60-year changes and mechanisms of Urumqi Glacier No.1 in the eastern Tianshan of China,Central Asia

ZhongQin Li,HuiLin Li,ChunHai Xu,YuFeng Jia,FeiTeng Wang,PuYu Wang,XiaoYing Yue

Tianshan Glaciological Station/State Key Laboratory of Cryospheric Sciences, Northwest Institute of Eco-Environment and Resources,Chinese Academy of Sciences,Lanzhou,Gansu 730000,China

ABSTRACT Worldwide examination of glacier change is based on detailed observations from only a small number of glaciers. The ground-based detailed individual glacier monitoring is of strong need and extremely important in both regional and global scales.A long-term integrated multi-level monitoring has been carried out on Urumqi Glacier No. 1(UG1)at the headwa‐ters of the Urumqi River in the eastern Tianshan Mountains of Central Asia since 1959 by the Tianshan Glaciological Sta‐tion,Chinese Acamedey of Sciences(CAS),and the glaciological datasets promise to be the best in China.The boundaries of all glacier zones moved up, resulting in a shrunk accumulation area. The stratigraphy features of the snowpack on the glacier were found to be significantly altered by climate warming. Mass balances of UG1 show accelerated mass loss since 1960,which were attributed to three mechanisms.The glacier has been contracting at an accelerated rate since 1962,resulting in a total reduction of 0.37 km2 or 19.3%from 1962 to 2018.Glacier runoff measured at the UG1 hydrometeoro‐logical station demonstrates a significant increase from 1959 to 2018 with a large interannual fluctuation,which is inverse‐ly correlated with the glacier's mass balance.This study analyzes on the changes in glacier zones, mass balance, area and length, and streamflow in the nival glacial catchment over the past 60 years. It provides critical insight into the processes and mechanisms of glacier recession in response to climate change. The results are not only representative of those gla‐ciers in the Tianshan mountains, but also for the continental-type throughout the world. The direct observation data form an essential basis for evaluating mountain glacier changes and the impact of glacier shrinkage on water resources in the in‐terior drainage rivers within the vast arid and semi-arid land in northwestern China as well as Central Asia.

Keywords:Urumqi Glacier No.1;glacier change;climate change;glacier zone;the Tianshan Mountains

1 Introduction

The shrinkage of alpine glaciers is on a global scale,and the rate of reduction appears to have acceler‐ated during the past several decades. From 1961 to 2016, alpine glaciers contributed 27±22 mm to global mean sea-level rise,accounting for 25%to 30%the ob‐served total. Their mass loss is equivalent to the sealevel contribution of the Greenland Ice Sheet, clearly exceeding that from the Antarctic Ice Sheet. Current mass-loss rates indicate that glaciers could almost dis‐appear in some mountain ranges within this century,while heavily glacierized regions will continue to con‐tribute to sea-level rise beyond 2100 (Watsonet al.,2015;Zempet al.,2019).Worldwide glacier retreat and associated future runoff changes raise major concerns over global water resources (Huss and Hock, 2018).However,examination and prediction of glacier change worldwide are based on detailed observations from a small number of glaciers due to the inaccessibility of many glacier areas (Dyurgerov and Meier, 2000;Zempet al., 2009).Thus, the ground-based detailed individu‐al glacier monitoring is essential and extremely impor‐tant for studying the processes and mechanism of gla‐cier changes on both regional and global scales.

Urumqi Glacier No. 1 (UG1) is a continental(cold) glacier located in the eastern Tianshan Moun‐tains,Central Asia.The glacier has been observed sys‐tematically on a regular basis back to 1959, which promises to produce the best glaciological datasets in China. Based on the data, a number of studies have been carried out on many aspects of UG1 since 1960.Classical international glaciology theories are put for‐ward based on maritime (temperate) glacier and ice sheet studies. Observations and investigations of UG1 fill in many gaps and constitute vital developments and contributions to international glaciology (Li, 2019).For example,according to observations of UG1,a gla‐cier zone division theory was established by combin‐ing the international glaciology theories with Chinese characteristics (Xie and Liu, 2011). Additionally,based upon observation of the artificial ice cave in UG1,Chinese glaciologists proposed that four mecha‐nisms are responsible for the movement of clod gla‐cier: glacier ice deformation, ice bed deformation,shearing, and bottom sliding (Liet al., 2018).The hy‐drological research of the Urumqi River Basin has formed the basic framework for the observation and experimentation of hydrometeorological processes in the glacier-covered basin, laying the foundation for studying the hydrological of inland river basins in China (Yang, 1991). This article summarizes the ob‐served changes of the glacier over the past 60 years and reveals the mechanisms related to these changes.This will greatly improve our understanding of alpine glaciers'response to climate change on a global scale.

2 Description of Urumqi Glacier No. 1

UG1 is a northeast-facing valley glacier located at the source area of the Urumqi River in the eastern Tianshan, the core area of Central Asia. UG1 is com‐posed of east and west branches,covering 1.95 km2in 1962 and 1.54 km2in 2017. It flanks the highest peak(Tianger Peak II), which, located in the southeastern Tianshan, has an elevation of 4,484 m above sea lev‐el. These two branches became two separated but re‐lated smaller glaciers in 1993 due to continued glacier shrinkage.

Situated at the center of the Eurasian continent, the eastern Tianshan is surrounded by vast desert areas with the Taklimakan to the south, the Gurbantunggut Desert in the Junggar Basin to the north, and the Gobi Desert to the east(Figure 1).With a typical continental climate, the westerly jet prevails high above the moun‐tains.The study area is under the influence of the Sibe‐rian anticyclonic circulation during winter, resulting in minimal precipitation in January. Over 90% of the pre‐cipitation occurs from April through September, with the maximum in July. Summer precipitation maxima coincide with snow and ice melt in the nival and ice zone (Li, 2019). As a summer-accumulation glacier with both accumulation and ablation happening in the summer,UG1 is quite different from glaciers in the Eu‐ropean Alps, where more precipitation takes place dur‐ing winter as snowfall to effectively protect the glacier from melting(Fujitaet al.,2008;Ludwiget al.,2016).

Figure 1 Geographic environment around the eastern Tianshan showing the vicinity of the study site,including the deserts,Gobis and the city of Urumqi

UG1 was selected for study for its representative‐ness and the fact that it serves as an important water supply for Urumqi,the provincial capital city of Xinji‐ang Uygur Autonomous Region, in 1959. Since then,it has been monitored by the Tianshan Glaciological Station (TGS), 3 km southeast of the glacier, of the Chinese Academy of Sciences (CAS). In the interna‐tional community, UG1 is one of the reference gla‐ciers in the World Glacier Monitoring Service(WGMS), and is considered a key element of the sys‐tem because of its unique geographical position in ar‐id Central Asia.Long-term monitoring of UG1,a con‐tinental-type (cold) glacier, thereby represents an irre‐placeable piece of the global mosaic in that it comple‐ments similar programs covering more maritime-type(temperate) glaciers and the ones in transitional cli‐mates within polar, temperate and tropical regions(Atsumu,2011).

3 Observation, methodology and data

The scope of observation includes various glacio‐logical elements, such as the property of snow stra‐tigraphy, glacier mass balances, terminus (glacier front) position and area changes, ice thickness, sur‐face velocity, englacial temperature and glacial cli‐matology, and glacial hydrology,etc..Glacier zone characteristic is one of the fundamental properties of mountain glaciers and demonstrates variations of abla‐tion area in different periods. Mass balance reveals the gain or loss in glacier mass and is a direct and im‐mediate indicator of climate evolution. Glacier length and area are well recognized as high-confidence indi‐cators of glacier changes?the comprehensive and de‐layed glacier responses to climate change.Detailed in‐dividual glacial runoff observation is imperative for evaluating glacier recession and water resource change on both regional and global scales. Hence,this study focuses on the long-term changes of a few key elements, including glacier zones, glacier mass balance, glacier length and area, and runoff of the UG1 catchment.

Physical properties of the snow-firn pack on UG1 and ice formation processes have been measured peri‐odically since 1959, including the composition of snow-firn stratigraphy, density, snow temperature,etc..Continuous datasets were obtained for four periods:1961?1962, 1980?1986, 2003?2007, and 2013?2017 from more than 80 snow-firn pits.Based on the prop‐erty of snow-firn stratigraphies, the range of super‐imposed-ice, and the position of snow line, the gla‐cier zones (ice formation zones) in different periods were classified according to Shumskii's theory(Shumskii,1964).

The mass balance of UG1 has been observed by the stake/snow pit method (Elderet al., 1992; Dy‐urgerovet al., 2000) since 1959. The observed data are available from 1959 through 2018, except for 1967?1979.The missing data from 1966 to 1978 was reconstructed by meteorological data collected from nearby stations (Zhanget al., 1981, 1984; Ji and Tang, 1994). The single-point specific mass balance has been measured monthly from May 1 to August 31 using the stake/snow pit method. The stake network consists of not less than 42 stakes drilled into the gla‐cier surface and evenly distributed with elevation.Using the data of specific mass balance and addition‐al snow pits, glacier-wide mass balance, annual net accumulation and ablation from September 1 in the previous year to August 31 are calculated by contour maps of accumulation and ablation.The mass balance data were published in annual reports of the Tianshan Glaciological Station from 1980 ?2018, the Glacier Mass Balance Bulletin (Nos. 1?12), and Global Gla‐cier Change Bulletin (Nos. 1 ?3) compiled by the World Glacier Monitoring Service (https://wgms.ch/global-glacier-state/).

Although initial observation of glacial terminus position began in 1959,regular monitoring did not oc‐cur until 1980. Terminus variations are measured in field surveys yearly. Since the glacial split in 1993 from melting, separate measurements have been per‐formed for the two branches. From 1962 throughout 2018,the area of UG1 was assessed 11 times by using different field survey techniques, including theodolite,plane table, ground-based stereophotogrammetry, to‐tal station, GPS?RTK, and laser scanning (Xuet al.,2019). These data can be found in the annual reports of the TGS from 1980 through 2018.

Streamflow and meteorological parameters are ob‐served at three hydro-meteorological stations at the headwaters of the Urumqi River. This study uses the runoff data from UG1 hydro-meteorological station to examine the long-term changes of glacier meltwater.Observation at the gauge station began in 1959 and was interrupted during 1967 ?1979; the missed data were reconstructed using the meteorological data from adjacent stations (Yang, 1991). Regular field da‐ta collection is performed from May to September each year. Specifically, the observed water level re‐cords are converted to discharges based on rating curves,which are used to evaluate daily,seasonal,and annual volumes in cubic meters. The annual runoff depth in mm is obtained by dividing the annual vol‐ume with the catchment area. Over 95% of the annual runoff at the stations occurs during the observation pe‐riod, as the stream is mostly frozen for the rest of the year. The runoff data had been compiled and internal‐ly published in the annual reports of the TGS from 1980?2018.

4 Results and discussions

4.1 Glacier zone and snow-firn stratigraphy

4.1.1 Decadal change of glacier zone

Glaciers have vertical zonality due to the large el‐evation difference.With increasing elevation, the hy‐drothermal conditions vary greatly, resulting in changing ice formation processes. There are two classical theories for categorizing glacier zones (also referred to as ice formation zones). Shumskii (1964)divided a glacier into seven zones that, starting from the glacier head, include recrystallization zone, re‐freezing-recrystallization zone, cold infiltration-re‐crystallization zone, warm infiltration-crystallization zone, infiltration zone, infiltration-freezing zone and ablation zone. In contrast, Paterson (1994) partitioned a glacier into five zones,i.e., dry-snow zone, percola‐tion zone, wet-snow zone, superimposed ice zone,and ablation zone.

As a glaciology studies pioneer in China, Xie(1965) applied Shumskii's glacier zone theory on UG1 and proposed a first classification based on the snow-firn stratigraphies, range of superimposed-ice,and the snow line position observed during 1961 ?1962.Afterwards, other researchers continued this ef‐fort for different periods according to the same theo‐ry. Figure 2 compares the distribution of glacial zones of UG1 in four representative observation periods:1961?1962 (Xieet al., 1965, 1989), 1980?1986 (Liu,1989), 2003?2007 (Liet al., 2011) and 2013?2017(this study). The glacier consisted of four zones dur‐ing 1961 ?1962: a recrystallization-infiltration zone,an infiltration zone, an infiltration-freezing zone, and an ablation zone. The recrystallization-infiltration zone was characterized by relatively poor development of melting, involving no more than 20% of the total mass of deposited snow, and minor meltwater with‐out penetrating through the annual snow layer(Shumskii, 1964; Xieet al., 1965). However, this zone was taken over by the infiltration zone and dis‐appeared by 1980 ?1986. Meanwhile, the boundar‐ies of all other zones moved up-glacier, and the ac‐cumulation area decreased.The zone map for 2003?2007 shows that in addition to the continued change of the zones consistent with a warmer cli‐mate, a small area at the glacier head of the east branch exhibits features of an infiltration-freezing zone. In 2013 ?2017, while the ablation zone ex‐panded, both the infiltration zone and infiltrationfreezing zone apparently shrunk further, especially at the west branch of UG1. The infiltration-freezing zone of the east branch of UG1 became very nar‐row in that period, implying that less and less ice was generated in late summer and autumn to super‐impose onto the glacier due to the accumulated ef‐fects of climate warming.

Figure 2 Temporal changes in glacier zones of UG1

4.1.2 The mechanisms for glacier zone changes

Glacier zones (ice formation zones), classified by snow stratigraphy and ice formation processes, are sensitive to climate change.They have changed appar‐ently under global warming, leading to the disappear‐ance of cold-type ice formation and its corresponding recrystallization zone or dry-snow zone. The transi‐tion of the recrystallization-infiltration zone into an in‐filtration zone results from a change in the stratigraph‐ic properties and the ice formation processes. This change has been observed at the PGPI (the Program for Glacier Processes Investigation) observation site located at 4,130 m above sea level on the east branch(Liet al.,2006).During the 1960s,the snow stratigra‐phy at the PGPI contained a number of thin ice layers,ice lenses, and thin dust layers. However, most of the ice layers ice lenses have melted with increased perco‐lation of meltwater in the current time. In addition,the meltwater accumulates the dust particles into thick layers.The proportion of coarse-grained firn, which is usually formed by infiltration water, was found to have increased from 40.7% to 74.8% during 1962 ?2006, indicating increased meltwater in the stratigra‐phy(Youet al.,2005;Wanget al.,2007).

4.2 Mass balance

4.2.1 Long-term trend of mass balance

Determined by mass/energy processes mainly oc‐curring on a glacier surface,mass balance presents the glacier's accumulation and ablation rates. It displays an undelayed response to climate variation and in-situ meteorological conditions (Xieet al., 1996; Braith‐waite,2002;Haeberli and Holzhauser,2003).Both an‐nual and cumulative mass balances of UG1 show ac‐celerated mass loss since 1960 (Figure 3). Two accel‐erated melting processes are visible. Beginning around 1985, the first accelerated melting brought about the change rate of the cumulative mass balance from?81 mm/a during 1960?1984 down to ?273 mm/a dur‐ing 1985?1996. Starting in 1997, the second one fur‐ther reduced the change rate to ?684 mm/a during 1997?2019. Compared to the first one, the second ac‐celerated melting was more vigorous. In 2010, the ice mass loss increased to 1,327 mm, the most since the data were available. Since 2011, the mass balance has shown a substantial annual fluctuation. After a slow‐down from 2011 to 2014,it has again surged to a state of high mass deficit. The cumulative mass balance of UG1 between 1960 and 2019 reached ?20,606 mm.

The observed annual mass balance of UG1 is found to coincide with the mean annual mass balance of 30 reference glaciers worldwide.This finding is im‐portant because it indicates that the investigation of UG1 can reveal the average mass balance change of mountain glaciers on a global scale.The generic drive forces for the accelerated wastage of UG1 may also apply to many continental mountain glaciers.

Figure 3 Annual net and cumulative mass balances of Glacier No.1 from 1960 to 2019

4.2.2 The mechanisms for accelerated mass loss

Glacier mass balance is determined by the energy budget (balance) on the glacier surface. The energy comes mainly from solar radiation. Therefore, a mass balance model coupled with a glacier surface energy balance model is required to explain the mass change.A previous study based on a mass balance model re‐vealed three mechanisms responsible for an increased mass loss of UG1(Liet al.,2011).The first is air tem‐perature rise during the melting season as described by positive degree days (PDD, the sum of daily mean air temperatures above the melting point in a year),which directly boosts the glacier melting rate.The sec‐ond is the ice temperature rise of UG1, which reduces the heat needed to raise the surface ice to its melting point and in the refreezing of the percolated meltwa‐ter.As a result, the risen ice temperature enhances the glacier's sensitivity to climate warming, making it an important contributor to the accelerated mass loss.The third is the albedo reduction of the glacier surface due to the continued growth of the ablation area. The positive feedback of glacier recession on the ablation area has most likely resulted in an accelerated mass loss. The low albedo in the ablation area was engen‐dered by mineral dust, which mainly originates from the surrounding Asian deserts.Another factor respon‐sible for the low albedo is the organic matter on the dust surface, such as living cyanobacteria, which pro‐liferates as temperature rises.

4.3 Terminus and area

4.3.1 Long-term changes of glacier terminus and area

The length reductions of UG1 are reflected by the retreat of its terminus, as illustrated in Figure 4. From 1980 to 1993,the terminus of UG1 continually retreat‐ed with an average rate of 3.6 m/a.When the glacier's east and west branches separated in 1993, two termini corresponding to the two branches began to exist(Fig‐ure 4). During 1994?2019, the mean retreat rates of the west and east branches' termini were 5.7 m/a and 4.9 m/a,respectively. Prior to 2011, the recession of the west branch was consistently faster. Afterwards,the shrinkage rate of the west branch terminus fell,whereas that of the east branch terminus surged, pass‐ing that of the west branch terminus and remaining as the one with a faster retreat rate,except for 2017.

Figure 4 Terminus retreat rates of Urumqi Glacier No.1 in the east Tianshan from 1980 to 2019

Figure 5 illustrates the morphometrical evolution of the glacier. Similar to terminus retreat, UG1 has been shrinking at an accelerated rate since 1962, with a reduction of 0.37 km2or 19.3% from 1962 to 2018,at an average loss rate of 0.007 km2/a. Specifically,the glacier contracted by 0.003 km2/a during 1962 ?1986 and 0.008 km2for 1994?2018,about three times that of the former period, indicating a pronounced ar‐ea reduction since 1994. The separation of the two branches obviously accelerated the shrinkage in both the area and termini of UG1 because it amplified the exposed area of the glacier front and thus enhanced melting(Liet al.,2011).

4.3.2 The mechanism and simulation of glacier geometric change

The geometric changes of a glacier, such as varia‐tions in area, length, and thickness, are governed by both the mass balance and the dynamic process relat‐ed to its topographical and thermo-dynamical parame‐ters.Alterations in glacier terminus and area represent the integration of many climate-related events, occur‐ring in as short a period as one year or over centuries,depending on the glacier's size and thickness and oth‐er factors (e.g., Harper, 1993; Pelto and Hedlund,2001; Liet al., 2010; Liet al., 2012; Liet al., 2019).These changes are indirect, filtered, and enhanced re‐sponses of the glacier to climate change. To simulate the geometric change,Liet al.(2019)carried out a de‐terministic dynamic modeling study. Based on the long-term observation datasets, they analyzed the re‐sponse of area, length, volume, and glacial runoff of UG1 by a High-order Ice Flow model coupled with a simplified surface energy/mass balance model. Cli‐mate forcing was specified by IPCC emission scenari‐os and by observed warming trends at nearby meteo‐rological stations.Also, the processes of glacier chang‐es under different climate scenarios in the future were systematically elucidated. Modeling results show that the glacier will recede substantially and ultimately vanish within the next 50?90 years under different cli‐mate scenarios(Figure 6).

Figure 5 Terminus and area variations of Urumqi Glacier No.1 from 1962 to 2018

4.4 Changes in streamflow

4.4.1 Long-term change

The runoff measured at the UG1 hydrometeoro‐logical station, as shown in Figure 7, has been ana‐lyzed by previous studies (e.g., Yang, 1991; Kanget al., 1997; Liet al., 2003; Liet al., 2010; Jiaet al.,2020). It demonstrates a significant increase from 1959 to 2018 with substantial interannual fluctuation.It can be roughly divided into two stages: 1959?1992 and 1993?2018 (Figure 7). During the first stage, the runoff values were relatively low, with an average of 157.39×104m3. In the second stage, the runoff rose consistently,particularly in its baseline.The mean val‐ue reaches as high as 271.78×104m3, an increase of 114.39×104m3, or 1.7 times that of the first stage. At present, the runoff stays at a high level with large in‐terannual variability.

Figure 6 Predicted variations of glacier length(a),area(b),volume(c),and runoff(d)in the future under different climate scenarios.Note that DXG1(DXG 1959?2004)and DXG2(DXG 1980?2004)represent two warming scenarios constructed by extending measured air temperature from the Daxigou Meteorological Station(about 3 km southeast of Urumqi Glacier No.1)in different periods

Figure 7 Runoff observed at the UG1 hydrometeorological station from 1959 to 2018

4.4.2 Factors influencing streamflow

In the catchment of the UG1 hydrometeorological station, glacier runoff (the runoff formed from glacier surface) accounts for 70% of the total. The temporal variation is inversely correlated with the glacier mass balance of UG1, which is closely related to climate warming (Liet al., 2010). The escalating runoff pri‐marily originates from the increased meltwater pro‐duction as a result of the glacier's mass loss.A previ‐ous study indicates that runoff depends on both tem‐perature and the timing and magnitude of precipita‐tion.It rapidly rises to a high level after 1992,coincid‐ing with the double-stepped increases in temperature and precipitation. The surge in precipitation increased the total amount of water entering the catchment,while the rise in temperature aggravated the melting of snow and ice in the glacierized area.

The glacier area reduction is also an important fac‐tor influencing the runoff because the glacier shrink‐age leads to decreased glacier runoff. According to Huss and Hock (2018), as glaciers recede, water is re‐leased from long-term glacial storage (mass balance).The annual runoff volume typically increases until a maximum is reached, often referred to as "peak wa‐ter",beyond which glacier runoff declines because the reduced glacier area can no longer support rising melt‐water volumes. In fact, UG1 contracted 21%, from 1.95 km2in 1962 to 1.54 km2in 2017. If the negative effect of area reduction on glacier runoff was enough to offset the positive impact of rising temperature on glacier runoff, the peak glacial runoff (tipping point)had already occurred. For UG1, it appeared that the peak runoff happened from 1997 through 2007 (Jiaet al.,2020).

5 Summaries

The glacier zone boundaries of Urumqi Glacier No. 1 moved up over the past 60 years, accompanied by a shrinking accumulation area. The recrystalliza‐tion-infiltration zone observed in the 1960s turned in‐to the infiltration zone in the 1980s. The features of snow pack stratigraphy on the glacier, including depth, structures, and composition, were significantly altered by climate warming. Both the annual and cu‐mulative mass balances show accelerated mass loss in UG1 since 1960, which is consistent with the findings of mountain glaciers worldwide and was attributed to three mechanisms. Similar to terminus shrinkage,UG1 has been contracting at an accelerated rate since 1962, resulting in a loss of 0.37 km2or 19.3% from 1962 to 2018, or 0.007 km2/a. The changes in glacier terminus and area, related glacier mass balance and dynamics processes are the glacier's response to many climate-related events occurring in as short a period as one year or over centuries. The runoff measured at the UG1 hydrometeorological station demonstrates a significant increase from 1959 to 2018 with sizeable interannual fluctuation.The rising runoff largely origi‐nates from the increased meltwater production due to the mass loss of the glacier, which depends on both temperature and the timing and magnitude of precipi‐tation. The reduction in the glacier area most likely has led to a decreased runoff. At present, the runoff stays at a high level with large interannual variability.The mechanisms of changes in Urumqi Glacier No. 1 may also apply to most continental mountain glaciers in the world.

Acknowledgments:

This research was funded by the National Natural Sci‐ence Foundation of China (Grant No. 41761134093),the Second Tibetan Plateau Scientific Expedition and Research (Grant No. 2019QZKK0201), the Strategic Priority Research Program of the Chinese Academy of Sciences (Class A) (Grant Nos. XDA20060201 and XDA20020102), and the State Key Laboratory of Cryospheric Sciences Open Research Fund (Grant No.SKLCS-ZZ-2020).