Response of glacier area variation to climate change in the Kaidu-Kongque river basin, Southern Tianshan Mountains during the last 20 years

1.   Introduction

As an important part of the Earth’s cryosphere, glaciers are extremely sensitive to climate and environmental changes, and can be considered as an indicator of climate variation. Glaciers are also called "freshwater towers", which are very important solid water resources, especially in arid and semi-arid areas, where glacial meltwater is the main source of supply for rivers. Since the middle of the 20th century, climate warming has caused glacier retreat and melting in mid- and high-latitude regions of the world. This continuous variation increases the risk of glacier disasters such as ice avalanche, glacial lake outburst, flood, and mudslide, which have important influences on climate regulation, water resources, and ecosystems in the watershed. Therefore, studying the characteristics of regional glacier changes and establishing their response relationships to climate change can not only identify the distribution characteristics of regional glaciers, but also understand the retreat/advance rate of glaciers under the background of climate warming, which is useful for regional water resources management and geological disaster prevention. This research is of great significance to the evolution of natural ecological environment and the sustainable development of social economy.

The traditional methods of glacier change studying are mainly based on position monitoring and field survey, which have high data continuity and measurement accuracy, although there are problems such as low temporal and spatial coverage, high measurement costs, long monitoring periods, and extremely difficult conditions. With the continuous development of remote sensing technology, researchers began to use remote sensing to study the dynamic characteristics of glaciers, which solved the shortcomings of traditional methods. Landsat is an important data source for glacier change studying, images acquired from Landsat satellite series, which have the longest observation records of the earth surfaces (from 1972 to present), with the advantages of wide coverage and good data quality (Zhou YS et al., 2012; Ke LH et al., 2017; Kaushik S et al., 2020). The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data carried by the Earth Observatory Satellite (EOS-Terra) is also a commonly used data source, mainly used for dynamic monitoring of glacier variations (Li SS et al., 2013; Ji Q et al., 2018). In addition, high-resolution data of QuickBird, ALOS, IKONOS, SPOT-5, Sentinel, GF, and Synthetic Aperture Radar (SAR) data have also been gradually used to study glacier changes in recent years (Goldstein RM et al., 1993; Zhou JM et al., 2009; Narama C et al., 2010; Adina R and Williams MW, 2012; Dong P et al., 2013; Worni R et al., 2013; Yang BY et al., 2016; Du WB et al., 2020). The study of glacier changes through remote sensing depends on the identification and extraction of glacier range, and common methods such as visual interpretation, thresholded band ratios, supervision and non-supervision classification, Normalized Difference Snow Index (NDSI), and principal component analysis (Wang GF et al., 2010; Yan LL and Wang J, 2013; Burns P and Nolin A, 2014; Zhang W, 2016). The thresholded band ratio method is the most widely used method in the glacier range identification, for Landsat TM/ETM+, the combination channels of Band 3 (B3, red: 0.63–0.69 μm), Band 4 (B4, near-infrated: 0.77–0.90 μm), and Band 5 (B5, short-wave infrared: 1.55–1.75 μm) is selected for threshold segmentation of digital number (DN) or reflectance ratio to obtain the glacier range. In 1987, the B4/B5 ratio of Landsat TM images was used for the first time to obtain the glacier range (Hall DK et al., 1987). The later studies analyzed the glacier extraction results of different band ratios and found that the accuracy of B3/B5 or B4/B5 is higher than other band combinations, B3/B5 has more advantages than B4/B5 in the extraction effect of shadows and thin snow covered areas (Paul F, 2000; Paul F and Kääb A, 2005; Bolch T and Kamp U, 2006; Andreassen LM et al., 2008). In a study of glacier changes in western Canada from 1985 to 2005, it was found that the B3/B5 ratio of Landsat TM had a good extraction effect, with an error within 3% to 4% and was mainly caused by seasonal snow cover (Bolch T et al., 2010). Schmidt and Nüsser also used the B3/B5 ratio threshold method to analyze the changes of high altitude glaciers in the Trans-Himalayas of Ladakh from 1969 to 2016 (Schmidt S and Nüsser M, 2017).

In China, glacier research started relatively late. The scientific investigation of glaciers was started in the 1950s. From 1978 to 1999, the First Chinese Glacier Inventory was carried out through field surveys and manual measurement methods (Liu CH et al., 2000). According to Landsat TM/ETM+ and ASTER remote sensing image data from 2006 to 2011, the Second Chinese Glacier Inventory completed 85.5% of the total area of glaciers in China using thresholded band ratio and visual interpretation methods (Liu SY et al., 2015). Ji Q et al. (2018) used the ratio threshold method and combined B3/B5 of Landsat TM/ETM+ and B4/B6 of Landsat OLI to extract the glacier range of eastern Nyainqêntanglha Range in Tibet Plateau, and analyzed the response of glacier changes to climate variation from 1999 to 2015. Ou JB et al. (2020) selected B3/B5 ratio threshold method to extract the multi-period glacier ranges of Que’er Mountins in southeastern Tibet Plateau from 1987 to 2016, and analyzed its change process and characteristics. Guan WJ (2020) also used Landsat images and combined with ratio threshold method to extract the glacier boundaries of every 5 years in western Kunlun Mountains from 1972 to 2019, and then calculated the glacier areas. The above studies indicate that the extraction of glacier ranges based on series remote sensing images of Landsat and combined with B3/B5 ratio threshold method has high efficiency, high precision, and less manual intervention, which is suitable for the study of glacier changes in a large regional scale.

The Tianshan Mountain range, which stretches across the central Asia-Inner Mongolia Plateau, is one of the mountain systems with the most glaciers in the world. Glacier meltwater in the mountains is also one of the important water sources in arid regions of central Asia. Except for the Urumqi Glacier No.1 in Tianshan Mountins, which has been continuously monitored for more than 50 years, the monitoring time periods of other glaciers such as Qingbingtan Glacier No.72 in Mt. Tuomuer Region of Tianshan Mountains, Haxilegen Glacier No.51 in Kuytun of Eastern Tianshan Mountains, and Hami Miaoergou Glaciers are relatively short (Li ZQ et al., 2018; Northwest Institute of Eco-Environment and Resources, CAS, 2018). Therefore, a large number of glacier studies in this region were conducted using remote sensing technology. Wang SJ et al. (2011) conducted a statistical analysis on glacier area changes in various basins of Tianshan Mountains from 1960 to 2010, and found that although there are different retreat rates of glaciers, the acceleration of glacier melting is obvious, and temperature and precipitation changes in dry and wet seasons have accelerated the retreat of glaciers. He Y et al. (2014) conducted a comparative analysis of glacier changes in eastern and western sections of north Tianshan Mountains from 1989 to 2011, and indicated that the glacier retreat area in eastern section is larger than that in western as well as retreat rates in all slope directions, this difference is due to more solid precipitation in western section, which can offset the melting of glaciers caused by the increase in summer temperature. Du WB (2014) extracted and analyzed the glacier information in eastern section of Tianshan Mountains in 1990, 2000, and 2010, studied the changes of glacier scale, area, length, orientation, and terminal elevation, and showed that the glacier retreat rate in 2000 to 2010 is twice than that in 1990 to 2000. Farinotti D et al. (2015) used satellite gravimetry, laser altimetry, and glaciological modelling and found that the total area and mass of Tianshan glaciers decreased by 18±6% and 27±15% respectively from 1961 to 2012, these values correspond to a total area loss of 2960±1030 km², and an average glacier mass-change rate of −5.4±2.8 Gt/year. Chen WH (2016) studied the glacier changes in Tuomuer area of western Tianshan Mountains from 1964 to 2011, and found decreasing in glacier number, area, and stability of spatial structure, which indicates a high retreat rate will remain in the future. Zhang QF et al. (2019) analyzed the glacial changes in Aksu basin in middle section of Tianshan Mountains from 1975 to 2016, and found that the total number, area, and volume of glaciers decreased by 202 (7.65%), 965.7 km² (25.88%), and 74.85–78.52 km³ (23.72%–24.3%), respectively, the main reason is the increase in temperature, and the rate of glacier retreat may be directly related to the increase in precipitation in mountainous areas. According to Zhao GN (2020), the Tianshan glaciers showed a continuous negative balance of melting from 2000 to 2018, and the retreat rate continued to increase from 2000 to 2010, while fluctuant decreased from 2010 to 2018. In general, the changes of Tianshan glaciers have been mainly retreat since the 1960s. However, due to the influence of geographical location and topographical condition, the glacier changes in various basins are quite different. In addition, there are also studies focused on the changes of the terminus of Tianshan glacier and debris-covered area (Li BL et al., 2004; Liu SY et al., 2006; Li SS, 2013; Li SS et al., 2013; Wang YY et al., 2014; Xu CH et al., 2016; Wang PY et al., 2017). However, the study on the watershed-scale glacier changes of southern Tianshan Mountains was relatively scarce in the past 20 years.

Kaidu-kongque River (Kaikong River for short) watershed is located in the southern Tianshan Mountains, which is one of the main sources of Tarim River watershed. The utilization of water resources in Kaikong River watershed is vital to the ecological protection and economic development of the main stream of Tarim River. The climate-driven glacier changes in the upper reaches of Kaikong River have a direct impact on river runoff, water supply, and agricultural irrigation. Once the upstream water decreases, it will directly affect the industrial and agricultural production and ecological environment protection in the downstream oasis. In recent years, under the influence of human activities such as climate change and rapid expansion of agriculture, the various elements of water cycle in the middle and lower reaches have drastically changed (Zhang JF et al., 2018), which has caused a series of water and ecological environmental problems such (Wang R et al., 2003; Mamt A et al., 2017), river dry up (He ZG, 2019), decline of groundwater level (Wubuli W et al., 2017; Wang XY et al., 2019; Zhang J et al., 2021), water salinization (Pan XZ, 2016), a large-scale death of natural vegetation mainly composed of Populus euphratica (Liu YQ, 2017), and the decline of desert riparian ecosystems (Liu JZ et al., 2018). Therefore, research on the response of glaciers to climate change is of great significance to the rational allocation of water resources and ecological environmental protection in Kaikong River basin. Based on Landsat remote sensing images, the objective of this study was to extract the glacier range of Kaikong River basin from 2000 to 2020, and discuss the glacier area variations and response mechanism to climate changes combined with meteorological data, which can provide a theoretical and scientific basis for rational response to climate change and prediction of temporal-spatial changes of water resources in northwestern inland basin, and strongly support the green and sustainable development of western region.

2.   Outline of study area

Kaidu-Kongque river basin (KKB) is a sub-basin of the Tarim River basin in Xinjiang Uygur Autonomous Regions (Fig. 1). It lies on the northeastern edge of the Taklamakan Deseart, and the southern ridge of the Tianshan Mountains with 700–4800 m above sea level (a.s.l). The areal extent of the basin lies between 82°57′ E to 90°39′ E longitude and 40°25′ N to 43°21′ N latitude covering an area about 9.16×10⁴ km². The study area is located in the middle Center Asia-Mongolian Plateau and it falls under the temperate continental arid zone. Mean average precipitation and evaporation are 62–66 mm and 1983–2010 mm, respectively.

Figure  1.  Location map of Kaidu-Kongque river basin.

DownLoad: Full-Size Img PowerPoint

The KKB is composed of Kaidu, Bosten and Kongque river basins. The main water sources for Kaidu river originate from ice and snow meltwater within the central Tianshan mountains and precipitation. The mainstream of Kaidu river is 1321 m, the annual average runoff is 3.36×10⁹ m³/a (Su R, 2019). Bosten Lake is the largest freshwater lake in inland China. It is the tail lake of Kaidu river and the source of Kongque river. The changes in lake level and volume show a slight decrease in recent years. Reduced runoff of Kaidu river is the dominant contributor to decreased lake level and volume (Zhang T et al., 2015; Mamt A et al., 2017; Lin CQ et al., 2020). The tail lake of Kongque river used to be Lop Nor. The lower reaches of Kongque river has been cut off, and the annual average runoff is 1.27×10⁹ m³/a (Zhang J et al., 2021).

Glaciers are mainly distributed in the northwestern and northern mid-high mountains of KKB. As reported in the Second Chinese Glacier Inventory (Liu SY et al., 2015), there were 464 modern glaciers in the study area, covering a total area of approximately 400.56 km², with elevation of 3440–4800 m a.s.l. Most of them were less than 1 km² (Table 1).

Table  1.  The statistics of glaciers in Kaidu-Kongque river basin.

Glacier size/km2	No. of glaciers	Glacier area/km2	Ratio of area/%
$\geqslant$ 10	1	14.9	3.73
$\geqslant$ 5–10	8	58.09	14.54
$\geqslant$ 2–5	22	69.29	17.34
$\geqslant$ 1–2	54	84.62	21.18
< 1	379	172.66	43.21

 | Show Table

DownLoad: CSV

3.   Materials and methods

3.1   Data

The suitability of satellite imagery for glacier mapping depends on cloud cover, the date of acquisition, the presence of seasonal snow, and illumination geometry at the time of acquisition (solar elevation and azimuth). Landsat data has irreplaceable advantages for long-term glaciers observation, and Landsat 7 is effective in glacier mapping due to its Enhanced Thematic Mapper plus (ETM+) sensor. And it has a short-wave infrared band (1.55–1.75 μm) that is extremely sensitive to ice and snow, so Landsat 7 images are widely used in glacier researches (Kaushik S et al., 2020; Kumar V et al., 2021). To ensure the accuracy of glacial boundary extraction, the following criteria were applied to imagery selection. First, the cloud coverage in an image had to be < 10%. Second, images acquired the ablation period of glaciers (May to October), when glacier areas were believed to be near to or at their maximal extent, were set as optimal choices to minimize the impact produced by seasonal snow changes (Ke LH et al., 2017; Di BG, 2019). Based on the above criteria, 25 Landsat ETM+ images with a spatial resolution of 30 m were downloaded from the United States Geological Survey (USGS, http://earthexplorer.usgs.gov), and used to map glacier extents in 2000, 2005, 2010, 2015 and 2020 (Table 2).

Table  2.  The information of Landsat 7 images utilized in Kaidu-Kongque river basin.

Path/Row	2000	2005	2010	2015	2020
	Date and cloud cover (mm/dd; %)
143/30	9/17; 3	8/14; 1	6/9; 1	8/26; 1	8/7; 2
144/30	7/31; 3	9/22; 6	8/3; 3	7/16; 2	8/30; 2
144/31	8/7; 1	9/22; 5	9/4; 0	7/16; 0	8/30; 8
145/30	5/10; 9	5/8; 12	9/11; 8	10/11; 13	8/5; 12
145/31	5/10; 1	5/8; 2	8/10; 2	5/20; 2	8/5; 7

 | Show Table

DownLoad: CSV

Other datasets used are: (1) Meteorological data from the China Meteorological Administration; (2) The second glacial catalogue data set of China (v1.0) (Liu SY et al., 2015). The air temperature and precipitation data used in this study are monthly data from the Bayanbulak Weather Station located in glacier accumulation area. Datasets of gridded daily temperature/precipitation in China (Version 2.0) were used to verify the meterological data accuracy (Fig. 2) Due to the temporal hysteretic response of glacier changes to climate change, glacier changes are 2–20 years later than climate change (Wang NL et al., 1992; Dyurgerov MB and Meier MF, 2000; Granshaw FD and Fountain AG, 2006; Wang X et al., 2008), large glaciers (length greater than 5 km) in the Northern Hemisphere have a lagging response for about 8–9 years and smaller glaciers for 2–3 years (Ding YJ et al., 1995). Most glaciers in this study area are shorter than 5 km in length. Therefore, 5-year is chosen to be a hysteretic response period, from 1990 to 2020. Glacier datasets were used to extract glacial attributions. Based on the Second Chinese Glacier Inventory Data (SCGI), a buffer distance of 10 km from the nearest glacier termini was adopted to determine the extent of glacier in this study. Glacier further than 10 km from the nearest glacier termini were excluded due to the error caused by seasonal snow.

Figure  2.  Location map of meteorological station and grided data in Kaidu-Kongque river basin.

DownLoad: Full-Size Img PowerPoint

3.2   Methods

3.2.1   Gap-filled processing

Landsat 7 (ETM+) was launched in California on April 15, 1999. Since June 2003, the sensor has acquired and delivered data with data gaps caused by the Scan Line Corrector (SLC) failure. When the Level-1 data are processed, the duplicated areas are removed, leaving data gaps. Although these scenes only have 78% of their pixels remaining after the duplicated areas are removed, these data are still some of the most geometrically and radiometrically accurate of all civilian satellite data in the world (USGS et al., 2003). Landsat ETM+ images were employed due to the continuity of data during the study period. The authors applied and validated an effective gap-fill algorithm developed by the USGS EDC (2004). This algorithm operates under the assumption that the same-class neighboring pixels around the unscanned pixels have similar spectral characteristics, and that these neighboring and unscanned pixels share patterns of spectral differences between dates. This algorithm has been widely applied in glacier researches because of its better effect and high accuracy (Qian LX et al., 2012; Du WB, 2014; Mohammdy M et al., 2014).

3.2.2   Pre-processing

Landsat ETM+ L1T products downloaded from the United States Geological Survey (USGS) Earth Resources Observation and Science (EROS) Center are already orthorectified and projected. This study converted Landsat ETM+ digital numbers (DN) to top-of-atmosphere (TOA) radiance using pre-processing tools in ENVI v.5.1. Atmospheric correction was performed to convert top-of atmosphere radiance to surface reflectance. For this, this study used the Radiometric Calibration and Flaash atmospheric correction modules in ENVI. The Mosaicking and Region of Interest modules were used for images stitching and clipping respectively due to the study area covering 5 scenes with different paths and rows.

3.2.3   Glacier boundary delineation

Automated water classification with high efficiency is necessary for large-scale glaciers mapping. Thresholding of ratio images is a robust and time effective approach and also enables identification of snow and ice in shadow (Bolch T et al., 2010). The B3/B5 was employed in this study in that it works better in shadows and with thin debris-cover (Han N, 2019; Ou JB et al., 2020; Ji Q et al., 2020). An empirical threshold of 2.3 (based on examination of B3/B5 histogram around the study glaciers and tests on different thresholds) was chosen to segment the B3/B5 images into glacier area and non-glacier area. Quality assurance procedures includes three steps: (1) Based on the expended buffer zone outlined by the SCGI data, the margins of glaciers were visually identified and modified with the aid of multi-temporal high-resolution images from Google Earth and Landsat images, (2) glaciers smaller than 0.02 km² were removed because of seasonal snow or cloud cover, topographic shadows.

3.2.4   Glacier area analysis method

The Annual Percentage of Area Changes (APAC) has been widely used to assess changes in glaciers over time (Wang SJ et al., 2011; Che YJ et al., 2018), given by Eq. (1):

where APAC is the annual percentage of area changes ($\%$/a), S₀ and ∆S are the initial area and change area of glaciers during the study period (km²) respectively, and ∆t is the time span of the study period.

3.2.5   Glacier-climate relationship

Linear-regression analysis is available to find variation tendencies between glacier area, temperature and precipitation. In addition, finding out the correlations among these data through Pearson’s correlation analysis identified the response of glacier fluctuation to climate change. A correlation coefficient (r) higher than 0.7 implies a strong correlation between two parameters, whereas 0.5 < r < 0.7 represents a moderate correlation at a significance level p < 0.05.

4.   Results

4.1   The change of glacier area

The variation of glacier area during the period of 2000–2020 is shown in Table 3. It shows that the total glacier area has shrunk from 421.62 km² to 398.21 km², lost 23.41 km² between 2000 and 2020, at a linear rate of 0.26%·per decade. The glacier retreat rate were not constant. According to APAC, the authors divided the change of glacier area into three periods (Period 1: 2000–2005; Period 2: 2005–2015; Period 3: 2015–2020). Period 1: Glacier area decreased from 421.62 km² in 2000 to 415.14 km² (−0.31%/a) until 2005. This indicates that glacier shrinkage is particularly obvious in Period 1. Period 2: APAC was estimated to be −0.25%/a between 2005 and 2010, and 0.21%/a between 2010 and 2015. The average glacier area is 410±5 km². The authors note that a slight decrease in the rate of deglaciation from 2005 to 2015. Period 3: Retreat rate, equivalent to an absolute value of APAC, was estimated to be 0.37%/a. This is higher than any previous rate of area loss between any two periods. The glaciers lost 1.86% of the total area during 2015–2020. The glacier shrinkage accelerated in recent five years.

Table  3.  Glacier area and area changes between 2000 and 2020.

Year	Glacier area/km2	Period	Area change/km2	Rate of area change/%	APAC/%/a
2000	421.62
2005	415.14	2000–2005	−6.48	−1.54	−0.31
2010	409.97	2005–2010	−5.17	−1.25	−0.25
2015	405.76	2010–2015	−4.21	−1.03	−0.21
2020	398.21	2015–2020	−7.55	−1.96	−0.37
		2000–2020	−23.41	−5.55	−0.26

 | Show Table

DownLoad: CSV

There are potential sources of error associated with our estimates of glacier area due to topographic correction, positional error, choice of threshold, debris-covered glacier, and expended buffer error. Compared with the SCGI, the results of this study show that the uncertainty of the glacier area estimated is about 10 km², and the error is about 2.5%.

4.2   Characteristics of temperature and precipitation

Glacial retreat is known to be a result of air temperature and precipitation. Rising temperatures and decreasing precipitation enhance the melt rates of snow and ice, and thereby facilitates the further reduction of the glacier mass (Oerlemans J et al., 2009). When the air temperature increases more than 0.5°C, the change of glacier mainly depends on temperature (Gao XQ et al., 2000).

4.2.1   Air temperature

Long-term meteorological data (1990–2020) was analyzed comprising of mean annual temperature (T) and 5-year moving average temperature (5-MAT). As shown in Fig. 3, the increasing trend of T and 5-MAT are not statistically significant (R²<0.2, p>0.05). T and 5-MAT increased at linear rates of 0.2°C per decade (0.02°C/a) and 0.3°C per decade (0.03°C/a), higher than the global warming trend of 0.15–0.20°C per decade (IPCC, 2014).

Figure  3.  Variation of mean annual temperature (T) and 5-year moving average temperature (5-MAT) between 1990 and 2020.

DownLoad: Full-Size Img PowerPoint

The inter-annual mean annual temperature has fluctuated noticeably during the study period, with the greatest increase of 2.57°C. In terms of inter-decadal variations, the time-series data reveal significant increasing trends of 5-MAT during the former 15 years (1995–2010) and the latter five years (2015–2020), compared to the negative trend between 2010 and 2015. In particular, the 5-MAT during 1995–2000 and 2010–2015 have a comparably larger increase with the change of 0.22°C/a, and a larger decrease with the rate of 0.23°C/a, respectively.

4.2.2   Precipitation

As for precipitation (Fig.4), the annual precipitation (P) and 5-year moving average precipitation (5-MAP) have a non-significantly increasing trend with rates of 2.07 and 1.54 mm/a (R²<0.3, p>0.05). The 5-MAP shows significant trend at 95% confidence interval and R²>0.5 between 2005 and 2020. The 5-MAP decreased around an average level of 6 mm/a during 2000–2005 and 2010–2015, except the other periods. The increase of 5-MAP is the most remarkable over 2015–2020, at the rate of 12.59 mm/a (R²=0.91, p<0.05).

Figure  4.  Variation of mean annual precipitation (P) and 5-year moving average precipitation (5-MAP) between 1990 and 2020.

DownLoad: Full-Size Img PowerPoint

4.3   Relationship between glacier area variation and climate change

Glaciers of KKB have retreated obviously within the study area. Glacier retreat and advance is sensitive to atmosphere. The interaction of glaciers and climate has been assessed through linear regression analysis of glacial and meteorological data (Figs. 5, 6). Glacier area followed a slightly negative variation to that of 5-MAP during the study period (Fig. 5). The variation trends between glacier area and mean annual temperature or annual precipitation (Fig. 6) are non-linear. In order to illustrate the relationship between glacier and climate, glacier area variation should be analyzed togehter with both air temperature and precipitation.

Figure  5.  Variation of mean annual temperature (T) and 5-year moving average temperature (5-MAT) in relation to glacier area changes (1995–2020).

DownLoad: Full-Size Img PowerPoint

Figure  6.  Variation of mean annual precipitation (P) and 5-year moving average precipitation (5-MAP) in relation to glacier area changes (1995–2020).

DownLoad: Full-Size Img PowerPoint

In principle, glaciers has preserved the hysteretic response of temperature and/or precipitation changes (Kumar V et al., 2021). The authors analyzed the variations of glacial area and 5-year moving average temperature and precipitation to understand the glacial sensitive to climate system. The noticeable variability in glacier area loss over the Period 1 (2000–2005) may be because of significant increasing trends of 5-MAT and 5-MAP during the former 9 years (1995–2003). The time-series climate data reveal an almost flat increasing trend in volatility of 5-MAT from 2003 to 2010, compared to a significant decreasing trend of 5-MAP over 2003–2007 and a sharp increasing trend of 5-MAP during 2007 the period of 2007–2010/2011. 5-MAT followed a similar variation to that of 5-MAP from 2010 to 2015. Trend of glacier area fluctuated slightly during the Period (2005–2015) due to variable tends of 5-MAT and 5-MAP between 2003 and 2015. The accelerated rate of deglaciation in Period 3 (2005–2015) due to the gradual increases in 5-MAT and 5-MAP during the former periods.

4.4   Links between glacier area variation and seasonal variation

The seasonal fluctuation is an important indicator of glacier melting, and the glacier ablation generally occurs when the air temperature is higher than 0°C (Kaushik S et al., 2020). Therefore, the possible changes in glacier area forced by seasonal variables were further accessed for glacier-climate relationship. Each year was divided into summer and winter according to whether the air temperature is above or below 0°C (Zhou ZH et al., 2017), representing the seasonal response of mean climatic variations.

4.4.1   Impact of summer

As shown in Fig. 7, the average summer temperature (ST) reaches up to 10.3°C in 2003, and ST of every five years from 1995 to 2020 is 7.48°C, 7.5°C, 7.24°C, 8.38°C and 7.84°C, respectively. The average summer temperature from 2015 to 2020 is significant higher than that of the former 20 years (1995–2015), and showing a sharper trend of ST during the same period (2015–2020). The mean summer temperature is increasing by 0.24°C over 2010–2015, higher than that of the global warming average of 0.15–0.20°C per decade (IPCC, 2014). Negative trend of ST is observed more significantly from 2015 to 2020 than other periods, and the mean ST decreased by 0.17°C.

Figure  7.  Variation of mean temperature (T) and 5-year moving average temperature (5-MAT) in relation to glacier area changes in summer (1995–2020).

DownLoad: Full-Size Img PowerPoint

Summer precipitation (SP) fluctuates greatly, with a large inter-annual variation range and a maximum difference of 160 mm (Fig. 8). The summer precipitation (249.52 mm per five years) of 1995–2000 is the lowest in the recent two decades. The increasing trends of SP are significant during 2005–2010 and 2015–2020, with the highest mean SP (about 270 mm) of these two periods during 25 years (1995–2020).

Figure  8.  Variation of mean precipitation (P) and 5-years moving average precipitation (5-MAP) in relation to glacier area changes in summer (1995–2020).

DownLoad: Full-Size Img PowerPoint

The area-season relationship was analyzed using the indexes of glacier area, 5-MAT and 5-MAT. The rapid loss of glacier area during the Period 1 might be mainly attributed to increasing 5-MAT and decreasing 5-MAP in summer between 1995 and 2005. The most probable explanation for passive deglaciation of Period 2 is the ongoing climate change in summer during the period of 2005–2015. Summer temperature (5-MAT) decreased and precipitation increased with small amplitudes over 2005–2015. Furthermore, the increasing trend of summer 5-MAT and the decreasing trend of summer 5-MAP are presented in Figs. 7 and 8. The average of summer 5-MAT in the recent decade (2010–2020) is more remarkable than that in the former 15 years (1995–2010). Significant trend of rising summer temperature coupled with precipitation change caused glacier melting, which led to loss of glacier area.

4.4.2   Impact of winter

The non-significant trends of winter mean temperature and 5-year moving average temperature from 1995 to 2010 at a low confidence level (R²<0.3, p<0.05) are shown in Fig. 9. During the period between 2010 and 2020 multi-year mean temperature in winter was higher than that the former 15 years (1995–2010), and multi-year mean precipitation presented increased trend (Fig. 10). According to the changes of winter temperature, precipitation and glacier area of different periods, it can be found that declining temperature and increasing precipitation between 1995 and 2000. There were an increasing trend of winter temperature and decreasing trends of winter precipitation and glacier area for the period of 2000 and 2005. The time-series climate data revealed decreasing trend of winter temperature and precipitation from 2005 to 2010, and a significant decreasing trend of winter temperature with an increasing trend of winter precipitation over 2010–2015, as well as a slightly decreasing trend of glacier area between 2005 and 2015. In addition, a significant increasing trend of winter temperature, a decreasing trend of winter precipitation and accelerated rage of the contemporary deglaciation were shown in Figs. 9 and 10. In general, increasing precipitation and decreasing temperature are beneficial for glacier storage in cold environment. Nevertheless, the present study shows the effect of winter temperature and precipitation on the loss of glacier area seem to be slight or not significant.

Figure  9.  Variation of mean temperature (T) and 5-year moving average temperature (5-MAT) in relation to glacier area changes in winter (1995–2020).

DownLoad: Full-Size Img PowerPoint

Figure  10.  Variation of mean precipitation (P) and 5-year moving average precipitation (5-MAP) in relation to glacier area changes in winter (1995–2020).

DownLoad: Full-Size Img PowerPoint

4.5   Response of glaciers to climate change

In principle, glacier shrinkage has preserved the response of temperature and/or precipitation changes (Kumar V et al., 2021). To interpret the relationship between the glacier ablation and climate change, the statistics on the correlation coefficients between glacier area and climatic variables were calculated and analyzed in this study (Fig. 11). Glacier area had the most significant relationship with summer 5-MAT and APAC, and non-significant correlation with other variables. The positive area-APCA relationship (p<0.05) indicates that APAC reflects the degree of glacier ablation, which was consistent with the actual situation. The negative correlation of summer 5-MAT with area (p<0.01) could be interpreted with increase in summer temperature diving deglaciation and the loss of glacier area. Further, there is a temporal hysteresis in the response of glacier shrinkage to climate change. APAC presents negative significant correlation with 5-MAT and summer 5-MAT, and positive correlation with winter 5-MAT at a high confidence level (p<0.01), indicating the glacier ablation is close connected with the variation of air temperature. It can be seen that rising temperature, especially summer temperature, seems to be more dominate than the precipitation, which leads to negative mass balance and loss of glacier area.

Figure  11.  Positive correlations are displayed in blue and negative correlations in red color. Color intensity and the size of the circle are proportional to the correlation coefficients. In the right side of the correlogram, the legend color shows the correlation coefficients and the corresponding colors.

DownLoad: Full-Size Img PowerPoint

The response of glaciers to climate change involves a series of complex processes, including solar radiation, air temperature, precipitation, wind and cloud cover, etc., which can affect the mass and energy balance of the glacier surface. Changes in the mass balance of glaciers will lead to changes in volume and thickness, which will affect the length of ice tongue and glacier area through internal deformation and basement sliding. The study by Gao XQ et al. (2000) reported the longer the temporal scale or the larger spatial scale, the more significant response of glacier advance/retreat to temperature. Generally, precipitation only contributes to loss of glacier area at short temporal scales (e.g. less than a decade) and small spatial scales. Several studies have shown that a 25% increase in annual precipitation is typically needed to compensate for the mass loss due to a uniform 1°C warming (Braithwaite RJ and Zhang Y, 2000; Oerlemans J, 2001; Oerlemans J et al., 2005). Glaciers of Tianshan Mountains are of summer accumulation type, which receive more accumulation in summer than in winter. Because climate in Center Asia-Mongolian Plateau is influence by summer monsoon, most of the annual precipitation is concentrated in summer, while winter is cold but is a relatively dry season. Summer accumulation type glaciers are very sensitive to changes in air temperature, which are considered to be much vulnerable to global warming (Naito N, 2011). The accelerated rate of glacier ablation of Tianshan Mountains in the recent decades due to rising air temperature is also reported by several studies (Du WB, 2014; Brun F et al., 2017; Che YJ et al., 2018).

In general, the increasing trend of summer temperature impacts strongly on glacier retreat in the study area during the recent decades. However, precipitation have no significant impact on glacier ablation over recent decades. In light of the dependence of glaciers on air temperature, glaciers in the KKB region will keep experiencing severe retreat if the warming trend continues, which will lead to hydrological and ecological environmental problems such as reduction of downstream runoff and lakes and desert encroachment, as well as the risk of floods and mudslides caused by glacier ablation.

5.   Conclusions

This study aimed to quantify the glacier area changes in Kaidu-Kongque river basin, south slope of Tianshan Mountains, Central Asia-Mongolian Plateau, to assess its response to the ongoing climate change. Based on the observation of the present study, the authors present following conclusions.

(i) Mapping the variations in extents of the glaciers in this study area reveals that a continuous trend of deglaciation during the past two decades (2000–2020). The total loss of glacier area during the period is 23.41 km², at a linear rate of 0.26% per decade. The highest loss of glacier area was observed for the period of 2000–2005 (1.54%) and 2015–2020 (1.96%).

(ii) Analysis of metrological data showed that slightly increasing trends of air temperature and precipitation in recent two decades. Annual and 5-year moving average temperate varied in the range of 0.02°C/a to 0.03°C/a, and annual and 5-year moving precipitation increased by 2.07 mm/a and 1.54 mm/a, respectively.

(iii) In terms of the response of glacier area to climate change, glaciers in the study area has preserved the hysteretic response of climate changes. The increasing trend of air temperature made a big contribution to loss of glacier area, particularly the increasing trend of summer temperature impacted strongly on glacier retreat in the study area during the recent decades. In light of the dependence of glaciers on air temperature, glaciers in the Kaidu-Kongque river basin will keep experiencing severe retreat if the warming trend continues.

(iv) Glacier retreat will affect the amount and stability of downstream runoff under the continuous warm-wet trend in Northwest China. Monitoring of glaciers at regular interval is required for mitigation of glacial hazards, regulation of runoff, understand regional climate. Furthermore, it is significant to the evolution of natural ecological environment and the sustainable development of social economy in Northwest China.

CRediT authorship contribution statement

Lu-chen Wang and Li-he Yin conceived of the presented idea. Lu-chen Wang wrote the manuscript with support from Kun Yu, Jun Zhang and Li-he Yin. Kun Yu aided in interpreting the results. Liang Chang contributed to metrological data collection. Kun Yu and Tao Tang designed the Fig. 1. All authors discussed the results and contributed to the final manuscript.

Declaration of competing interest

The authors declare no conflict of interest.

Acknowledgment

This work was supported by the project of China Geology Survey (DD20190315), Innovation Capability Support Program of Shaanxi (2019TD-040), “Integration of Groundwater Resources Assessment Results in Key Areas of Northwest China” program, and Key Laboratory of Groundwater and Ecology in Arid and Semi-arid Areas of China Geological Survey.