Measurement of light-absorbing particles in surface snow of central and western Himalayan glaciers: spatial variability, radiative impacts, and potential source regions

. We collected surface snow samples from three different glaciers – Yala, Thana, and Sachin – in the central and western Himalayas to understand the spatial variability and radiative impacts of light-absorbing particles. The Yala and Thana glaciers in Nepal and Bhutan, respectively,


Introduction
Black carbon (BC) is a distinct type of carbonaceous material that is formed primarily in flames. BC particles in the atmosphere are generally produced by the incomplete combustion of fossil fuel, biofuel, and biomass. BC is only a minor contributor to aerosol mass but has great climatic interest as a strong absorber of solar radiation (Quinn et al., 2008;Ramanathan and Carmichael, 2008). In addition to warming, BC particles can interact with clouds, changing their microphysical properties and thus impacting the climate (Wang et al., 2018;Bond et al., 2013). Besides this, several studies in the past have highlighted the role of BC in the cryosphere .
The cryosphere is one of the most sensitive indicators of climate change. The temperature rise in cryospheric regions is generally larger than that in other regions on the global scale (Pepin and Lundquist, 2008;Kang et al., 2010;You et al., 2021). BC particles are deposited on the glaciers or snow cover surface, decreasing the surface albedo and absorbing more solar radiation (Warren and Brandt, 2008;He et al., 2017), which accelerates snowmelt and ice melt and triggers albedo feedback (Flanner et al., 2009;Hansen and Nazarenko, 2004;Kang et al., 2020). The forcing produced by BC and other light-absorbing particles (LAPs) further affects the regional climate (Flanner et al., 2009;Ji et al., 2015) and leads to complex responses of the Earth's climate system (Hansen et al., 1997). The largest climate forcing from BC in the snow is estimated to occur over the Tibetan Plateau (TP) and Himalayas (Flanner et al., 2009;Ji et al., 2015).
Mountain glaciers are the most important freshwater resources to the inhabitants of arid and semi-arid regions (Hock, 2005;Mayer et al., 2006). The Himalayas is considered the world's largest freshwater reservoir outside the polar regions (Immerzeel et al., 2010;Marcovecchio et al., 2021). The economy and lives of millions of people in the region are influenced by the changes in mountain river discharge downstream of the Himalayas (Vaux et al., 2012). Lack of in situ data, the low resolution of emission inventories, and coarse model resolutions prevent an accurate evaluation of LAP impacts on snow albedo and radiative forcing. Many glaciers have retreated in the region due to climate warming (Zhang et al., 2009;Kang et al., 2010;Yao et al., 2019) and possibly due to LAP-induced surface darkening (Flanner et al., 2009;Qian et al., 2011;Kang et al., 2019). Glacier retreat in the TP and the Himalayan region has serious consequences because snow and runoff from this region are sources of major rivers in Asia, and the availability of freshwater resources has profound effects on human health and agriculture (Immerzeel et al., 2010). However, large uncertainties remain regarding glacier retreat driven predominately by the deposition of BC and other LAPs (Bolch et al., 2012;Kang et al., 2020).
Snow albedo is an important indicator of the surface energy budget over a snow-covered area. Small changes in surface snow albedo can have large impacts on surface warming due to the rapid feedback involving changes to sublimation, snow morphology, and melt rates (Bond et al., 2013). The concentration of LAPs in surface snow is a major factor that affects snow albedo. BC and other LAPs present in the snow reduce the albedo in the visible portion of the electromagnetic spectrum (Flanner et al., 2007). Besides the concentration of pollutants deposited on the surface of the snow, multiple other factors, such as the solar zenith angle (SZA), snow grain size, snow grain shape, snow surface texture, snow density, and snowpack thickness, can also affect snow albedo (He and Flanner, 2020). The radiative transfer model used for the albedo has brought a better understanding of snow optical properties in the shortwave spectrum (He and Flanner, 2020;Tripathee et al., 2021). We estimated the spectral snow albedo using the online Snow, Ice, and Aerosol Radiative (SNICAR) model (Flanner and Zender, 2006). The model was originally developed by Flanner et al. (2007), was further updated by He et al. (2018) and Dang et al. (2019), and has been widely used in simulating the impacts of LAPs on snow albedos (Qu et al., 2014).
Here we present the mass concentration of BC, waterinsoluble organic carbon (OC), and mineral dust in surface snow from the ablation and accumulation zones of selected glaciers, located in three different countries (Nepal, Bhutan, and Pakistan) on the southern slope of the Himalayas. The Yala and Thana glaciers were selected from the central Himalayas, while the Sachin glacier was selected from the western Himalayas. To reasonably compare the results (mass concentrations and optical and radiative properties) across the central and western Himalayas, samples were collected on similar dates of the same seasons (pre-monsoon and postmonsoon). We investigate the spatial variability in BC, OC, and mineral dust concentrations due to differences in the source region, meteorology, deposition, and post-deposition processes. The measured mass concentrations were compared to regional model simulations. The associated changes in surface snow albedo and radiative forcing (RF) by mineral dust and BC in surface snow were estimated using the SNICAR model. We also aim to identify the potential source regions of pollution reaching sampling sites using tracertagged model simulations.  Table 2). The Yala glacier is a plateau-shaped glacier that has an elevation range between 5160 and 5750 m a.s.l. The length of the Yala glacier is 1.5 km facing the northwest. The glacier is located away from the residential area and is mostly covered by snow, especially during the winter season. Details about the metrological conditions at the Yala glacier are available in Rai et al. (2019) and Gul et al. (2021). The Thana glacier is a gently sloping glacier with slight debris cover and an elevation range between 5250 and 5700 m a.s.l. The length of the glacier is about 5 km, facing the southwest. The Thana glacier is mostly covered by fresh snow, especially during the winter season. The Sachin glacier has a gentle slope with dense debris cover in its ablation area with an altitude range from 3105 to 4976 m a.s.l. The length of the Sachin glacier is around 8 km facing northeast. In general, the Sachin glacier is a low-elevation and relatively debris-covered glacier compared to the central Himalayan glaciers (Yala and Thana). Precipitation in the central Himalayan glaciers (Yala and Thana) was higher than that of the western Himalayan glacier (Sachin), especially from April to October each year (Table 2). Surface air temperature over the Yala and Sachin glaciers was higher than that of the Thana glacier. The geographical location of the selected glaciers and snow sampling locations are shown in Fig. 1. Besides the difference in altitude, latitude, and meteorology of the selected glaciers in the central and western Himalayas, there is also a difference in the surface conditions shown in Fig. S1 in the Supplement.

Snow sampling and analysis
Surface snow samples were collected from the central and western Himalayan glaciers during May and September 2016. Samples were taken from the ablation and accumulation zones of the selected glaciers; however, a few snow sam-ples were also collected from the surrounding nearby areas of the Yala and Sachin glaciers. The Sachin glacier samples were of relatively aged snow and had less snow thickness as compared to the samples collected from the Thana and Yala glaciers (Fig. S1). At each sampling location, Whirl-Pak bags were used to collect samples from the upper 0-10 cm of depth (approximately 2 L, unmelted). The samples were kept frozen until they were melted and filtered in through the quartz filters near the sampling site. The snow density was measured with a small density kit. The snow grain size was measured through a hand lens (25×) with an accuracy of 0.02 mm. The same sampling protocol was used for all three selected glaciers. A detailed description of the sampling procedure is described in Y. . Quartz filters were used to measure the mass concentration of BC, OC, and dust in the collected samples. BC and OC present in snow samples were analyzed by a filter-based thermaloptical analysis method using a DRI ® Model 2005 analyzer (Chow et al., 1993). Filters were analyzed at the State Key Laboratory of Cryosphere Sciences, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences. Before starting the analysis, a piece of the sampled filters was put in an oven for a few minutes to eliminate the water vapor content and volatile organic compounds. Further detailed information on the instrument and analysis method can be referred to in earlier studies (Gul et al., 2018).

Estimation of snow albedo reduction and radiative forcing
The online snow simulation model SNICAR (Flanner et al., 2007, http://snow.engin.umich.edu/, last access: 20 April 2022) was used to estimate snow albedo calculation for the collected samples. The model has been used by multiple studies in the past (Gul et al., 2018;Zhang et al., 2018). Albedo was simulated based on an hourly SZA at the sampling site with an averaged mass concentration of BC, dust, and other input parameters such as snow grain size, snow density, and snow depth from measurements. We computed broadband snow albedo for direct solar incident radiation under mid-latitude, winter, clear-sky conditions (Table S1 in the Supplement). Depending on geographical location, 10 to 15 SZAs were used (between 0 and 90 • ) during instantaneous daytime albedo simulation. Albedo was simulated in four categories: (1) broadband albedo with BC and dust in snow; (2) broadband albedo with BC in snow only; (3) broadband albedo with dust in snow only; and (4) broadband albedo with the absence of BC and dust, which was considered a reference albedo. Radiative implications caused by snow darkening due to BC and dust deposition were investigated using the albedo reduction and the radiative transfer model Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART) (Ricchiazzi et al., 1998). To evaluate the amount of additional solar radiation absorbed by the snow in the presence of BC and dust, we estimated the mean solar irradiance and its characteristics via SBDART, which has been used in the past (Yang et al., 2015). According to the location of the sampling site, the characteristics of the atmospheric profiles such as water vapor, aerosols, or ozone were set in the model. RF for the snow samples was estimated by following Eq. (1): where R in-short denotes incident shortwave solar radiation for selected SZA and α x denotes the reduction in albedo due to BC, dust, or both, as simulated by the SNICAR model.

Potential source region of pollutants
Glaciers of the Himalayan Karakoram and Hindu Kush (HKH) region are located at high altitudes as compared to the sources of the major pollutants. LAPs including BC and dust can be transported from urban areas towards glaciated areas (Yasunari et al., 2010;Kang et al., 2019). Multiple approaches, including climate circulation modeling, combinations of bottom-up inventories, and back air trajectories, have been used in the past to determine the possible source regions of pollution in the HKH region. To identify the potential source region of pollution arriving at the observation sites, we used the Weather Research and Forecasting (WRF) model coupled with Chemistry (WRF-Chem version 3.9.1.1) simulations (Grell et al., 2005). The model uses region-tagged tracers for different regions across the world. WRF-Chem simulations were used to estimate BC mass concentration in surface snow and deposition of BC particles on three selected glaciers (Yala, Thana, and Sachin). We archived the hourly model results for instantaneous BC deposition and concentration in snow. The horizontal grid spacing of the model was 20 km × 20 km with 35 vertical levels stretching from the surface up to 50 hPa (∼ 20 km). The updated Model for Ozone And Related chemical Tracers (MOZART) was applied for the gas-phase chemistry (Knote et al., 2014), while aerosols in the WRF-Chem were simulated via the Model for Simulating Aerosol Interactions and Chemistry (MOSAIC) (Zaveri et al., 2008). We use the Global Data Assimilation System (GDAS) from the National Centers for Environmental Prediction (NCEP) for the meteorological initial and boundary conditions. We used the Fire INventory from NCAR (FINN; NCAR is the National Center for Atmospheric Research), Emission Database for Global Atmospheric Research developed for Hemispheric Transport of Air Pollution (EDGAR-HTAP), and MEGAN (Model of Emissions of Gases and Aerosols from Nature) for biomass burning emissions, anthropogenic emissions, and online biogenic emissions, respectively. For chemical boundary conditions, we used the NCAR global CAM-chem simulation dataset (https://rda.ucar.edu/datasets/ds313.7/, last access: 22 June 2022). Key meteorological variables such as winds, temperature, and water vapor above the planetary boundary layer (PBL) were nudged every 6 h towards the NCEP GDAS reanalysis fields to reduce temporal error growth in meteorological variables. We used the Community Land Model (CLM) scheme for the land component in WRF-Chem. The CLM can simulate BC concentration in snowpack and its effects on snow albedo (Flanner et al., 2007). We used online coupled BC deposition fluxes from the atmosphere component of WRF-Chem with the CLM following Zhao et al. (2014). We also implemented a taggedtracer method (Kumar et al., 2015) to track anthropogenic BC emissions from 10 different Asian countries surrounding the TP areas, as well as BC emissions from Asian biomass burning and the domain boundary (i.e., areas outside Asia). The 10 tracked anthropogenic emission source regions include China, India, Nepal, Pakistan, Afghanistan, Bhutan, Bangladesh, Myanmar, southeast Asia, and the rest of Asia.
The aim of the model simulation was to estimate the BC mass concentration in surface snow, deposition of BC particles, and the source contribution to BC deposition on snow.

Concentrations of light-absorbing particles in surface snow
The average mass concentration of LAPs in surface snow of the Yala glacier was 358 ng g −1 for BC, 904 ng g −1 for OC, and 22 µg g −1 for dust in spring (May), and there were relatively low concentrations of 69 ng g −1 for BC, 177 ng g −1 for OC, and 4 ng g −1 for dust during autumn (September). These mass concentrations of BC and OC in surface snow were comparable to the study result conducted on the Yala glacier in May 2017 . High LAP concentration in the pre-monsoon is due to an effective transport mechanism from the Indian subcontinent and an additional source such as forest fires . The average surface concentrations of BC, OC, and dust in the Thana glacier samples during the autumn season were 39, 115 ng g −1 , and 34 µg g −1 , respectively. Possible reasons for the lower concentration at the Thana glacier may be due to the relatively high elevation of the sampling location and relatively fresh snow. A strong effect of LAPs (BC and dust) has been observed at lower elevations in comparison to higher elevations . The average concentrations of BC, OC, and dust measured in the selected western Himalayan glacier (Sachin) during May were 2381, 3896, and 101 ng g −1 , respectively, and were relatively high during October with values of 5314 ng g −1 for BC and 546 µg g −1 for dust (Gul et al., 2018). The observed average mass concentrations in the western Himalayas were higher than those in the central Himalayas. The BC mass concentration difference might be due to the difference in snow type, precipitation rate, local emission, the elevation of sampling sites, meteorology, and BC deposition over the glacier surfaces. Post-monsoon dry deposition of LAPs over the surface of the snow was an important factor. The pollutant source regions for the central and western Himalayas are different. In the case of the central Himalayas, pollutants emitted during premonsoon convection and multiple forest fire events are effectively lifting and transported towards the central Himalayan glaciers. Due to strong inversion in winter, most of the pollutants become stuck near the surface, whereas in the monsoon, pollutants are scavenged by rain. Thus the pre-monsoon is a very significant period in the transport process in the central Himalayas. Snow samples collected from the western side of the Himalayas were aged as compared to the central part; post-deposition ion (or enrichment) of LAPs over the snow surface increased the concentration in the snow . The majority of the samples from the western Himalayan side were from ablation zones of the glacier, where concentrations of LAPs are higher as compared to the accumulation zone of the glacier. Li et al. (2017) showed a strong negative relationship between the elevation of glacier sampling locations and the concentration of LAPs. Therefore strong melting of surface snow and ice in the glacier ablation zone could lead to BC enrichment, which causes high BC concentrations . In the case of western Himalayan glacier sites, snow samples were collected long after the snowfall and the concentration of pollutants would also have increased in the surface snow due to dry deposition. The surface concentrations of the individual samples collected from the Yala, Thana, and Sachin glaciers during May and September 2016 are shown in Fig. 2 and Table S2. BC and OC concentrations on our selected glaciers with a comparison to other glaciers of the TP and the surrounding region are shown in Fig. 3 and Table S3. It was observed that the concentrations of BC, OC, and dust in the central Himalayan glaciers (Yala and Thana) were comparable to other reported results. In the past, similarly high concentrations were reported in the region such as the Tian Shan (Tibetan) glacier  (Niu et al., 2020;Zhang et al., 2017); central Asian glaciers ; and Tian Shan .
The yellow boxes of Fig. 2 show the WRF-Chem-modeled BC concentrations in surface snow at the three measurement glacier sites during the measurement periods. Compared to the observations shown with red boxes in Fig. 2, model results reasonably capture the spatial and seasonal patterns and variables of the observed BC in the snow with a relatively small magnitude. The modeled variation at the Sachin site during the October sampling periods is much larger than that of the observations (Gul et al., 2018). The discrepancies between model results and observations are due to model uncertainties from (1) the relatively coarse grid spacing that may not capture the transport over the complex TP terrain, (2) the underestimated anthropogenic emissions that are not representative of the measurement periods, and (3) deficiencies in model physical parameterizations that affects BC transport and deposition. The WRF-Chem model implicitly accounts for the surface impurity enrichment during snow ablation by using a low meltwater scavenging efficiency for BC. However, we notice that this meltwater scavenging efficiency parameter could be associated with large uncertainties  due to the lack of direct observational constraints. We also note that the observed variation at each site shown in Fig. 2 includes both the temporal and the subgrid variabilities derived from multiple sampling locations surrounding each site (Fig. 1). In contrast, all the measurement locations at each particular glacier site are located within a single model grid. As a result, the model is unable to resolve this subgrid information and hence only includes the temporal variability for each selected site.

Surface snow albedo and radiative forcing
The minimum daytime absolute albedo reduction due to combined BC and dust, BC only, and dust only was in the range of 1.03 %-13.44 %, 0.48 %-12.42 %, and 0.12 %-2.12 %, respectively. The maximum daytime albedo reduction due to combined BC and dust, BC only, and dust only was in the range of 1.98 %-24.97 %, 1.05 %-24 %, and 0.25 %-4.8 %, respectively. The lowest and highest contributions in albedo reduction were observed in the central Himalayas (September) and the western Himalayas (May), respectively. Snow albedo reduction (%) derived from samples collected from the Yala glacier (during May 2016) and the Thana glacier (during September 2016) was in the range of 0.13 %-3.82 % and 0.90 %-1.99 %, respectively. A significant difference in daytime albedo reduction between the western and central Himalayas was mainly due to the difference in mass concentrations of pollutants and snow age. The pollutant concentrations in the western Himalayan samples (Sachin glacier) were higher, resulting in higher albedo reduction as compared to the central Himalayan (Yala and Thana glaciers) samples. The average elevation difference between central and western sampling sites was greater than 1000 m, where a high concentration of pollution is expected at the low-elevation glacier of the western side as compared to the central part of the Himalayas. Snow samples collected in the central part of the Himalayas (Yala glacier) were much fresher as compared to the samples collected from the western side (Sachin glacier). Dust and other pollutants were visible over the surface of the Sachin glacier (Fig. S1). Aged snow had increased density, enlarged grain size, and increased concentration of BC and dust particles due to dry deposition on the snow surface. In the case of all sampling sites, the impact of BC on snow albedo reduction was greater than the impact of dust except at the Thana glacier, where the impact of dust was higher than that of BC (Fig. 4a). This may be due to a different dust type in Thana samples. Daytime snow albedo reductions (%) due to BC only, dust only, and both BC and dust are given in Fig. 4a.
The daytime instantaneous RF (W m −2 ) ranged from 0.076 to 39.65 for the Yala glacier in May 2016, 0.006 to 18.26 for the Yala glacier in September 2016, 0.0 to 11.48 for the Thana glacier in September 2016, and 0.03 to 96.48 for the Sachin glacier during May 2016. RF for the western Himalayas (Sachin glacier) was quite high as compared to the central Himalayan glaciers (Yala and Thana glaciers). The radiative effect on the Sachin glacier was much more than that of other selected glaciers mainly due to low albedo and increased temperature. Zhang et al. (2017) reported that a reduction in albedo by 9 % to 64 % can increase the instantaneous RF by as much as 24.05-323.18 W m −2 . In the case of all sampling sites, the impact of BC on RF was greater than the impact of dust except at the Thana glacier, where the impact of dust was higher than that of BC (Fig. 4b). Therefore, BC can be a major pollutant in the snow responsible for reducing albedo and increasing warming in the selected glaciers. BC was the dominant factor in snowmelt in the Yala and Sachin glaciers; however, dust was the dominant factor in Thana glacier samples. According to Kaspari et al. (2014), RF caused by mineral dust was greater than that of dust. The BC and dust had low importance for RF in fresh snow (central Himalayas -Thana glacier) as compared to aged snow (western Himalayas -Sachin glacier). In the northern TP, BC played an important role in RF (C. , while in the central TP and Himalayas, dust was more important than BC (Kaspari et al., 2014). The average instantaneous RF caused by the combined contribution of BC and dust (BC + dust), only BC, and only dust is shown in Fig. 4b as a function of surface snow types. Variation in the RF and albedo change for a particular pollutant type was due to variation in the SZA. Figure 5 shows the contributions of different BC emission sources to the BC in snow from WRF-Chem tagged-tracer simulations. For the Yala site, it is dominated (> 50 %) by anthropogenic emissions from India and Nepal for both May and October, while the biomass burning contribution (> 20 %) increases largely in May primarily due to the spring burning activities in northern India (Kumar et al., 2011). In September, China's contribution also increases to > 20 % at Yala. For the Thana site, it is dominated (> 60 %) by anthropogenic emissions from China and India in September, while anthropogenic emissions from Bhutan and Myanmar each contribute about 10 %. The Sachin site is predominantly affected by anthropogenic emissions from India and Pakistan (total contribution > 80 %), while the spring biomass burning only contributes > 10 % in May. Overall, the source contributions show large variation depending on the site locations and sampling seasons, but there is a consistent India contribution of 20 %-40 % across all the sites and seasons.

Discussion on uncertainty in measurements, albedo, and potential source identification of pollutants
The possible uncertainties in the present research were related to measurements, sampling, analysis, albedo, and RF estimation. When sampling at remote rural sites, sample preservation, filtration, and transport can modify the results if proper standard protocols are not followed. During laboratory analysis via thermal optical techniques, several uncertainties may be related to separating OC from BC in the sample . The level of generated uncertainty depended on the temperature protocol, sample type (residential cookstoves, diesel exhaust, rural aerosols, and urban aerosols), the amount of dust loading on the filter, and the analysis method. The overall accuracy in the measurement of OC, BC, and total carbon concentrations was estimated considering the mass contributions from field blanks and the analytical accuracy of concentration measurements. The uncertainty in the OC and BC mass concentrations was extracted through the standard deviation of the field blanks . OC in snow can produce minor warming (Yasunari et al., 2015), but in this research albedo reduction from OC was not quantified. In albedo simulation and RF estimations, snow grain size and texture can produce large uncertainty. We measured/considered the physical grain size in this re-search, which does not have the same as the effects as optical grain size. Optical grain size defines the amount of solar radiation absorbed/scattered by the snow. We assumed a spherical shape for the snow grains, which may affect the results because the albedo of non-spherical grains is higher than the albedo of spherical grains (Dang et al., 2016;He et al., 2018). The contribution of pollutants generated from local sources can be important , but this was not included in the global emission inventories; thus we were unable to capture emissions at the local scale. Therefore contributions of local sources may be underestimated by coarse-resolution models. High-resolution models and emission inventories at the local scale are required to capture local emissions.

Conclusions
The average mass concentrations of LAPs in the samples collected from the Sachin, Yala, and Thana glaciers were in the range of 835 to 3545 ng g −1 for BC and 35 to 253 µg g −1 for dust, 23 to 2529 ng g −1 for BC and 1.5 to 196 µg g −1 for dust, and 21 to 127 ng g −1 for BC and 1.5 to 67 µg g −1 for dust, respectively. Overall the concentrations of BC and dust varied from 21 ng g −1 and 1.5 µg g −1 , respectively, in fresh snow to 3545 ng g −1 and 253 µg g −1 , respectively, in the aged snow. Mass concentrations of BC, OC, and dust in the samples collected from the western Himalayas was much higher than the average concentration in the central Himalayas mainly due to difference in snow age, elevation, and meteorology. In the accumulation area of glaciers, the deposition of pollutants is expected to be lower, enrichment influences are less marked, and measured values are likely to be lower. Pollutant concentrations were likely underestimated in the earlier studies, particularly when there was strong surface melting. Dust and other pollutants were visible on aged-snow surfaces in the western Himalayan glacier, indicating considerable enrichment during snow aging. WRF-Chem-modeled BC concentrations in surface snow were similar to the observed BC in the snow with a smaller magnitude. Based on observed pollutants, snow albedo reduction (%) in the central Himalayas was in the range of 0.48 %-3.6 % for BC and 0.13 %-1.99 % for dust, much lower than that of the western Himalayas. BC was the major component responsible for the albedo reduction, and the dust had little effect except at the Thana glacier. In the case of the Thana glacier, the impact of dust was higher than that of BC. The daytime instantaneous radiative forcing (W m −2 ) ranged from 0.076 to 39.65 (Yala glacier during May 2016), 0.006 to 18.26 (Yala glacier during September 2016), 0.0 to 11.48 (Thana glacier during September 2016), and 0.03 to 96.48 (Sachin glacier during May 2016). The average albedo reduction due to the combined effect of dust and BC on the western Himalayan side (Sachin glacier) was 0.372, which was ∼ 15 times higher than that of the central Himalayan part (Yala glacier). Similarly, the radiative forcing in the western Hi-malayas was ∼ 6 times higher than that of the central Himalayan part. Observation showed that the potential source regions of pollutants for the western and central Himalayas were different. Western Himalayan glaciers were mostly affected by long-range transport via the westerlies; however central Himalayan glaciers were affected by relatively local winds from Nepal, Bhutan, India, and China. For the western Himalayan glaciers, the emissions from central Asian and south Asian countries (particularly Pakistan and India) are more important source regions. Data availability. Precipitation and temperature data were used from https://doi.org/10.24381/cds.f17050d7 (Hersbach et al., 2019).
Author contributions. CG contributed with data collection, data analysis, image processing, and draft writing. SK and SPP contributed with conception or design of the work and critical revision of the manuscript and approved the final version. XW ran the model part, provided data for the model, and wrote the draft for the model part of the paper. CH ran the model part, provided data for the model, and contributed with critical revision of the manuscript. YX ran the model part, provided data for the model, and contributed with critical revision of the manuscript. IK collected data collection from one site and undertook analysis. SM collected date from one site and undertook analysis and revision. RK contributed with critical revision of the model part of the manuscript and image analysis for the model part. GD contributed with writing, language editing, and revision.
Competing interests. The contact author has declared that neither they nor their co-authors have any competing interests.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Acknowledgements. This study was partially supported by the core funds of ICIMOD contributed by the governments of Afghanistan, Australia, Austria, Bangladesh, Bhutan, China, India, Myanmar, Nepal, Norway, Pakistan, Sweden, and Switzerland. We thank Faiza Gul, Aditi Mukherji, and Arnico Panday for their useful comments and guidance. We are also grateful to the staff of the National Center for Hydrology and Meteorology in Bhutan for organizing the Thana glacier expedition in 2016. We would like to acknowledge high-performance computing support from Cheyenne provided by NCAR's Computational and Information Systems Laboratory, sponsored by the National Science Foundation. NCAR is operated by the University Corporation for Atmospheric Research under the sponsorship of the National Science Foundation.
Financial support. This study was supported by the Second Tibetan Plateau Scientific Expedition and Research program (grant no. 2019QZKK0605), the Chinese Academy of Sciences (XDA20040501, QYZDJ-SSW-DQC039), and the State Key Laboratory of Cryosphere Sciences (SKLCS20 ZZ-2021).
Review statement. This paper was edited by Anne Perring and reviewed by two anonymous referees.