1 Introduction

Snow avalanches are among the cryosphere’s most violent land surface activities that profoundly influence mountain environment dynamics, evolution, and biodiversity (Schweizer, Bruce Jamieson et al.2003; Bebi et al. 2009; Korup and Rixen 2014). Snow avalanches seriously threaten transportation corridors, transmission lines, pipelines, and human activities in cold mountainous regions worldwide (Wastl et al. 2011; Reichel and Frömming 2014; Cui et al. 2021; Gaume and Puzrin 2021). In recent years, snow avalanches have been highly active in the Tianshan Mountains and its surrounding high mountains in Asia, killing more than 1,600 persons from 2009 to 2019 (Hao, Huang et al. 2021). Multiple catastrophic disasters with more than 100 deaths were caused by regional snow avalanche events triggered by heavy snowfall and sharp air-temperature increases (Ganju and Dimri 2004; Rafiq and Mishra 2018; Hao, Huang et al. 2021; Caiserman et al. 2022). Furthermore, human activities, such as mineral exploration and the construction of transportation and electric power facilities, are expanding to cold regions with the social and economic development in the High Mountain Asia, which challenges the snow avalanche risk management and regional sustainable development.

Regional snow avalanche events are defined as the release of a series of snow avalanches in a specific area within a short period (Schweizer, Kronholm et al. 2003; Techel and Schweizer 2017; Hao et al. 2022) and are often triggered by heavy snowfall, significant air-temperature increase, earthquakes, and strong winds (Mitterer and Schweizer 2013; Rafiq and Mishra 2018; Frigo et al. 2021). The frequency and type of snow avalanche release depend on the snowpack stratigraphy, terrain characteristics, snowfall, air temperature, and external triggering factors (Fromm et al. 2018; Frigo et al. 2021; Hao, Mind'je et al. 2021). The regional snowfall and snow cover change significantly with climate change (O’Gorman 2014; Le Roux et al. 2021), affecting the frequency and scale of snow avalanches. From 1946 to 2010, a 12% decrease in the proportion of powder dry snow avalanches had occurred in the French Alps (Eckert et al. 2013). The frequency and hazards of snow avalanches at mid-high altitudes have increased with climate warming in the western Himalayas as well (Ballesteros-Cánovas et al. 2018). The period of snow avalanche activity has shortened at lower altitudes, and the snow avalanche hazards have increased at higher elevations in the French Alps (Giacona et al. 2021). In addition, because higher snow densities in wet avalanche debris are likely to interfere with the respiration of buried victims, asphyxia and trauma as causes of snow avalanche death have increased with climate warming (Strapazzon et al. 2021). Wet snow avalanches are gradually becoming more prevalent due to ongoing global warming, and the snow avalanche risk at mid-high altitudes is increasing. Therefore, it is indispensable to assess the snow avalanche activity in a warming climate and to propose strategies to deal with snow avalanche disasters under different warming scenarios, which will facilitate road route planning to reduce the threat of snow avalanches and support integrated disaster risk management needs (Ballesteros-Cánovas et al. 2018; Shi et al. 2020; Strapazzon et al. 2021).

Many snow avalanches are the consequence of interactions between the terrain, snowpack, air temperature, and snowfall, so their characteristics are similar under the same climatic conditions (Schweizer, Bruce Jamieson et al. 2003; Hao et al. 2018; Gratton et al. 2020). Snow climates have been defined and extensively used to describe the general regional meteorology and snowpack characteristics of avalanche activity areas, aiming to provide helpful background information concerning the snow avalanche activity for snow avalanche risk management (Mock and Birkeland 2000). Snow avalanches are frequently associated with failures in a weak layer formed by depth hoar (Fukuzawa and Akitaya 1993; Reiweger and Schweizer 2013). Depth hoar will grow quickly and can persist for months in a continental snow climate where the density and shear strength of snow is characterized by colder temperature and less snowfall than those in a maritime snow climate with relatively warmer temperature and heavier snowfall (Shandro and Haegeli 2018). Therefore, snow avalanches in continental snow climate regions are more likely to be triggered under the same snowfall, earthquake, and strong wind conditions because the snowpack structure is weaker than that in other snow climate regions (Armstrong and Armstrong 1987; Hao et al. 2018). Snow avalanche activity characteristics and regional climate changes vary significantly under different snow climatic conditions, implying the differentiation of climate change impacts on snow avalanche activity.

Existing studies have also reported that the behavior of snow avalanches has changed. There are significant regional differences in marine and transitional snow climates (Ballesteros-Cánovas et al. 2018; Gratton et al. 2020; Giacona et al. 2021). The temperature increase rate in the Tianshan Mountains with a typical continental snow climate was 0.34 °C/10 years from 1960 to 2017, significantly higher than the global average temperature increase rate (Li et al. 2020). Significant snow depth and snow phenology changes have been observed in the Tianshan Mountains due to climate warming (Li et al. 2019; Wu et al. 2019). Theoretically, the response of snow avalanche activities to climate change in the Tianshan Mountains should be very apparent compared with other snow climate regions. However, changes in the snow avalanche frequency, size, and type under climate change in the region are rarely reported because of the lack of methods and long continuous historical observations. Snow avalanches in the Tianshan Mountains have been studied over the past several decades (Hu et al.1992; Liu et al. 2021) by focusing on identifying the paths and the triggering factors of snow avalanche activity to enable disaster managements and optimize road routes (Hu et al. 1992; Hao et al. 2018; Hao, Mind'je et al. 2021; Liu et al. 2021), while neglecting the temporal variability of snow avalanche activity in active areas. Therefore, how snowfall, air temperature, snowpack, and other factors shape the temporal distribution of snow avalanches remains unclear, limiting the assessment and understanding of snow avalanche activity under climate change.

In this study, we collected the meteorological and snowpack observation data for the period of 1968–2021, and snow avalanche activity data for the snow seasons from 2011 to 2021 along a transportation corridor in the central Tianshan Mountains in order to analyze temporal changes in snow avalanches and to examine the impacts of climate change on snow avalanches under continental snow climate conditions. The three objectives of this study were: (1) to determine the triggering conditions for regional snow avalanche activities; (2) to quantify the temporal distribution of the regional snow avalanche activities and determine how meteorological and snowpack factors shape the distribution of snow avalanches; and (3) to determine the trends of snow avalanche frequency and type with climate change. The findings will help in the rational arrangement of snow avalanche relief resources to improve snow avalanche risk management and provide essential information to understand the impact of climate change on global snow avalanche hazards.

2 Data and Methods

This section introduces the location of the meteorological, snowpack, and avalanche observation network in the central Tianshan Mountains, and the complete datasets collected in the past 50 years. Geomorphological setting and climate of the study area are summarized in Sect. 3.1, detailed descriptions of meteorological, snowpack, and avalanche datasets are given in Sect. 3.2, and the methods of data analysis are presented in Sect. 3.3.

2.1 Geomorphological Setting and Climate of the Study Area

The unique location of the central Tianshan Mountains helps the westerly circulation produce abundant snowfall during the cold season (Guo and Li 2015; Hao 2022). In addition, ice and snow erosion depressions are scattered throughout the region due to ancient glaciation, which provides favorable conditions for active avalanches (Hao et al. 2018; Hao, Huang et al. 2021). As a result, avalanches cause much trouble in constructing and operating roads through the central Tianshan Mountains. The G217 national highway (Fig. 1a), for example, is a crucial route that connects the northern and southern Xinjiang Uygur Autonomous Region. Because snow avalanches here paralyze traffic and cause casualties every winter and spring, the G217 Tianshan Mountains section (the Duku Highway) is closed from November to May. The G218 national highway (Fig. 1a) is also a vital traffic corridor that connects the western and eastern parts of Xinjiang through the Tianshan Mountains and extends to Central Asia and Europe. However, snow avalanches have seriously damaged the transportation systems and infrastructure of this traffic corridor in the Tianshan Mountains, causing severe losses to the regional economy and hindering regional development.

Fig. 1
figure 1

Location of the study area and the snow avalanche activity along a transportation corridor in the Tianshan Mountains. a Location of the Tianshan Station for Snowcover and Avalanche Research (TSSAR); b snow avalanche deposit blocking the G218 national highway; and c air-temperature variations at the TSSAR, 1968–2021

Full size image

The Tianshan Station for Snowcover and Avalanche Research (TSSAR), established in 1967, was built in the snow avalanche-prone areas along the Tianshan Mountains traffic corridor to understand avalanche risks and disaster prevention in the region (Fig. 1a). The TSSAR is located in the middle of the K329–K350 section of G218, with an altitude ranging from 1,700 to 2,000 m above sea level. A section of roughly 20 km experiences the worst conditions, with frequent traffic paralysis resulting from snow avalanches on G218 due to the surrounding steep slopes (Fig. 1b). Snow avalanches produce powerful shockwaves that can wreak havoc on highway infrastructure (Hao et al. 2022).

Here the snow season lasts from November to March because of significant snow presence ( Hao, Mind'je et al. 2021). Station observations show that the average annual temperature was approximately 2.0 °C from 1968 to 2021 and increasing significantly at a rate of 0.37 °C/10 years (Fig. 1c), which was consistent with the rapid warming rate of 0.34 °C/10 years experienced by the entire Tianshan Mountains from 1960 to 2017 (Li et al. 2020). We used a snow climate classification scheme (Mock and Birkeland 2000) to define and examine the snow climate in the study area. The analysis indicates that the 54 snow seasons had a typically continental snow climate and that the snow climate remained stable.

2.2 Data Collection

The location and the surrounding environment of the TSSAR observation station remained unchanged during the study period. An observation record of the meteorological and snow physical characteristics from 1968 to 2021 was compiled, covering temperature, atmospheric pressure, water vapor pressure, relative humidity, precipitation, evaporation, average surface temperature, and snow depth.Footnote 1 There are 64 snow avalanche-prone sites along the K329–K350 section of G218, and the details of these snow avalanche paths have been described by Yang et al. (2022) and Hao, Mind'je et al. (2021).

Because of the inadequate protective infrastructure, there are many locations on the highway with traffic cutoffs as a result of avalanches during the snow season. Consequently, when snow avalanches seriously affect road traffic conditions, the road is closed to reduce the threat of avalanches to vehicles and passengers. The staff of the traffic management department observes and records avalanche information along this section and takes emergency measures to eliminate the avalanche risk (Hao et al. 2022). This study statistically analyzed the snow avalanche data from 2011 to 2021. The traffic management department is primarily concerned with snow avalanche events that seriously affect highway traffic conditions and ignores snow avalanche events that have little impact on traffic. Therefore, only data concerning snow avalanche events that seriously affected road traffic conditions were used for the statistical analyses in this study.

2.3 Methods

Heavy snowfall, sharp air-temperature increase, earthquake, and strong wind often trigger a series of snow avalanches in a specific area within a short period (Techel and Schweizer 2017; Abermann et al. 2019; Hao, Mind'je et al. 2021). Because similar meteorological and environmental conditions lead to comparable snow avalanche situations, snow avalanche events can be classified into events triggered by snowfall, temperature rise, and other factors (that is, earthquake and strong wind). Based on meteorological variables strongly correlated with snow avalanche activity, all of the snow avalanche events from 2011 to 2021 were classified with hierarchical cluster analysis (Hao et al. 2018).

Snowfall-triggered snow avalanches are defined as snow avalanches that occur after or during a snowfall event, and are significantly positively correlated with three days of accrued snowfall before the snow avalanche release, which is considered the best threshold for forecasting a snowfall-triggered snow avalanche (Schweizer, Bruce Jamieson et al. 2003; Gauthier et al. 2010; Graveline and Germain 2016; Gauthier et al. 2017). To identify the snowfall conditions that trigger regional snow avalanches, we counted three days of accrued snowfall before all snowfall-triggered snow avalanche events and analyzed the quantitative relationship between the three days of accrued snowfall and the avalanche frequency.

High temperatures that result from air temperature increases cause the infiltration and accumulation of snowmelt water in the snowpack, leading to a noticeable loss of snowpack strength that can trigger snow avalanches (Ancey and Bain 2015). In this scenario, the snow is characterized by a relatively high liquid water content, so temperature-rise-triggered snow avalanches are wet snow avalanches ( Hao, Mind'je et al. 2021). The daily mean air temperature is generally used to distinguish snow avalanches from non-snow avalanche days and to forecast wet snow avalanches related to temperature increases (Baggi and Schweizer 2009; Peitzsch et al. 2012). Accordingly, the mean air temperature during an avalanche release is analyzed to identify the triggering conditions. After determining the meteorological conditions that triggered each regional snow avalanche, we analyzed the changes in the meteorological conditions that triggered regional snow avalanches from 1968 to 2021 to identify the impacts of climate change on snow avalanche activity.

Given the changes in snowpack and meteorological conditions at different stages of the snow season, the frequencies and types of avalanches also changed ( Hao, Mind'je et al. 2021). To quantify the frequency of snow avalanches during the snow season, we defined the relative frequency of snow avalanche activity as:

$$F_{i} = \frac{{n_{i} }}{N}(i = 1,2, \cdot \cdot \cdot ,m)$$
(1)

where \(F_{i}\) denotes the relative frequency of snow avalanche events in the \(i\)th time interval. A snow season is divided into m time intervals, with half a month being the time interval in this study; \(n_{i}\) is the cumulative number of regional snow avalanche events in the \(i\)th time interval from 2011 to 2021, and \(N\) is the total number of regional snow avalanche events from 2011 to 2021.

In this study, we adopted linear regression to estimate the linear tendencies of air temperature, snowfall, and snow avalanche parameters, using the F-test to determine the significance of the parameters’ time series. In addition, the intensity of the interannual changes in various meteorological and snow avalanche activity parameters was measured with the relative standard deviation (RSD) to assess the degree of dispersion of variables.

$$RSD = \frac{1}{{\mathop x\limits^{ - } }}\sqrt {\frac{{\sum\limits_{i = 1}^{n} {(x_{i} - \mathop x\limits^{ - } )^{2} } }}{(n - 1)}}$$
(2)

here, \(\mathop x\limits^{ - }\) denotes the average of all the samples, \(x_{i}\) is the value of the \(i\)th sample, and \(n\) is the total number of samples. A larger RSD value indicates a more drastic change in the parameter.

3 Results

This section presents the analyzed results of snow avalanche activity including the relationship between snowfall, air temperature, and snow avalanche frequency, and the temporal distribution of snow avalanche activities.

3.1 Relationship Between Meteorological Conditions and Snow Avalanche Activity

Compared with other snow seasons, the 2012/2013 snow season was characterized by the highest daily average snow depth (88.3 cm), a slightly lower mean air temperature (− 8.4 °C), and the most frequent snow avalanche activity (nine regional snow avalanche events), severely damaging public transportation (Fig. 2). Conversely, the lowest daily average snow depth (37.4 cm), a slightly higher mean air temperature (−4.9 °C), and only two regional snow avalanche events occurred in the 2011/2012 snow season. The contrasting comparison indicated that avalanche activity was more intense during a snow season with higher daily average snow depth. Furthermore, the statistical analysis showed that the number of regional snow avalanche events in the snow season was significantly positively correlated with the total snowfall (R = 0.85, P < 0.001) and daily average snow depth (R = 0.86, P < 0.001).

Fig. 2
figure 2

Variations in the maximum, minimum, and mean daily air temperatures, mean daily snow depth, and dates of regional snow avalanche releases for the 10 snow seasons (from November to March) of 2011/2012 to 2020/2021 in the Tianshan Mountain study area. Gray areas daily temperature range, Red lines mean daily air temperature, Black lines mean daily snow depth, Blue ellipses date of regional snow avalanche release

Full size image

The statistical analysis revealed that 50% of the total regional snow avalanche events were triggered by snowfall, possibly occurring throughout the snow season. Of the snowfall-triggered snow avalanche events, 25% occurred during or after snowstorms, with a snowfall of more than 14.6 mm within three days. Snowfall of more than 21.40 mm within three days triggered approximately 50% of the total snowfall-triggered snow avalanche events (Fig. 3a). The avalanche events triggered by a sharp rise in temperature and high solar radiation accounted for 31% of the total regional snow avalanche events. The daily mean temperatures ranged from −4 °C to 4 °C, and 87.5% of snow avalanches occurred when the daily mean temperature was between −3 °C and 3 °C (Fig. 3b).

Fig. 3
figure 3

Statistical relationship between the frequency of the snow avalanche activity and the meteorological conditions in the Tianshan Mountain study area. a Frequency of snowfall-triggered snow avalanche activity (F) versus 3 days of accrued snowfall (S); and (b) frequency of temperature-rise-triggered snow avalanches versus the daily mean air temperature

Full size image

3.2 Temporal Distribution of Snow Avalanche Activity

Snow avalanches began to appear in November and disappeared entirely by the end of March (Fig. 4a). A total of 46% of snow avalanche events occurred from February 15 to March 15, higher than during other periods. A Gaussian bimodal wave distribution characterized regional snow avalanche activity frequency. The first wave of regional snow avalanche activity was from November 1 to December 31, accounting for 23% of the total number of snow avalanches. The peak of the first wave occurred in the first half of December, accounting for 12% of the total snow avalanche events. The second wave occurred from January 1 to March 31, accounting for 77% of the total snow avalanches. The peak of the second wave occurred during the second half of February and the first half of March, accounting for 21% of the total snow avalanches. The following equation can express the temporal distribution of regional snow avalanche activity.

$$F_{i} = 2.6 + 31.1e^{{ - 6.9(i - 2.6)^{2} }} + 16.1e^{{ - 0.3(i - 8.4)^{2} }} (i = 1,2, \cdot \cdot \cdot ,m)$$
(3)

where, \(F_{i}\) denotes the relative frequency of a regional snow avalanche event of the \(i\)th time interval, the snow season is divided into m time intervals, and i = 1 denotes the first half of November.

Fig. 4
figure 4

Cumulative temporal distribution of snow avalanche activity frequency from 2011/2012 to 2020/2021 in the Tianshan Mountain study area. Cumulative temporal distribution of (a) the total snow avalanche activity; (b) snowfall-triggered snow avalanche activity; and (c) temperature-rise-triggered snow avalanche activity frequency

Full size image

The frequency of snowfall-triggered snow avalanche releases was characterized by high-low-high bimodal waves in a given snow season (Fig. 4b). Taking January 1 as the time turning point, the first wave of snowfall-triggered snow avalanche activity occurred from November 1 to December 31, accounting for 31% of the total number of snowfall-triggered snow avalanches. The second wave of snowfall-triggered snow avalanche activity occurred from January 1 to March 31, accounting for 69% of the total snowfall-triggered snow avalanches. Few snow avalanches were triggered by snowfall in the second half of December.

Temperature-rise-triggered snow avalanche events occurred from February 1 to March 31 (Fig. 4c). Snow avalanche events occurring primarily in the first half of March accounted for 56% of the total number of temperature-rise-triggered snow avalanche events. The active period of temperature-rise-triggered snow avalanches coincided with the transition period in the air temperature from below to above 0 °C (see Fig. 2).

4 Discussion

This section first presents the key factors that influence snow avalanche activity and discusses their influence on the temporal distribution of regional snow avalanches based on the previous sections. Then, the precipitation and air temperature changes in the past 50 years are shown, and the changing trends of snow avalanche frequency and type with climate change are discussed. Finally, avalanche risk mitigation strategies and future perspectives are summarized.

4.1 Factors Influencing the Temporal Distribution of Regional Snow Avalanches

Snow avalanches originate from the destruction of the snow layer, which likely fails on hillsides under the coupling effects of air temperature, snowfall, and wind (Schweizer, Bruce Jamieson et al. 2003; Dreier et al. 2016; Fromm et al. 2018). Accordingly, the evolution of snowpack and weather determine the temporal distribution of regional snow avalanches.

4.1.1 Impact of Snowfall and Snowpack on the Temporal Distribution of Regional Snow Avalanches

Our results disclosed an increasing trend in the frequency of regional snow avalanche releases with growing snow depth and snowfall, in agreement with previous studies (Esteban et al. 2005; Bocchiola et al. 2008; Reardon et al. 2008; Reuter et al. 2016; Veitinger and Sovilla 2016). In addition, snow avalanches are easily released under low snowpack stability (Reuter et al. 2016). The snowpack stability depends on the shear stress and strength of the weak layer or the basal snow layer and will decrease with the layer’s increasing shear stress (Schweizer, Bruce Jamieson et al. 2003; Reuter et al. 2016). Therefore, an increase in snow depth results in high shear stress on the weak layer or the basal snow layer, reducing the snowpack stability and indirectly increasing the probability of snow avalanche release (Bocchiola et al. 2008; Reuter et al. 2016; Veitinger and Sovilla 2016; Hao et al. 2022). Figure 5 shows that the snow depth first increased and then decreased during the snow season, reaching its peak in the second half of February. When there were no temporal differences in snowfall and air temperature during a snow season, the probability of snow avalanche release first increased and then decreased during a snow season, and snow avalanches occur most frequently in the second half of February.

Fig. 5
figure 5

Mean daily snow depth during the snow seasons from 2011/2012 to 2020/2021 in the Tianshan Mountain study area. The blue line indicates the mean daily snow depth, and the gray bars represent the standard error of the mean

Full size image

Snowfall is the critical factor that affects the temporal distribution of snowfall-triggered snow avalanches. Snowfall can directly increase the shear stress of the weak layer or the basal snow layer, reducing the snowpack stability and indirectly increasing the probability of snow avalanche release (Schweizer, Bruce Jamieson et al. 2003; Gauthier et al. 2010; Bair et al. 2012; Veitinger and Sovilla 2016). We found that snowfall shaped the temporal distribution of snowfall-triggered snow avalanche activity prior to January. During the early period of the snow season, the snowfall amount was higher, roughly distributing with a single peak (Fig. 6). At the same time, the snow depth had negligible impact on the snowpack stability because of the shallow snowpack (Fig. 5). Consequently, the temporal distribution of snowfall-triggered avalanches prior to January had a single peak, synchronous with the snowfall (see Figs. 4b and 6).

Fig. 6
figure 6

Temporal distribution of the mean snowfall per half month during the snow seasons from 2011/2012 to 2020/2021 in the Tianshan Mountain study area. The gray bars and black line represent the snowfall and its smoothed trend, respectively

Full size image

However, there were no significant temporal differences in snowfall amount since January, except for the relatively low snowfall amount in the first half of January (Fig. 6). The snow depth increased steadily from January to February, followed by a rapid decrease in March due to the combined effects of the accumulation and melting of snow (see Fig. 5). Therefore, the temporal distribution of snowfall-triggered avalanche activity since January had a single peak distribution synchronous with the snow depth, with the most frequent snow avalanche events occurring in late February when snow depth was deep (see Figs. 4b and 5). Snowfall shaped the temporal distribution of snowfall-triggered snow avalanche activity prior to January, while the snow depth dominated the temporal distribution after January. Therefore, the temporal distribution of the snowfall-triggered snow avalanche activity had a bimodal distribution because of the combined effects of snowfall and snow depth (see Fig. 4).

4.1.2 Impact of Air Temperature on the Temporal Distribution of Regional Snow Avalanches

Temperature-rise-triggered snow avalanches occurred from February 15 to March 31 (see Fig. 4c), when the average air temperature increased significantly from below to above 0 °C (Fig. 7). Simultaneously, snowmelt caused a sharp drop in the snow depth (see Fig. 5). The infiltration from surface snowmelt water due to significant temperature increase reduces the shear strength of the weak layer or the basal snow layer, resulting in weakening snowpack stability and driving snow avalanche release (Mitterer and Schweizer 2013; Dreier et al. 2016; Ceaglio et al. 2017; Fromm et al. 2018).

Fig. 7
figure 7

Daily air temperature during the snow seasons from 2011/2012 to 2020/2021 in the Tianshan Mountain study area. The blue line and gray bars indicate the daily air temperature and the mean standard error, respectively

Full size image

With its enhanced infiltration, the snowmelt water continuously accumulated and reduced the slope snowpack stability from the beginning to the middle of the transitional period from below to above 0 °C. But the snowpack stability increased with a dramatic snow depth decline due to the significant snowmelt at the end of the transitional period, despite the prolonged growing infiltration of snowmelt water at the snow-soil interface. The snowpack stability, therefore, first decreased and then increased due to the accumulated snowmelt water infiltration and the declined snow depth. Hence, the probability of temperature-rise-triggered snow avalanche occurrence first increased and then decreased from February 15 to March 31 (see Fig. 4c).

4.1.3 Combined Impacts of the Evolution of Snowpack and Meteorological Factors on the Temporal Distribution of Regional Snow Avalanches

The temporal distribution of regional snow avalanches is determined by the superimposition of the distributions of the snowfall-triggered and temperature-rise-triggered snow avalanches and other factors. The temporal distribution of the snowfall-triggered snow avalanche activity had a bimodal distribution due to the superimposed effects of the snowfall and snowpack development, while the temporal distribution of the temperature-rise-triggered snow avalanche activity had a single-peak distribution due to the combined effects of the air temperature and snowpack development. The results indicate that snowfall and significant air temperature increases triggered 81% of the total regional avalanche events. Furthermore, the frequency superimposition of regional avalanche activity triggered by snowfall and temperature rises exhibits a bimodal distribution. Therefore, the regional snow avalanche activity had a bimodal distribution (see Fig. 4a).

4.2 Impact of Snowfall Change on the Snow Avalanche Frequency

Around 50% of the snow avalanches were directly triggered by heavy snowfall in the study area. The frequency and amount of snowfall determine the frequency and scale of snow avalanche activity, and the snowfall trend directly affects the snow avalanche risk on long temporal scales (Hebertson and Jenkins 2003; Reardon et al. 2008; Christophe et al. 2010). There was no significant precipitation and snowfall trend in the study area from 1968 to 2021 (Fig. 8). However, the RSD of the snow seasonal snowfall (0.40) since the beginning of the twenty-first century was roughly 1.4 times that (0.29) before the twenty-first century (Fig. 8). Only 9% of the snow season snowfall exceeded 75% of all data prior to the twenty-first century. In contrast, 15% of the snow season snowfall exceeded 75% of all data since the twenty-first century, indicating an increasing trend toward extremely high snowfall frequency (Fig. 8). According to the observations at the TSSAR from 1968 to 2021, snowfall events that exceeded the snowfall thresholds of 14.6 mm and 21.4 mm for three consecutive days induced 25 and 50% of snowfall-triggered snow avalanches, respectively. The frequency of snowfall events at both snowfall thresholds showed a slight and insignificant increase, implying an increase in snowfall-triggered snow avalanches (Fig. 9).

Fig. 8
figure 8

Variations in the precipitation and snowfall at the Tianshan Station for Snowcover and Avalanche Research (TSSAR) from 1968 to 2021. The blue and red dashed lines indicate the linear trend of the seasonal precipitation and snowfall, respectively

Full size image
Fig. 9
figure 9

Variation in the number of snowfall events from 1968 to 2021 in the Tianshan Mountain study area. The black bars indicate the number of snowfall events with more than 21.4 mm snow for three consecutive days, and the red line indicates the linear trend. The gray bars indicate the number of snowfall events with more than 14.6 mm snow for three consecutive days, and the blue line indicates the linear trend

Full size image

The snowfall events exceeding 21.4 mm snow for three consecutive days increased at a higher rate than those exceeding 14.6 mm snow for three consecutive days (Fig. 9), meaning that the increase in the number of snowfall events with more high snowfall amounts was more pronounced. Snowfall events with high snowfall amounts within a relatively short period commonly will induce larger snow avalanche sizes and more significant hazards (Reardon et al. 2008; Schweizer et al. 2020; Hao, Mind'je et al. 2021). The growing extreme high snowfall events in the study region implied that the frequency of large-scale and high-impact snow avalanches was increasing. There were five snowfall events with more than 21.4 mm of snowfall for three consecutive days prior to the twenty-first century, meanwhile six events were observed since the twenty-first century. Accordingly, the number and scale of snow avalanches induced by heavy snowfall have been higher since the twenty-first century than before. In the 2001/2002, 2010/2011, and 2012/2013 snow seasons, the snowfall events exceeding 21.4 mm snow in three consecutive days were much higher than the average, resulting in frequent massive snow avalanches. The TSSAR was buried and severely damaged twice by snow avalanches that were released in a nearby valley in February 2000 and 2010.

More heavy snowfall events will drive greater impact snow avalanches that will enhance snow avalanche risk. Given that the traffic flow and road networks are expanding with the social and economic development in the Tianshan Mountain area, improving the capacity of avalanche disaster prevention and mitigation is of paramount importance. However, it is challenging to meet the needs of avalanche disaster prevention and mitigation just by relying on fixed snow avalanche protection projects because of their high construction costs and unfavorable construction conditions in cold regions. Therefore, avalanche disaster prevention methods must be adjusted, optimized, and upgraded. A snow avalanche early warning system, for example, will enable the emergency department to improve the emergency plan and site proposal measures, such as evacuating people and closing roads to avoid high-impact snow avalanches as much as possible. So we should strengthen the avalanche monitoring and early warning capability based on the existing avalanche prevention system to cope with the increasing snow avalanche risk in a warming climate.

4.3 Impact of Climate Warming on the Temporal Distribution of Regional Snow Avalanches

Temperature-rise-triggered snow avalanches occurred in the study area from February 15 to March 31 (see Fig. 4c) during the transition period of air temperature from below to above 0 °C. The active stage of temperature-rise-triggered snow avalanches was consistent with the period of apparent temperature increase. The combined effects of the increasing solar radiation and temperature accelerated snowpack melt with a significant increase in the snowpack water content, indicating a wet snow state. Therefore, temperature-rise-triggered snow avalanches occurred during the snowmelt period and represented wet snow avalanches. We found that the temperature at the TSSAR increased significantly at a rate of 0.37 °C /10 years (see Fig. 1c), with a prolonged snowmelt period and rising daily average temperature in the snow season, which has been observed for the entire Tianshan Mountains as well as in other regions worldwide (Peng et al. 2013; Beniston et al. 2018; Yang et al. 2019). The advancement of the snowmelt period means that the active time period for wet snow avalanches triggered by dramatic warming will occur earlier in the snow season.

The time period with a daily average temperature from –3 °C to 3 °C has shifted significantly forward in the snow season (Fig. 10); 87.5% of the temperature-rise-triggered snow avalanches occurred within this temperature interval. During the snow seasons at the TSSAR from 1968 to 2021, the temperature range of wet snow avalanches triggered by the sharp air-temperature increase was primarily concentrated after March 20, with most wet avalanches occurring from March 10 to March 30 of 1968−2000, and from March 1 to March 20 of 2000−2021. The time period of wet snow avalanche activity gradually advanced from the second half of March to the first half of March. As a result, the wet snow avalanche activity has significantly advanced, coinciding with our observations that wet snow avalanches triggered by apparent warming from 2011 to 2021 frequently occurred in the first half of March. Here we take the median of the period with a daily average temperature from −3 °C to 3 °C as the active time for wet snow avalanches. When the regional temperature increases by 1 °C, the active time for wet snow avalanches advances by approximately 8 days. Increasing winter air temperatures drove the advance and prevalence of wet avalanche activity in the Tianshan Mountains, given its continental snow climate, similar to previous findings reported in snow climate regions worldwide (Ballesteros-Cánovas et al. 2018; Giacona et al. 2021; Strapazzon et al. 2021). Therefore, it is urgent to adjust the risk prevention and control time for wet snow avalanches and rationally arrange wet snow avalanche relief resources to improve the regional risk management of wet snow avalanche disasters.

Fig. 10
figure 10

Variations with a daily average temperature from −3 °C to 3 °C during the snow seasons at the Tianshan Station for Snowcover and Avalanche Research (TSSAR) from 1968 to 2021. The gray bars represent the periods with a daily average temperature from −3 °C to 3 °C, and the blue triangle indicates the median of that period

Full size image

5 Conclusion

Understanding the temporal distribution of regional snow avalanche activity and its changes with climate change can improve snow avalanche risk management in areas with frequent snow avalanche disasters. We investigated the regional snow avalanche activity in the central Tianshan Mountain study area and analyzed the relationship between the snowfall, air temperature, snow depth, snowpack dynamic, and the frequency of the regional snow avalanche activity. The results revealed that the snow avalanche frequency changed dramatically during the snow season and was characterized by a Gaussian bimodal distribution. The temporal distribution of the regional snow avalanche activity consisted of the superimposed effects of the interactions between the snowfall, snow depth, air temperature, and snowpack development.

Ongoing climate change has been impacting the regional snowfall and air temperature, thereby changing the behavior of the snow avalanche activity. We found that snowfall events that exceed the snowfall threshold of 21.4 mm for three consecutive days induced about 50% of snowfall-triggered snow avalanches, enhancing the snowfall-triggered snow avalanche risk at the TSSAR from 1968 to 2021. In addition, the time period of wet snow avalanches triggered by a sharp air-temperature increase gradually advanced from the second half of March prior to 2000 to the first half of March since 2000 due to the warming climate. These findings highlight the need to immediately propose avalanche risk mitigation strategies to improve the regional capacity to adapt to climate change.

Our findings demonstrate how snowfall, air temperature, and other factors control the fluctuations in the snow avalanche activity, hopefully providing insights into the impacts of climate change on the snow avalanche activity in the Tianshan Mountains. However, it is challenging to systematically examine the processes of the dynamic change chain of climate-snowpack-snow avalanche activity due to the lack of observational data related to snowpack and snow avalanche activity. Moreover, it is difficult to understand the response mechanism of snow avalanche activity to climate change, which dramatically limits the prediction ability of snow avalanche risk with climate change over a broader study region. Therefore, we will update the incomplete data concerning physical snow characteristics and snow avalanche activities using tree rings and snowpack models to reveal the response mechanisms of snow avalanche activity to climate change, to enable a more accurate prediction of snow avalanche risks under different climate models and to support the regional disaster risk prevention of snow avalanches.