Research on the LHG Glacier began in 1958 and the nearby glacier station was one of the earliest adopted for glaciology research in China. During 1958–1962, researchers monitored local meteorology and hydrology, and performed glacier mapping and MB, and ice temperature surveys. Unfortunately, the station was closed in 1962 because of a glacial flood. Only in the 1970s–1980s were some short-term observations performed in relation to studies on glacier change and its influence on the water resource. The Qilian Shan Station of Glaciology and Ecologic Environment was rebuilt in 2005 to focus on research in the fields of glaciology (Chen et al., 2018; Sun et al., 2014, 2012), ecology, and the atmospheric environment (Dong et al., 2019, 2018a, b ) of this remote area. This study used measurements of the glacier and the atmosphere to run the melt model.
During 1976, the ground stereo photogrammetry and mass balance measurements were carried out by previous glaciologists (Sun and Xie, 1981), the glacier-wide mass balance was 330 mm w.e. during the year.
Since 2010, in-situ MB measurements on the LHG Glacier have been determined on a monthly basis during May–September. Approximately 40 plastic stakes have been drilled into the ice to measure glacier ablation (Fig. 1). In the accumulation zone, snow pits were dug to measure the depth and density of the snow. Measured ice ablation was converted into water equivalent assuming an ice density of 900 kg·m-3.
Meteorological data from the LHG Glacier have been available since 2009 from an automatic weather station (AWS). The station is located at the elevation of 4 200 m a.s.l., 1.6 km from the terminus of the LHG Glacier (Fig. 1). Half-hourly air temperature (T) was measured using a Vaisala 41382 sensor (Campbell Scientific Inc.). Precipitation (P) was measured using an all-weather precipitation gauge (Geonor T-200B) without heating. The sensors were connected to a low-temperature resistant (-55 ℃) data logger (CR1000, Campbell Scientific Inc., USA). Detailed descriptions of instruments can be found in Sun et al. (2014). Figure 2 shows the monthly average air temperature and accumulated precipitation at the AWS during 2009–2015. The average air temperature was -5.9 ℃ and the yearly total amount of precipitation was 344 mm w.e. (water equivalent).
To derive T since 1960, data were collected from six adjacent weather stations (Fig. 1) within a horizontal distance of 250 km from the LHG Glacier. The data were observed by the national meteorological observation network of the China Meteorological Administration (Table 1). Stations of Dunhuang, Yumenzhen, and Jiuquan lie on the northern slopes of the Qilian Mountains, the Tuole is in the alpine area, and the Da Qaidam and Lenghu are located on the southern slopes. These stations were established in the 1950s and records of daily T were available. Strict quality control was applied by the National Meteorological Information Center (Wang, 2004).
Station name Latitude (N) Longitude (E) Elevation (m a.s.l.) Dunhuang 40°09′ 94°41′ 1 139 Yumenzhen 40°16′ 97°02′ 1 526 Jiuquan 39°46′ 98°29′ 1 477 Tuole 38°48′ 98°25′ 3 367 Da Qaidam 37°51′ 95°22′ 2 982 Lenghu 38°44′ 97°20′ 2 766
Table 1. List of the names and the latitude, longitude, elevation, and observation periods of stations used in this study
Data from the six stations were interpolated to the site of the LHG Glacier AWS. Having compared several interpolation techniques, (e.g., the inverse distance weighted, spline, and ordinary kriging methods and the multiple regression equation), Du et al. (2011) found the ordinary kriging method had the best performance compared with observed T at the AWS. We reconstructed T at the AWS using the ordinary kriging interpolation method, as suggested by Du et al. (2011). The R2 was 0.995 and the root mean square error (RMSE) was 0.51 between the reconstructed monthly T and that measured by the AWS during 2009–2015 (p < 0.001, n=72).
The westerlies dominate the climate of the middle and western Qilian Mountains, whereas the climate of the eastern Qilian Mountains is dominated by both the westerlies and southeastern monsoon (Morrill et al., 2003). We compared precipitation between the LHG Glacier and the surrounding national stations and found the best match with Tuole. Figure 3 shows the amount of annual P during 1961–2015 at Tuole.
2.1. Mass Balance Data
2.2. Meteorological Data
2.2.1. Air temperature
During the past six decades, T has increased continuously (Fig. 3), from -7.3 ℃ during 1961–1970 to -5.8 ℃ during 2001–2015. The LHG Glacier has experienced dramatic mass loss despite the gradual increase of P. The MB change was found to correlate well with the variations of T (R=0.78). To discuss the response of glacier MB to climate change, we divided the 55 years of the study period into six periods according to 10-year intervals. The MB was near to balance during 1961–1970, thus we took this period as the background reference period. Figure 7 shows the decadal variations of T, P, and MB relative to 1961–1970. Before 2000, T increased gently at the rate of 0.3 ℃· (10a)-1. Then, during 2000–2010, it increased sharply at the rate of 0.6 ℃· (10a)-1. Because of the function of P, the glacier did not experience consistent aggravated mass loss. During 1971–1980, T increased by 0.4 ℃, while P increased by 19 mm w.e. (9%); thus, the MB increased positively by 35 mm w.e.. During 1981–1990, the negative increase of MB, attributable to the 0.6 ℃ increase of T, was approximately offset by the 50 mm w.e. (25%) increase of P. In subsequent years, the negative increase of MB was substantial despite the increase of P. During 2000–2010, T increased by 1.5 ℃ and P increased by 88 mm w.e. (44%). Consequently, the MB increased negatively by -425 mm w.e.. After 2010, the increase of T remained at 1.5 ℃, while P increased by 125 mm w.e. (0.62); thus, the MB increased negatively by -312 mm w.e..
Figure 7. Decadal variations of air temperature, precipitation, and mass balance compared with the average during 1961–1970.
The increase of T led to reduction of the ratio of snow accumulation to total P. Table 2 shows the decadal glacier-wide ratio of snow accumulation to total precipitation, which varied from 0.98 during 1961–1970 to 0.85 during 2001–2010. Generally, the MB of maritime glaciers is less sensitive than that of continental glaciers to variation of P. One reason for this is the smaller ratio of snow accumulation to total P for maritime glacier-wide mass balance. glaciers compared with continental glaciers. For instance, the MB of the Parlung No. 94 Glacier which was a maritime glacier on the southeastern TP was found approximately two to three times more sensitive to a change in T of 1 ℃ than to 30% variation in P (Yang et al., 2013). However, for the LHG Glacier, a 30% increase in total P during 2010–2011 could approximately offset the change in MB attributable to an increase in T of 1.5 ℃ (Chen et al., 2017). Zhu et al. (2017) compared MB sensitivities to T and P in relation to the Parlung No. 4 (southeastern TP), Zhadang (southern TP), and Muztag Ata No. 15 (eastern Pamir) glaciers. Their results showed higher MB sensitivity to T by the Parlung No. 4 and the Zhadang glaciers, and higher MB sensitivity to P by the Muztag Ata No. 15 Glacier. It was considered that the most important factor determining the different sensitivities of glacier MB to change in T was the difference in the ratio of snowfall to total P.
Period Ratio 1961–1970 0.98 1971–1980 0.96 1981–1990 0.94 1991–2000 0.90 2001–2010 0.85 2011–2015 0.86
Table 2. Decadal average glacier-wide ratio of snow accumulation to total precipitation
LAPs in snow and ice could accelerate glacier melt. Based on sampling of snow and ice in combination with the use of an energy and mass balance model, Li et al. (2016) indicated that BC could account for 37% of summer melt on the LHG Glacier. Sun et al. (2017) found that the existence of LAPs in surface ice could cause an increase in net shortwave radiation of 7.1–16.0 W·m-2 in the ablation zone during June–September, which could result in glacier melt of 1 101–2 663 mm w.e.. Research on lake sediment cores (Han et al., 2015) and ice cores (Wang et al., 2015) has indicated that the amount of BC in northern China has more than doubled since the 1980s (Fig. 8). The DDF is a mathematical expression that reflects the conditions of the glacial surface and of the atmosphere, and BC is one factor that influences the glacial surface condition. Considering such a substantial increase of BC concentration in the snow and ice since the 1980s, it is important to determine whether the MB before 1980s was negatively overestimated when the DDF was calibrated after the 1980s, or whether the MB after the 1980s was negatively underestimated when the DDF was calibrated by measurements before the 1980s.
Figure 8. (a) Reconstructed mass balance of Laohugou Glacier No. 12, and measured mass balance of (b) the Qiyi Glacier and (c) the Urumqi Glacier No. 1; (d) black carbon concentration from an ice core located in eastern Pamir (Wang et al., 2015); (e) elemental carbon concentration from sediment core of Lake Qinghai. Straight lines show the average value before and after 1983.
Figure 8 shows a comparison between simulated MB on the LHG Glacier and measured MB on the Qiyi Glacier and the Urumqi Glacier No. 1. The MBs of the LHG and Qiyi glaciers correspond very well because the two glaciers are separated by only a short distance and they have similar climatic setting. The three glaciers all have experienced intense mass loss since the 1980s, but the magnitude of the losses has varied. The straight lines in Fig. 8 represent the average MBs before and after 1983. It can be seen that the difference in MB of the LHG Glacier before and after 1983 was much less than that of either the Qiyi Glacier or the Urumqi Glacier No. 1. It is demonstrably improper to attribute it to increased BC after the 1980s because the responses of glacier MB to climate change which are influenced by many factors such as the ratio of snowfall to total P, differences in melt energy, and seasonal distribution of P, they varied between the different glaciers (Zhu et al., 2017).
We search glaciers with long-term field MB observations and corresponding simulations. Table 3 presents information regarding three such glaciers. The Golubina glacier in western Tianshan is included in this glacier list in addition to the Qiyi Glacier and the Urumqi Glacier No. 1. Figure 9 shows since the beginning of the 1980s, the TP has experienced overall warming and moistening of the surface air, solar dimming, wind stilling, and more frequent occurrence of deep cloud (Yang et al., 2014). Chen et al. (2018) indicated that increased cloud could cause a decrease of net radiation direct to the glacier surface. Although higher humidity and lower wind speed could depress the energy output of the latent heat flux from the surface, latent heat is comparatively insignificant relative to radiation in relation to glacier melting (Sun et al., 2014, 2012). Therefore, we could rule out the possibility of other meteorological variables leading to the smaller difference.
Table 3. Information of glaciers with long-term field mass balance observations and sources of related mass balance simulations
Figure 9. Comparisons between simulated and measured mass balance of (a) the Qiyi Glacier, (b) the Urumqi Glacier No. 1, and (c) the Golubina Glacier. Red straight lines show measured average mass balance before and after 1983. Blue straight lines show simulated average mass balance before and after 1983.
To quantify intensification of glacier melt caused by the increase of BC after 1980s, we simulated MB of the Urumqi Glacier No. 1 (Fig. 10). The inputs of monthly air temperature and precipitation were from the Daxigou meteorological station with a distance of 2.5 km from the terminus of the glacier, and the data of measured MB was from Dong et al. (2013). The DDF was calibrated by measured MB between 1961–1983 with a constant value of 6.2 mm·d-1·℃-1.
Figure 10. Reconstructed and measured mass balance of Urumqi Glacier No. 1. Red straight lines show measured average mass balance before and after 1983. Blue straight lines show simulated average mass balance before and after 1983.
The total RMSE was 256 mm w.e. between simulation and measurement, and the simulated MB well matched the measured MB during 1961–1983. The mean measured MB was more negative by 220 mm w.e. than simulated MB during 1984–2015, which was equivalent to 26% of glacier ablation (mass balance+snow accumulation).
The surface albedo of TP glaciers derived from remote sensing imagery has exhibited a decreasing trend during 2001–2011 (Qu et al., 2014; Ming et al., 2012), which could partly be explained by the increase of LAPs (Qu et al., 2014; Ming et al., 2012). Cui et al.(2013, 2010) reported an increasing trend of DDF after the 1980s and they attributed it to surface darkening. Given such a dramatic increase of BC since the 1980s and the well-known influence on glacier melt, we assume that the warming has mainly contributed to the intense glacier melt since the 1980s, but it has been aggravated by the increase of BC in snow and ice.