004 event was below 0 °C; the temperature before the 2007 and 2016 events fluctuated around 0 °C, while the minimum temperature was above 0 °C before the 2019 event. Based on the temperature characteristics before the four events and considering the temperature gradients (0.57~0.6 ℃/100 m in the QTP) (Yao et al. 2000; Wu and Zhang 2008; Wu et al. 2010), it can be inferred that the temperatures in the four events may have contributed to frost heave and freeze-thaw cycles, respectively. Prolonged low-intensity rainfall preceded the three avalanche events during the rainy season, which may have lubricated sliding surfaces and reduced resistance, thereby potentially promoting landslides to some extent (Yin et al. 2017; Senthilkumar et al. 2018; Kääb et al. 2021). However, the pre-failure rainfall was discontinuous and too low in intensity, with the total precipitation in the week preceding the four events being , , , and  (Fig. 7). Given the large daily and monthly temperature differences while low-intensity rainfall (Figs. 5b and 7), temperature may play a more critical role than rainfall in the short-term period preceding ice-rock avalanches. While in the long term, the failures of the glacier slopes are controlled by a combination of increased temperature and rainfall from climate warming. Paul (2019) and Kääb et al. (2021) also indicated that the repeated glacier collapses and surges in the Amney Machen Mountain Range were mainly controlled by debris input from the headwall caused by regional temperature, which partly supports our viewpoint.

Therefore, long-term climate change and short-term meteorological conditions are the main driving factors behind the ice-rock avalanches in the Amney Machen Range. A variety of factors, including glacier retreat, de-buttressing at the base, crevasse expansion, permafrost degradation, rock degradation, as well as sliding surface lubrication all contributed to the avalanches. Among them, glacier retreat, de-buttressing, and crevasse expansion seem to be the main factors. Their changes to the overall morphology and stress state of the glacier are usually considered to be the most significant factors in large glacier failure (Paul 2019; Kääb et al. 2021; Li et al. 2024). They are primarily influenced by long-term climate conditions and play an important role in ice-rock avalanches under conditions of climate warming, especially regional warming in the QTP (Li et al. 2024). Meanwhile, permafrost degradation, rock degradation, and sliding surface lubrication also contribute to ice-rock avalanches. They lead to the formation of potentially low-strength sliding surfaces and further lubrication and softening, thereby increasing the probability of ice-rock avalanches. However, the impact of these three factors on glacier stability is often localized, gradual, and more likely to cause rock-ice avalanches with high rock content at the glacier terminus (Huggel et al. 2009; Krautblatter et al. 2013; Scherler et al. 2013; Coe et al. 2018; Deline et al. 2021). Considering the large seasonal span, high-altitude source area, and low rock debris content in the Amney Machen Mountains, the first three factors may be more critical than the latter three and are more likely to play a significant role.

Glacier retreat and ablation pattern

In recent years, mountain glaciers have undergone dramatic ablation and retreat worldwide, resulting in an increased frequency of glacier-related disasters, and glacier retreat and its ablation pattern contribute significantly to the formation and pattern of glacier disasters (Huggel et al. 2012; Roe et al. 2017; Chiarle et al. 2021; Deline et al. 2021; Stuart-Smith et al. 2021; Zou et al. 2023). According to field investigation and interpretation of high-resolution remote sensing imagery in 2021, the total glacier area in the Amney Machen Range is 83.36 km2, with elevations ranging from 4456 to 6282 m, and a maximum elevation drop of over 1800 m (Fig. 8a).

Fig. 8
figure 8

Glacier distribution and velocity in the Amney Machen Mountains: a glacier elevation in Amney Machen Mountains, b glacier slope in Amney Machen Mountains, c average glacier velocity from 2014 to 2023 in Amney Machen Mountains obtained by offset tracking, d elevation and slope distribution of high glacier velocity areas in Amney Machen Mountains from 2014 to 2023, and e average glacier velocity in Amney Machen Mountains from April to October 2016 and January to July 2019 obtained by offset tracking. The black lines represent glacier boundaries as well as mountain ridges

Full size image

The overall topography of the glacier region is characterized by a steep upper section and a more gentle lower section. The slope gradient ranges from 40 to 80° in the upper section and from 0 to 30° in the lower section (Fig. 8b). The major river systems on the southwest side of the Amney Machen Mountains include the Qinglong Gully and its tributaries the Qianlong and the Xiaoma Gullies. The Qinglong Gully flows northward, mainly recharged by meltwater and atmospheric precipitation.

Offset-tracking technology, based on Sentinel-1 data, monitored the glacier surface flow velocity in the Amney Machen Mountains from 2014 to 2023. A significant spatial variation in surface velocity is observed, closely related to topographic relief (Fig. 8c, d). High-velocity areas (> 0.1 m/day) are primarily concentrated in the middle glaciers, at elevations between 5000 and 5800 m and slopes ranging from 10 to 50°, and the maximum glacier flow rate reaches 0.24 m/day. As the elevation increases, the glacier flow velocity gradually decreases, with a velocity of approximately 0.07 m/day at the top of the glacier, while it also diminishes as the elevation decreases, approaching 0 m/day at the glacier terminus. The high-altitude steep upper glacier and the low-altitude gentle glacier are typically referred to as the accumulation zone and the ablation zone, while the transition zone near the equilibrium line between these two zones can be regarded as the equilibrium zone (Benn and Lehmkuhl 2000; Hock 2010). The material accumulation and downward driving force from the accumulation zone glacier in the Amney Machen Mountains provide the material source and dynamic basis for the ablation zone glacier, while the melting and disintegration of the ablation zone glacier cause de-buttressing and increase the instability of the accumulation zone glacier. The above factors together lead to a high surface velocity in the equilibrium zone located between the accumulation zone and ablation zone. Low friction over glacial ice and the lubrication effect from rain and meltwater make the glacier even more mobile (Huggel 2009; Khamis et al. 2015; Benn et al. 2019; Kääb et al. 2021). This makes it easier for high-speed glacier flow to happen on a gentler slope (0–10°) (Fig. 8d), with lower friction and stronger mobility being significant features of glacier movement. The distribution of glacier flow velocity of the Maqengangri Glacier in the half-year prior to the 2016 and 2019 ice-rock avalanches (Fig. 8e) shows that the high-velocity areas are mainly located in the upper source area and avalanche-prone area. This led to the formation of trailing edge tensile crevasses and a reduction in lower support, ultimately causing glacier instability and ice-rock avalanches, and also implied the possibility of another avalanche.

A supervised classification of remote sensing images can be used for fast glacier mapping. We use Landsat 8 OLI images (https://glovis.usgs.gov) for sample training and SVM classification to identify glaciers in the Amney Machen Mountains. The trend of glacier area changes was obtained (Fig. 9) by combining field investigations and visual discrimination for error correction. The interpretation results showed an overall retreating trend in the glaciers of the Amney Machen Mountains, primarily due to the recession of valley glacier tongues (Fig. 9a). Furthermore, there were distinct differences in glacier retreat across various elevation ranges. The classified statistics on the proportion of glacier retreat area at different elevations showed that glacier retreat predominantly occurred in the mid-low elevation glaciers between 4400 and 5600 m, accounting for 88.6% of the area. This was due to climate warming-induced glacier retreat and snowline rise mainly occurring in mid-low elevation regions, and the rate of glacier retreat gradually slowed down with increasing elevation (Fig. 9b). The total glacier area decreased from \(95.93 {\text{km}}^{2}\) in 2014 to \(83.36 {\text{km}}^{2}\) in 2021, with an average annual retreat area of \(1.80 {\text{km}}^{2}\) (Fig. 9c). The overall retreating trend was aligned with the warming climate (increasing annual temperature and precipitation) shown in Fig. 5a. In addition, the retreat rate has accelerated in recent years (2017–2021), potentially increasing the hazard of glacier disasters.

Fig. 9
figure 9

Glacier area change trends in Amney Machen Mountains from 2014 to 2021: a supervised classification results based on sample training and SVM classification, b the proportion of glacier retreat area at different elevation ranges, and c glacier area change trends in Amney Machen Mountains

Full size image

A stereo image-based DEM differencing was utilized to analyze the glacier elevation changes. The findings showed that the glaciers in the Amney Machen Mountains changed significantly from 2007 to 2021. The average cumulative change is − 9.13 m, and the average annual decrease is 0.65 m, indicating a rapid glacier ablation from 2007 to 2021 (Fig. 10a). The Maqengangri Glacier (point P1) showed the most significant thickening, with a cumulative value of 336.2 m. The glacier terminus at the Xiaoma Gully (point P2) showed maximum thinning, where the cumulative thinning is 182.8 m. Combining Fig. 10a and b, it can be observed that glacier thickness variations in the Amney Machen Mountains are elevation-dependent. The mid-lower glaciers, particularly glacier tongues, showed significant thinning due to ablation (blue dotted box). In contrast, upper glaciers on both sides of mountain ridges exhibited insignificant melting, as well as local freezing and thickening (red dotted box). This particular ablation pattern further reduced the stability of glacial glacier stability slopes, as demonstrated by the Xiaoma Gully slope in Fig. 10c. The inconsistent melting, thinning, or thickening of the mid-lower and upper glaciers made the already steep slope gradient in the source area even steeper, increasing the downward driving force and speeding up the movement process. Meanwhile, de-buttressing due to vanishing glaciers, melting-induced crevasses, and rock mass deterioration further accelerated the instability of the glaciers and increased the frequency of various glacier hazards.

Fig. 10
figure 10

Glacial retreat and ablation pattern from 2007 to 2021 in the Amney Machen Mountains: a, b changes in glacier elevation in the Amney Machen Mountains and the Xiaoma Gully and c elevation profile in 2007 and 2021 along section line AA′ in Fig. 10b

Full size image

In summary, the Amney Machen ice-rock avalanches occurred under the influence of the combined effects of long-term climate warming, short-term meteorological fluctuations, glacier retreat, and ablation patterns. Figure 11 illustrates the possible triggering mechanisms and evolution. The source area at Xiaoma Gully was initially in an overall stable state (Fig. 11a), but bedrock outcrops appeared due to the minor glacial detachment from 1987 to 1995 (Paul 2019), forming a weak spot in the glacier (Fig. 11b). Then, under long-term warming, glaciers continuously retreated and snowlines persistently rose. The steeper glacier slope, increased debris input and downward stress from the top, formation of trailing edge tensile crevasses, and weakening of basal support, caused the gradual transition of glacier in the source area into an unstable state, which ultimately induced small-scale ice avalanche events occurred in 2001 and 2003 (Fig. 11c) (Zou et al. 2023).

Fig. 11
figure 11

Formation mechanisms and dynamic evolution of the Amney Machen multiple ice-rock avalanches: a initial stable state of the glacier, b minor glacier detachment from 1987 to 1995, c minor ice-rock avalanches in 2001 and 2003, d repeated ice-rock avalanches from 2004 to 2016, and e further extension of tensile crevasses in the remaining glacier

Full size image

Eventually, the coupled effects of further climatic warming and short-term meteorological fluctuations repeatedly triggered four ice-rock avalanche events in February 2004, October 2007, October 2016, and October 2019 (Fig. 11d). The short-term meteorological factors likely mainly involved freeze–thaw damage and cycle due to short-term meteorological fluctuations, as well as lubrication and softening of bedrock due to meltwater and rainfall. All ice-rock avalanches developed in the northwestern slope of the Maqengangri Glacier, with the source area gradually expanding towards the southeast. The preceding event caused the leading-edge exposure and trailing-edge tensile deformation, resulting in the formation of concentrated, parallel large tensile crevasses in the remaining glacier. After repeated impacts of glacial avalanches, the tensile crevasses expanded, leading to decreased stability, and the remaining glacier gradually became an avalanche-prone area. Under continuous global warming and glacier retreat, the avalanche-prone mass could transform into another catastrophic ice-rock avalanche (Fig. 11e).

Discussion

Potential impact of seismic events

Seismic events are often an important trigger for landslide events, e.g., the 2008 Wenchuan earthquake triggered over 15,000 landslides, resulting in over 20,000 deaths (Yin et al. 2009). The Amney Machen Mountains are located in the East Kunlun Fault Zone, which is one of the largest and most seismically active fault zones in China. In 2001, the \({M}_{\text{w}}\) 7.9 Hoh Xil earthquake occurred here, triggering four large-scale ice avalanches (Zou et al. 2023). The Amney Machen mountain area has a well-developed geological structure, with the Amney Machen suture zone running through it, and is located adjacent to the North Qaidam suture zone, the North Qilian suture zone, and several large faults (Jiang et al. 2021). Therefore, it is necessary to investigate the potential impact of seismic events on ice-rock avalanches’ occurrence.

The time-distance-magnitude (TDM) distribution of earthquakes from 1993 to 2023 within 200 km of the ice-rock avalanche source area was plotted according to historical records (Fig. 12a). In the past 30 years, there were only two strong earthquakes with a magnitude exceeding \({M}_{\text{w}}\) 6.0, in 2021 at a distance of 110.5 km and with \({M}_{\text{w}}\) 7.3 and in 2000 at a distance of 66 km and with \({M}_{\text{w}}\) 6.1. Within 50 km, there were two events with magnitude exceeding 5.0 (both had magnitude 5.1), occurring at a distance of 9 km in 2009 and 49 km in 2021. The magnitude and spatio-temporal correlation of these events did not reveal a direct association with ice-rock avalanches (Fig. 12a). Furthermore, on December 18, 2023, a magnitude of \({M}_{\text{w}}\) 6.2 earthquake struck Linxia Prefecture, Gansu Province (35.70°N, 102.79°E), resulting in 151 deaths (Fig. 12b). The intense shaking triggered several destructive landslides in both Gansu and Qinghai. Although the Linxia earthquake occurred 320 km away from Xiaoma Gully, its impact on the Xiaoma Gully glacier needs to be observed due to its immense destructive power. After rough analysis, no detectable displacement or crack extension was observed in the avalanche-prone area after the Linxia earthquake (Fig. 12d), which indicates that the avalanche-prone area remained in a stable state.

Fig. 12
figure 12

Potential impact of seismic events on the Amney Machen ice-rock avalanches: a TDM distribution of seismic events within 200 km of the Amney Machen ice-rock avalanches from 1993 to 2023 and b impact of the Linxia earthquake on the avalanche-prone area

Full size image

Time series analysis of the avalanche-prone area

Under the influence of continued climate warming and glacier retreat, the Xiaoma Gully avalanche-prone area in the Amney Machen Range has the potential to experience ice-rock avalanches again. The SBAS-InSAR time-series analysis of the Xiaoma Gully avalanche-prone area based on the 2021–2023 Sentinel-1A descending orbit data assessed the likelihood of recurrence of ice-rock avalanches in recent years. The results showed that Xiaoma Gully has remained overall stable in recent years, with localized high deformation rate areas in the glacier zone (Fig. 13a). Attention should be directed to the regions with high deformation rates inside the avalanche-prone area and its trailing edge, which may be categorized into ablation and accumulation zones. The ablation zone was mainly distributed in the area above the avalanche-prone area, where tensile crevasses were generated and propagated due to climate warming and the traction force of the avalanche-prone mass. The unstable ice-rock mass, slowly detached from the headwall, supplying debris material into the avalanche-prone area and causing top loading. The accumulation zone, located at the mid-lower portion of the avalanche-prone area, consists of sporadic but high-deformation areas caused by ice-rock debris inputting from above. When the material accumulation and upper loading exceed the critical point, the avalanche-prone mass will destabilize (Deline et al. 2021; Allen et al 2022).

Fig. 13
figure 13

Time series analysis from 2021 to 2023 for the avalanche-prone area in Amney Machen Mountains: a displacement velocity, b cumulative displacement in ablation area, c cumulative displacement curve in accumulation area, and d planet remote sensing imagery of the avalanche-prone area in 2021 and 2023

Full size image

Figure 13b and c shows the total deformation at monitoring points in the ablation zone (points A, B, C, D) and the accumulation zone (points H, I, J, K). Affected by seasonal changes, temperature fluctuations, and extreme weather, the cumulative deformation in the Xiaoma Gully ablation zone and accumulation zone exhibits slight fluctuations (Li et al 2018; Brencher et al 2021). However, from 2021 to 2023, the deformation rate in and around the avalanche-prone area remained low. The avalanche-prone area experienced slight overall ablation, while the area above it showed slight accumulation, without signs of accelerated deformation. This indicated that the avalanche-prone mass exhibited a stable state and no well-developed slip surface had been formed. They showed no major changes in morphology or obvious tensile crevasse propagation in the avalanche-prone area or trailing edge (Fig. 13d). Therefore, although the Xiaoma Gully avalanche-prone area has a long-term possibility of recurrence, the likelihood of a large-scale event occurring in the short term is relatively low, with only potential for local small-scale ice avalanches. Of course, this is a reasonable assumption that does not take into account major earthquakes or extreme weather events.

Conclusion

Major ice-rock avalanche events in the QTP often exhibit characteristics in both time and space, extending spatially and temporally. A comprehensive analysis incorporating field investigation, meteorological record, and high-resolution remote sensing imagery reveals the triggering mechanisms and spatio-temporal evolution of the Amney Machen repeated ice-rock avalanches. Furthermore, the study also performs a time series analysis for the avalanche-prone mass. The main conclusions are drawn as follows:

  1. (1)

    Long-term climate warming and short-term meteorological fluctuations have caused a series of catastrophic ice-rock avalanches from a common source in the Amney Machen Mountains. The region has undergone significant climate warming, with mean annual temperatures rising 0.30 °C per decade and annual precipitation increasing by 23.3 mm per decade. Additionally, typical climatic features include concurrent rainy and hot seasons alongside significant diurnal temperature differences. The melting and collapse of lower glaciers due to regional warming have resulted in the destabilization and de-buttressing of the ice-rock system. The expansion and merging of fractures caused by freeze–thaw cycles have weakened the bedrock and potentially formed loose debris sliding surfaces. Furthermore, meltwater and rainfall have introduced an external driving force to the ice-rock system, accelerating sliding across lubricated and softened glacier beds. In summary, regional climate warming, freeze–thaw cycles, rainfall, and meltwater have collectively triggered Amney Machen ice-rock avalanches. Moreover, the presence of tensile crevasses and debris materials generated from previous events acted as material and dynamic foundations, facilitating subsequent events. Freeze–thaw effects, including frost heave and freeze–thaw cycles, may have contributed to the four ice-rock avalanche events.

  2. (2)

    Tracking spatio-temporal glacier evolution characteristics in the Amney Machen Mountains using offset-tracking, SVM classification, and DEM differencing indicates rapid glacier retreat in recent years. The glacier surface velocity reaches 0.24 m/day, the glacier retreat area is 1.80 km2/a, and the glacier elevation decreases by an average of 0.65 m/a. The Amney Machen Mountains glaciers show a pattern of ablation and thinning in the lower part and locally freezing and thickening in the upper part, with a maximum cumulative thickening of 336.2 m at the Maqengangri Glacier and a maximum cumulative thinning of 182.8 m at the source area. The ablation pattern of glaciers in the Amney Machen Mountains has increased the debris input and upper loading, causing additional downward driving forces in the source area. Simultaneously, the de-buttressing at the base and the formation of melt-induced crevasses due to glacier retreat have also exacerbated the hazard of different glacial disasters.

  3. (3)

    Spatio-temporal correlation analysis has been performed on seismic events within 200 km of the source area in Amney Machen Mountains over the past 30 years, and no direct connections were found between seismicity and the ice-rock avalanches. Time-series analysis of the Xiaoma Gully avalanche-prone mass indicated continuous low deformation rates from 2021 to 2023 without acceleration, suggesting a low likelihood of another large-scale ice-rock avalanche in Xiaoma Gully soon.