1. Introduction

In a warming climate, snowlines retreat earlier in the season and to higher elevations (Oerlemans, Reference Oerlemans2001). As a result, more bare-ice is becoming exposed at the glacier surface and for longer periods of time (Box and others, Reference Box, Fettweis, Stroeve, Tedesco, Hall and Steffen2012, Di Mauro and Fugazza, Reference Di Mauro and Fugazza2022). This growing exposure has implications for the surface energy balance and melting rates (Hartl and others, Reference Hartl, Felbauer, Schwaizer and Fischer2020, Di Mauro and Fugazza, Reference Di Mauro and Fugazza2022, Gunnarsson and Gardarsson, Reference Gunnarsson and Gardarsson2023). Radiative transfers control the surface energy balance (Greuell and Smeets, Reference Greuell and Smeets2001, Klok and Oerlemans, Reference Klok and Oerlemans2002) and, among these transfers, short-wave radiation is usually the dominant energy source for extra-tropical glaciers (Braithwaite, Reference Braithwaite1995, Oerlemans, Reference Oerlemans2001, Marshall, Reference Marshall2014). The albedo of the exposed snow, firn and ice modulates the absorption of short-wave radiation, resulting in distinct albedo zones on glaciers. The variability of bare-ice albedo critically impacts the extent and amount of surface melt in ablation areas, where the net mass loss occurs (Hall and others, Reference Hall, Ormsby, Bindschadler and Siddalingaiah1987, Hartl and others, Reference Hartl, Felbauer, Schwaizer and Fischer2020). Conversely, bare-ice albedo is influenced by radiative transfers, as the net energy budget alters the ice surface properties, on which albedo relies. Such feedback mechanisms generate self-feeding melt cycles and explain the amplified response of glaciers to atmospheric changes (Johnson and Rupper, Reference Johnson and Rupper2020). Rising temperatures play a key role in these melt cycles, as higher air temperatures enhance the energy input to the surface energy budget of ablation areas by modulating long-wave radiation and sensible heat fluxes. On bare-ice, this contributes to warmer ice, larger grain sizes and meltwater presence on the ice—all of which decrease the bare-ice albedo of glaciers (Azzoni and others, Reference Azzoni, Senese, Zerboni, Maugeri, Smiraglia and Diolaiuti2016, Fugazza and others, Reference Fugazza, Senese, Azzoni, Maugeri and Diolaiuti2016, Tedstone and others, Reference Tedstone, Cook, Williamson, Hofer, McCutcheon, Irvine-Fynn, Gribbin and Tranter2020). Therefore, with the increased and prolonged exposure of bare ice to rising temperatures, the influence of bare-ice albedo variability on glacier surface energy balance and melting rates is expected to grow.

Multiple studies have investigated glacier-wide albedo (Brock and others, Reference Brock, Willis, Sharp and Arnold2000b, Jonsell and others, Reference Jonsell, Hock and Holmgren2003, Gascoin and others, Reference Gascoin, Gudmundsson, Adalgeirsdóttir, Pálsson, Schmidt, Berthier and Björnsson2017, Marshall and Miller, Reference Marshall and Miller2020, Gunnarsson and others, Reference Gunnarsson, Gardarsson, Pálsson, Jóhannesson and Sveinsson2021, Di Mauro and Fugazza, Reference Di Mauro and Fugazza2022). The temporal variability of glacier-wide albedo provides valuable information on the snow-free and snow-covered fractions of glaciers (Davaze and others, Reference Davaze, Rabatel, Arnaud, Sirguey, Six, Letreguilly and Dumont2018). Therefore, the annual minimum albedo averaged over the glacier can be used to derive the annual equilibrium line altitude (Brun and others, Reference Brun, Dumont, Wagnon, Berthier, Azam, Shea, Sirguey, Rabatel and Ramanathan2015, Shaw and others, Reference Shaw, Ulloa, Farías-Barahona, Fernandez, Lattus and McPhee2021). However, the temporal variability of glacier-wide albedo does not capture the variability of albedo of the bare-ice that remains exposed during ablation season. This variability significantly impacts glacier melt, as Naegeli and Huss Reference Naegeli and Huss(2017) found an average decrease of $-0.14~\mathrm{m~w.e.~a}^{-1}$ in the mass balance of 12 Swiss glaciers for every 0.1 reduction of bare-ice albedo. A similar sensitivity of mass balance to bare-ice albedo variability has been found for glaciers in the Chilean Andes (Barandun and others, Reference Barandun, Bravo, Grobety, Jenk, Fang, Naegeli, Rivera, Cisternas, Münster and Schwikowski2022). Yet, few studies have investigated this variability (Cutler and Munro, Reference Cutler and Munro1996, Jonsell and others, Reference Jonsell, Hock and Holmgren2003, Dumont and others, Reference Dumont, Gardelle, Sirguey, Guillot, Six, Rabatel and Arnaud2012, Naegeli and Huss, Reference Naegeli and Huss2017, Naegeli and others, Reference Naegeli, Huss and Hoelzle2019, Hartl and others, Reference Hartl, Felbauer, Schwaizer and Fischer2020, Tedstone and others, Reference Tedstone, Cook, Williamson, Hofer, McCutcheon, Irvine-Fynn, Gribbin and Tranter2020, Barandun and others, Reference Barandun, Bravo, Grobety, Jenk, Fang, Naegeli, Rivera, Cisternas, Münster and Schwikowski2022, Rossini and others, Reference Rossini, Garzonio, Panigada, Tagliabue, Bramati, Vezzoli, Cogliati, Colombo and Di Mauro2023) and bare-ice albedo is often assumed to be constant in space and time in mass balance modelling (Klok and Oerlemans, Reference Klok and Oerlemans2002, Machguth and others, Reference Machguth, Purves, Oerlemans, Hoelzle and Paul2008). A growing body of research is exploring the potential of remote sensing, in situ measurement and unmanned aerial system technology to monitor albedo and linked glacier surface properties (Naegeli and others, Reference Naegeli, Huss and Hoelzle2019, Hartl and others, Reference Hartl, Felbauer, Schwaizer and Fischer2020, Tedstone and others, Reference Tedstone, Cook, Williamson, Hofer, McCutcheon, Irvine-Fynn, Gribbin and Tranter2020, Barandun and others, Reference Barandun, Bravo, Grobety, Jenk, Fang, Naegeli, Rivera, Cisternas, Münster and Schwikowski2022, Gunnarsson and Gardarsson, Reference Gunnarsson and Gardarsson2023, Gunnarsson and others, Reference Gunnarsson, Gardarsson and Pálsson2023, Rossini and others, Reference Rossini, Garzonio, Panigada, Tagliabue, Bramati, Vezzoli, Cogliati, Colombo and Di Mauro2023). Mass balance models have been shown to improve when using the output of such monitoring efforts to replace constant albedo values with satellite-based spatially distributed albedo in Barandun and others Reference Barandun, Bravo, Grobety, Jenk, Fang, Naegeli, Rivera, Cisternas, Münster and Schwikowski(2022) and Naegeli and others Reference Naegeli, Damm, Huss, Schaepman and Hoelzle(2015). Still, to date, the spatio-temporal variability of bare-ice albedo has not been incorporated into such models.

A better understanding of the sub-seasonal and interannual spatio-temporal variability of bare-ice albedo is key to effectively integrate temporal and spatial albedo distributions into glaciological models. With this in mind, the present study aims to gain insights into the spatio-temporal variability of bare-ice albedo, along with its drivers and its influence on glacier melt. We focus on the bare-ice albedo of Abramov Glacier in Central Asia. The region has warmed by 0.37–0.43 $^\circ\text{C}$ per decade over the period 1979–2011 (Hu and others, Reference Hu, Zhang, Hu and Tian2014). The number of summer days (days with daily maximum temperature $ \gt 35^\circ\text{C}$) has increased by more than 10 days per year between 1999–2019 compared to 1950–79, while icing days (days with daily maximum temperature below $0^\circ\text{C}$) have decreased by 15 days per year (Fallah and others, Reference Fallah, Didovets, Rostami and Hamidi2024). Simultaneously, snow melted earlier in the season in the Amu Darya River Basin from 1985 to 2008 (Zhou and others, Reference Zhou, Aizen and Aizen2017). Snow depth has also decreased by 20% in Central Asia over the past 70 years, associated with a warming of $1.2^\circ\text{C}$ (Fallah and others, Reference Fallah, Didovets, Rostami and Hamidi2024). As a result of those changes, the bare ice of Central Asian glaciers is exposed to solar radiation, impurity deposition and rising temperatures for longer periods of time (Schmale and others, Reference Schmale, Flanner, Kang, Sprenger, Zhang, Guo, Li, Schwikowski and Farinotti2017, Zhang and others, Reference Zhang, Gao, Kang, Sprenger, Tao, Du, Yang, Wang and Meng2020). Data are scarce for glaciers of Central Asia, but Abramov benefits from extensive meteorological and mass balance records (Hoelzle and others, Reference Hoelzle, Azisov, Barandun, Huss, Farinotti, Gafurov, Hagg, Kenzhebaev, Kronenberg and Machguth2017). Here, we explore bare-ice albedo variability using a time series of albedo scenes over the ablation area acquired during melt season (July–September) from 1999 to 2022. We investigate (1) the sub-seasonal temporal and spatial variability of bare-ice albedo of Abramov, (2) the interannual variability of bare-ice albedo under the changing conditions for the period from 1999 to 2022, (3) the role of air temperature in modulating bare-ice albedo and (4) the influence of bare-ice albedo on variations of melting rates.

2. Study site

Abramov Glacier ( $39.50^\circ\,\text{N}, 71.55^\circ\,\text{E}$) is located in the Pamir-Alay range in Kyrgyzstan, Central Asia, and feeds the Koksu river, a tributary of the Amu Darya (Denzinger and others, Reference Denzinger, Machguth, Barandun, Berthier, Girod, Kronenberg, Usubaliev and Hoelzle2021). This is a north-facing valley glacier, with longitudinal supra-glacial medial moraines and a significant load of debris and impurities, such as light-absorbing particles and dust, on the lower tongue (Fig. 1). The glacier extends over $21.27\,\text{km}^2$ (Mattea and others, Reference Mattea, Machguth, Berthier and Hoelzle2023) between 3650 and 5000 m a.s.l. Mass balance measurements started already in the 1960s, but the collapse of the Soviet Union brought the measurement campaigns to an end in 1999 (Suslov and Krenke, Reference Suslov and Krenke1980, Glazyrin and others, Reference Glazyrin, Kamnyansky and Pertziger1993, Pertziger, Reference Pertziger1996, Hoelzle and others, Reference Hoelzle, Azisov, Barandun, Huss, Farinotti, Gafurov, Hagg, Kenzhebaev, Kronenberg and Machguth2017, Kronenberg and others, Reference Kronenberg, Machguth, Eichler, Schwikowski and Hoelzle2021). The monitoring was re-initiated in 2011 (Hoelzle and others, Reference Hoelzle, Azisov, Barandun, Huss, Farinotti, Gafurov, Hagg, Kenzhebaev, Kronenberg and Machguth2017). An automatic weather station (AWS) and an automatic terrestrial camera (Mobotix M25) were installed at 4100 m a.s.l. about 1.5 km from the glacier terminus (Fig. 1a) (Schöne and others, Reference Schöne, Zech, Unger-Shayesteh, Rudenko, Thoss, Wetzel, Gafurov, Illigner and Zubovich2013, Barandun and others, Reference Barandun, Huss, Usubaliev, Azisov, Berthier, Kääb, Bolch and Hoelzle2018). The mean air temperatures measured at the station from 2011 to 2022 are $5.78^\circ\text{C}$ in July, $4.89^\circ\text{C}$ in August and $1.89^\circ\text{C}$ in September.

Abramov has lost mass in recent decades. Kronenberg and others Reference Kronenberg, van Pelt, Machguth, Fiddes, Hoelzle and Pertziger(2022) calculated a mass balance of $-0.27~\mathrm{m~w.e.~a}^{-1}$ for the period 1968/69–2019/20. Concurrent with this mass loss, Kronenberg and others Reference Kronenberg, Machguth, Eichler, Schwikowski and Hoelzle(2021) estimated that net accumulation had increased simultaneously to ablation over the last 60 years. The glacier has experienced occasional surge behaviour, such as between 1972 and 1975 (Emeljyanov and others, Reference Emeljyanov, Nozdryuhin and Suslov1974, Suslov and Krenke, Reference Suslov and Krenke1980, Glazyrin and others, Reference Glazyrin, Kamnyansky and Pertziger1993) and possibly in 2018 (Mattea and others, Reference Mattea, Machguth, Berthier and Hoelzle2023, Reference Mattea, Berthier, Dehecq, Bolch, Bhattacharya, Ghuffar, Barandun and Hoelzle2024). Our study area is the lower third of the glacier, covering $5.1~\text{km}^2$ between 3626 and 4071 m a.s.l. In this area, bare ice is exposed for most of summer and remains unaffected by the variations in surface types in the transition region between ablation and accumulation area.

Figure 1. Right panel: Map showing the glaciated areas (dark blue) in Central Asia. Left panel: Overview of Abramov glacier, on a section of a Sentinel-2 scene (20 August 2019) overlaid with elevation data from Shuttle Radar Topography Mission (SRTM) imagery. The study area is outlined (light blue), using the Global Land Ice Measurements from Space (GLIMS) dataset, modified to reflect the 2020 glacier terminus as delineated by Enrico Mattea in Mattea and others Reference Mattea, Berthier, Dehecq, Bolch, Bhattacharya, Ghuffar, Barandun and Hoelzle(2024). The automatic weather station (AWS) location and the manually delineated supra-glacial medial moraines are also indicated.

3. Data

4. Methods

4.3. Analysis of interannual variability

We used non-parametric Mann–Kendall tests (Mann, Reference Mann1945, Kendall, Reference Kendall1975) to detect pixel trends in the summer, July, August and September albedo time series (dataset 1). We used three confidence levels (75%, 85% and 95%) to provide a comprehensive overview of changes in bare-ice albedo while maintaining robust results. We assessed the rate of albedo change using the Theil–Sen slope estimator (Theil, Reference Theil1950) for pixels with trends at 85% confidence level. In addition, we calculated the mean difference in bare-ice albedo for each pixel during July, August and September (dataset 1) of two sub-periods: 1999–2010 and 2011–22.

Figure 2. (a) Time series of Abramov’s average bare-ice albedo for all years with a minimum of three scenes between 1999 and 2022. (b) Spatial distribution of 1999–2022 average bare-ice albedo for July, August and September. (c) Spatial distribution of 1999–2022 average bare-ice albedo difference between July/August and August/September.

5. Results

5.1. The sub-seasonal cycle of bare-ice albedo

Average bare-ice albedo at Abramov decreases from July to August and increases from August to September (Fig. 2a). The magnitude of seasonal changes in bare-ice albedo depends on elevation and presence of debris and impurities on the ice (Fig. 2b, c). From July to August, both the variability and the values of bare-ice albedo decrease significantly ( $\leqslant -0.06$) at high elevation (between 3950 and 4071 m a.s.l.) on the tongue. However, bare-ice albedo remains constantly low at low elevation (between 3650 and 3850 m a.s.l.) between July and August (Fig. 2c). From August to September, the albedo of debris-free ice increases more significantly ( $\geqslant +0.06$) than that of the ice in the area of the medial moraines (between +0.01 and +0.05; Fig. 2c).

Figure 3. (a) Spatial distribution of all significant bare-ice albedo trends (75%, 85% and 95% confidence level) in the 1999–2022 time series of average pixel-by-pixel summer bare-ice albedo. Bare-ice albedo time series of a pixel displaying a decreasing (b) and increasing (c) trend. For both (b) and (c), the time series include all scenes, which are in chronological but non-uniform time spacing.

Figure 4. The rate of change of bare-ice albedo from 1999 to 2017 for pixels showing trends at the 85% confidence level for (a) July, (b) August and (c) September.

5.3. Air temperature impact on bare-ice albedo

We detected significant correlations between all assessed air temperature parameters and mean bare-ice albedo (Table 1). Some of these correlations are illustrated by scatter plots in Appendix 1, Fig. A3. Moderate negative correlations can be observed between mean bare-ice albedo and mean air temperature of the 6 h, 12 h, 24 h, 48 h, 60 h and 72 h time windows prior to albedo acquisition; the correlations are strongest with the average air temperature of the 48 h and 60 h time windows ( $\tau = -0.44$). Moderate negative correlations are also present between albedo and night minimum air temperature ( $\tau = -0.36$). Positive moderate correlations are apparent between bare-ice albedo and the sum of negative degree hours prior to albedo acquisition; the correlation is strongest with the sum of negative hours in the 48 h time window (τ = 0.45) and in the 120 h time window (τ = 0.47). Also, positive moderate correlations exist between bare-ice albedo and the sum of sign changes of air temperature (freezing/non-freezing conditions) during the time prior to albedo acquisition; the correlation is strongest with the sum of sign changes of air temperature in the 48 h time window prior to albedo acquisition (τ = 0.48).

Table 1. Kendall τ derived between Abramov average bare-ice albedo and the following parameters: average air temperature ( ${}^\circ\mathrm{C}$), nighttime minimum temperature ( ${}^\circ\mathrm{C}$), the sum of negative degree hours and the sum of sign changes of air temperature for various time intervals prior to albedo acquisition. All relationships are statistically significant at the 99% confidence level with p-values < 0.01

Figure 5 illustrates that high temperatures coincide with low bare-ice albedo (Fig. 5a, b, c, d), whereas low temperatures, negative degree hours and sign changes of air temperature correspond to high average bare-ice albedo (Fig. 5a, b, c, d). For example, the highest mean bare-ice albedo occurs on 23 September 2020 after 60 h of negative temperatures occurring in the last 5 days and four sign changes of air temperature in the past 48 h (Fig. 2d1).

Figure 5. Average bare-ice albedo over the tongue for each scene (orange line) of (a) 2014, (b) 2018, (c) 2019 and (d) 2020. Average albedo of the tongue, including albedo values above 0.39 for each scene (light orange area) and average air temperature of the 48 h prior to albedo acquisition (dark blue) and sum of negative hours in the 120 h prior to albedo acquisition (light blue) for the 4 years (a–d). Spatial distribution of bare-ice albedo after cold conditions in a1 (15 September 2014) and d1 (23 September 2020) and after warm conditions in b1 (9 August 2018) and c1 (12 August 2019).

5.4. Bare-ice albedo impact on glacier melt

We detected a negative correlation between the mean summer bare-ice albedo and summer melt ( $\tau = -0.40$) and a moderate negative correlation between the summer minimum mean bare-ice albedo and summer melt ( $\tau = -0.53$) from 1999 to 2022 (see Table 2). We also found a strong negative correlation between the summer minimum mean bare-ice albedo and summer melt ( $\tau = -0.71$) from 2012 to 2022, the period during which melt was measured in situ. Figure 6 illustrates that most years with low melt correspond to years with high minimum summer mean bare-ice albedo and vice versa. The direction and the rate of changes in melt rates is mirrored in the minimum summer mean bare-ice albedo (Fig. 6). A negative moderate correlation is also apparent between the mean bare-ice albedo of July, August and September and the respective melt within each month ( $\tau = -0.40$). Similarly, a negative moderate correlation is present between the mean bare-ice albedo of individual scenes and daily surface melt ( $\tau = -0.43$). These relationships between bare-ice albedo and melt rates across various timescales are illustrated in Appendix 1, Fig. A4, and the correlation values between bare-ice albedo and mass balance are provided in Appendix 1, Table A1. Appendix 1, Fig. A5 depicts the relationship between minimum summer mean bare-ice albedo and glacier-wide annual mass balance. Most years with a positive mass balance correspond to years with high minimum summer mean bare-ice albedo and vice versa (see Appendix 1, Fig. A5).

Table 2. Kendall τ between 1999 and 2022 summer melt and mean summer bare-ice albedo over the tongue and summer minimum mean bare-ice albedo over the tongue, between monthly (July, August, September) surface melt and monthly bare-ice albedo over the tongue, between daily surface melt (albedo acquisition dates) and the average bare-ice tongue albedo per scene. Kendall τ between 2012 and 2022 summer melt and summer minimum mean bare-ice albedo over the tongue, corresponding to the in situ melt measurement period. All relationships are statistically significant at the 99% confidence level with p-values < 0.01

6. Discussion

The bare-ice albedo on Abramov shows a sub-seasonal cycle which reaches its lowest values and variability in August compared with those in July and September. The reduction of albedo between July and August is pronounced at high elevation on the Abramov’s tongue, while albedo remains consistently low at the snout. Our results reveal an inverse relationship between seasonal temperature and bare-ice albedo. Yet, a time lag exists between the peak insolation in July, the warmest period from mid-July to mid-August, and the minimum bare-ice albedo at high elevation on the tongue in August. The time lag is likely attributed to elevation-dependent refreezing events in July, which maintain a high and fluctuating albedo at higher elevation. In August, the rise in air temperature along the altitudinal gradient reduces refreezing, leading to decreased albedo values and variability over the entire tongue. The decrease of albedo between July and August can also be related to prolonged exposure of ice to impurity deposition (Azzoni and others, Reference Azzoni, Senese, Zerboni, Maugeri, Smiraglia and Diolaiuti2016) and to a decrease in ‘washing-out’ of impurities by rainfall (Azzoni and others, Reference Azzoni, Senese, Zerboni, Maugeri, Smiraglia and Diolaiuti2016, Fugazza and others, Reference Fugazza, Senese, Azzoni, Maugeri and Diolaiuti2016). Azzoni and others Reference Azzoni, Senese, Zerboni, Maugeri, Smiraglia and Diolaiuti(2016) also observed a disconnection between peak of insolation and the Forni glacier’s most intense darkening. The authors relate the phenomena to persisting snow when insolation is strongest. Whereas snow often persists in ogives and crevasses at higher elevation (between 3950 and 4071 m a.s.l.) on Abramov’s tongue in July, snow cover changes cannot explain the delay between temperature, insolation and darkening. However, the role of snow in washing-out impurities might contribute to maintain a higher albedo in July. At lower elevation (between 3650 and 3850 m a.s.l.), the tongue of Abramov is preserved from refreezing events and is exposed to warm conditions as early as July, contributing to low and consistent albedo at the snout. This exposure enhances energy inputs to the surface and raises the ice temperature to melting point. The meltwater coating and ponds darken the ice and render it vulnerable to accelerated darkening through the albedo feedback (Paul and others, Reference Paul, Machguth and Kääb2005, Senese and others, Reference Senese, Diolaiuti, Mihalcea and Smiraglia2012, Azzoni and others, Reference Azzoni, Senese, Zerboni, Maugeri, Smiraglia and Diolaiuti2016).

In September, the rise of bare-ice albedo coincides with reduced sunlight, lower temperature and more negative degree hours. The impact of meltwater on the reflectivity of the ice is reduced, and recurrent refreezing of residual meltwater and weathering crust formation can generate high-reflective ice surfaces (Marshall and Miller, Reference Marshall and Miller2020, Tedstone and others, Reference Tedstone, Cook, Williamson, Hofer, McCutcheon, Irvine-Fynn, Gribbin and Tranter2020, Barandun and others, Reference Barandun, Bravo, Grobety, Jenk, Fang, Naegeli, Rivera, Cisternas, Münster and Schwikowski2022). Additionally, the extensive meltwater in August can have washed-out the impurities on the ice and thus increased the albedo for September. The brightening in September is more pronounced on clean ice than on dirty ice because of the reflective properties of clean ice, which quickly generate the necessary conditions for refreezing (Marshall and Miller, Reference Marshall and Miller2020).

The trends of bare-ice albedo around the medial moraines reveal a dynamic shift towards the east. This shift is likely the result of changing ice dynamics related to a possible surge initiation in 2018 (Mattea and others, Reference Mattea, Machguth, Berthier and Hoelzle2023). Lateral displacement and foliations of medial moraines, such as the one on Abramov, have been observed on other surging glaciers (Post, Reference Post1972, Wilson and others, Reference Wilson, Carrión and Rivera2016, Lovell and Fleming, Reference Lovell and Fleming2023). The significant trends between 1999 and 2017, unaffected by the glacier dynamics, reveal a widespread darkening of the ice in July, August and September. This darkening is likely related to the atmospheric warming recorded for Central Asia (Parry, Reference Parry2007), which increases the frequency of melting events, maintains water-coating on the ice for extended periods and suppresses the occurrence of freezing events (Cutler and Munro, Reference Cutler and Munro1996, Naegeli and others, Reference Naegeli, Huss and Hoelzle2019). The darkening of the glacier tongue in July can also be attributed to the earlier onset of melting of the snow cover in the Pamir-Alay (Zhou and others, Reference Zhou, Aizen and Aizen2017). This onset results in bare ice exposed to insolation, warm conditions and impurity deposition earlier in the season and for an extended period. The darkening observed at high elevation in September may be related to the gradual delay in the elevation-dependent refreezing events descending over the tongue (Peng and others, Reference Peng, Zhou, Zhang and Zou2019).

The detection of potential brightening and darkening of the ice over the full study period remains challenging due to short-term trends and processes that alter the glacier surface. These changes include processes such as the movement of morainic material or the displacement of meltwater streams on the ice. These short-term trends illustrate the importance of observing bare-ice albedo over extended time periods, as underlying darkening of the ice might otherwise be missed.

The progressive darkening of the tongue has critical implications for the melt. The sensitivity of the mass balance to bare-ice albedo variability stems from the crucial role that bare-ice albedo plays in the surface energy budget, triggering melting, thus mass loss and, vice versa, refreezing events that retain mass. These events lead to darkening or brightening phenomena on the ice and contribute to amplified feedback loops of mass loss or gain. The consistency of the correlation between bare-ice albedo and melt over the different time scales highlights that the relationship between the two variables is robust, which indicates that even short-term variability of bare-ice albedo can significantly impact glacier mass loss. Therefore, monitoring bare-ice albedo variability shows potential to provide valuable insights into the response of glaciers to climate change.

Figure 6. Summer minimum mean bare-ice albedo over the tongue (dark red) and summer melt (orange) from 1999 to 2022.

7. Conclusion

The bare-ice albedo of Abramov exhibits sub-seasonal and interannual spatio-temporal variability. Changes in air temperature control the temporal variability of bare-ice albedo during the ablation season. We observe inverse correlations between short-term average air temperature and the average bare-ice albedo on the tongue. Recurrent refreezing at high altitude on the tongue of Abramov is believed to generate the lag between the highest temperatures in July and the lowest average bare-ice albedo in August. Concentrations of impurities, debris and water on the ice dictate the spatial variability of bare-ice albedo during the ablation season. Our results indicate that the albedo of debris-free ice experiences a more pronounced increase than that of the ice in the vicinity of the medial moraines when temperatures and insolation drop from August to September. Bare-ice albedo decreased on Abramov in July and August from 1999 to 2017. These trends are likely related to higher air temperatures and the extension of the snow-free period, both of which can contribute to more frequent and widespread meltwater presence at the glacier surface. Changes in ice dynamics have also occurred on the glacier in the last decade. A lateral displacement of moraines in 2018 suggests that Abramov entered a dynamic phase related to surging activity (Mattea and others, Reference Mattea, Machguth, Berthier and Hoelzle2023), which is indeed reflected in the albedo values and trends.

Simplified versions of the spatio-temporal variability of bare-ice albedo have been used so far in glaciological modelling (Brock and others, Reference Brock, Willis and Sharp2000a). Yet, the significant correlations between variations of bare-ice albedo and glacier melt demonstrate the sensitivity of Abramov Glacier to changes of bare-ice albedo. Further research on the drivers of bare-ice albedo variations and the impact of those variations on glacier mass balance are required. With such inputs, bare-ice albedo variations could be integrated effectively into glaciological models to reduce uncertainties in melt modelling and its resulting contribution to river runoff.