1 Introduction

Mineral dust may have significant effects on climate by absorbing and scattering solar radiation as well as by modifying the optical properties of clouds(Forster et al., 2007). The long-range transport and deposition of mineral dust over the open oceans also provides micronutrients to marine organisms such as phytoplankton(Martin et al., 1991), and thus modulates the carbon cycle. In addition,mineral dust directly alters the rate and timing of snow/glacier melts(Painter et al., 2007) and ,on a regional scale,acts as a primary source of nutrients and toxic chemicals in ecosystems(Field et al., 2010).

Asia is second only to the Sahara desert as the largest aeolian dust source on Earth,contributing to the transport and deposition of dust in the surrounding and downwind areas every year. Many studies have focused on the impact of this dust on the air quality in downwind regions(e.g.,Zhuang et al., 2001)as well as the biogeochemical cycling effect of this dust in remote seas(Duce et al., 1980). However,the influence of this dust on the cryospheric areas of central Asia,which are sensitive to global change and vulnerable to disturbance,has not been extensively studied. Identifying the spatial distribution of dust deposition is an essential step in underst and ing the effect of dust on this cryospheric area. Over the past decades,several studies have focused on dust deposition in central Asia. For example,Wake et al.(1994)identified the spatial characteristics of dust deposits in 11 mountain glaciers in High Asia and found that the glaciers in the northern mountainous area of the Tibetan Plateau(TP)bounded by the Tanggula Mountains contained 10 times higher insoluble particle mass concentrations than did the southern mountainous area. An analysis of the ion concentrations in snow pit samples from the TP led to similar conclusions(Xiao et al., 2002). Based on aerosol collection from three sampling sites,Zhang et al.(2001)estimated the annual mean dust flux at the TP to be approximately 100 g/(m²·a),which is lower than that of the other Chinese arid regions(200-300 g/(m²·a)), and suggested that the TP itself is not a major source of Asian dust. Despite these studies,the characteristics of dust deposition in cryospheric areas of western China remain poorly understood due to its huge surface area. In addition,it remains unclear whether this spatial feature has been affected by recent climate change.

In 2008,a field study called "the Biogeochemical Cycles in the Cryosphere of Western China" was conducted to characterize and quantify the evolution of dust and pollution in the cryospheric areas of western China and evaluate their potential effects on the regional biogeochemical cycle(Xu et al., 2012; Zhang et al., 2012; Yu et al., 2013). The current study mainly focuses on the spatial characteristics of dust deposition on the mountain glaciers in western China.

From September 2008 to September 2010,a total of 15 snow pit samples were collected during field expeditions on 13 different glaciers in western China. The locations of the sampling sites,from north to south,are presented in Table 1 and shown in Figure 1. The elevations of the sampling sites roughly increased from the north to south,with the lowest and highest elevations being 3,605 m a.s.l.(Musidao Glacier: MSD) and 6,518 m a.s.l.(East Rongbuk Glacier: QM),respectively. Most of the sampled glaciers were far from population-center regions except for sites on the peripheral mountains of western China,such as Tienshan Glacier(TS),which is approximately 120 km south of Urumqi City, and Yulong Mountain(YL),which is approximately 25 km north of Lijiang City. All of the snow pits were dug in the accumulation zone of the glaciers(where the accumulation of snowfall exceeded the losses from ablation)so that the data would be less affected by snow melting and local dust sources. The equilibrium line altitude(ELA)of each glacier,which was lower than that of each sampling site,is also presented in Table 1. Samples were collected into Whirl-Pak bags(Whirl-Pak^®,Nasco,USA)using a Teflon ice scoop at approximately 10-cm vertical intervals by personnel wearing non-particulating clean suits,masks, and gloves. The density,temperature, and physical description of the snow pit stratigraphy were also obtained during sample collection. The samples were maintained at a temperature below 0 ℃ using a cooler box with ice packs during the 1-2 days of transport. After arriving at the laboratory,the samples were stored at −18 ℃ until analysis.

Figure 1 A map showing the locations of snow pits in western China from which study samples were collected, 2008 to 2010. The base map shows the distribution of desert and semi-desert regions as well as the locations of glaciers and general patterns for winter and summer circulation systems in this region. The desert data were obtained from the U.S. Geological Survey(http://store.usgs.gov/) and the glacier data were obtained from the Glacier Inventory of China(Guo et al., 2014)

Figure options

The climatic features of western China are briefly described here(Figure 1). During the boreal winter,the East Asian winter monsoon dominates most of the northern parts of western China(e.g.,sites MSD,TS,MS, and LHG),whereas the westerly winds move south and dominate most of the southern study area(e.g.,sites DKMD,GL,JMYZ,QM, and ZD). During the boreal summer,the Asian monsoon(including the Indian monsoon and East Asian monsoon)dominates most of western China. The westerly winds shift to the north and influence the northern part of western China(e.g.,sites MSD,TS, and MS). Spring and autumn are seasons of transition for these climate systems. The summer monsoon provides most of the annual precipitation in western China(more than 70%). The amount of precipitation decreases from east to west and from south to north along the moisture trajectory,as illustrated by the annual accumulation rate(water equivalent)at each sampling site(Table 2). For example,the annual precipitations at the sampling sites DML and YL are approximately 2,500-3,000 mm,but only 200-500 mm at MSD and TS. MSD receives rainfall throughout the year,with a large portion of winter precipitation derived from the North Atlantic(Tian et al., 2007).

Prior to analysis,the samples were melted at room temperature in a clean room. Once completely melted,the samples were gently agitated for homogenization and then partitioned into prewashed low-density polyethylene(LDPE)bottles. A portion of each sample was used for water-insoluble particle analysis using a 300-channel Coulter Counter(Multisizer III)in a Class 100 clean room. Prior to analysis,a 2.5-mL sample was diluted at a 1:4 ratio with a 2% NaCl electrolyte. The particle sizes reported here had spherically equivalent diameters and ranged from 1 to 30 μm, and the total mass of the insoluble particles was calculated from the volume size distribution assuming an average density of 2.5 g/cm³. The deposition flux of dust was calculated by multiplying the water equivalent thickness by the insoluble particle mass concentration and major water-soluble inorganic ion(Na⁺,NH₄⁺,K⁺,Mg²⁺,Ca²⁺,SO₄²⁻, and Cl⁻)concentration of each sample. The residuals of the insoluble particle analysis were filtered through a nuclepore membrane(polycarbonate,with a pore size of 0.22 μm and diameter of 47 mm)using precleaned filter holders. The insoluble particles could generally be uniformly deposited on the filter by agitating the solution before filtering. The filters were dried at room temperature and divided into two aliquots for proton-induced X-ray emission(PIXE) and X-ray diffraction(XRD)analyses. Three procedural blanks were obtained by filtering approximately 100 mL of Milli-Q water, and all data reported in this work were blank-corrected. All of these pre-operation procedures were performed in a Class 100 laminar flow bench situated inside a Class 1,000 clean laboratory.

The filtered samples were analyzed by performing PIXE analysis in a 2×1.7 MV t and em electrostatic accelerator at the Institute of Low Energy Nuclear Physics,Beijing Normal University. Details for the PIXE analysis can be found in Zhang et al.(2011). Briefly,the PIXE analysis was conducted using 2.5-MeV proton bombardment with a beam current of 30-40 nA. The evaluation of X-ray spectra was done by the GUPIXWIN program,a code developed by Guelph University in Canada. PIXE can detect Si,Al,Fe,K,Mg,Ca,Ti,Mn,V,Cu,Ni,Zn, and Cr with analytical detection limits of <1 µg/kg. Note that the detection limits for the analytes reported here were governed by their uncertainties in the blank filters(three folds of st and ard deviation),rather than the instrumental detection limits.

The samples were prepared for XRD analysis using a back-loading preparation method. Mineralogical analyses were performed by XRD using a JDmax 12-kW powder diffractometer with a Bragg-Bentano θ:2θ configuration,Cu target tube, and graphite monochromator. The samples were analyzed within the range of 10°-70°,with a two-tube voltage of 40 kV,tube electric flow of 100 mA,scan rate of 8°/min, and a wide sampling step interval of 0.02°. XRD plots and MDI Jade 5 software were used for peak recognition,mineral identification, and peak intensity calculations. The data were identified according to the PDF2-2 Release of the 2004 database(JCPDS - International Centre for diffraction Data(2004)). The relative phase amounts(weights %)were estimated using the Rietveld method with the AutoQuan software(Seifert Analytical X-ray,AutoQuan,version 2.7.0.0).

The in situ descriptions of the stratigraphy for each snow pit are shown in Figure 2. Note that the stratigraphy for DKMD and Yala are not presented due to their shallow depth(approximately 20 cm). The depth of the snow pits varied from 70 cm(GLDD)to 295 cm(YL),depending on the amount of precipitation,collecting time, and the distance from the ELA at each site. Most of the snow pits consisted of new snow,ice layers, and firn. Except for QM,at least one dust layer appeared in each snow pit. For example,the TS snow pit had three dust layers, and JMYZ presented two dust layers that formed during dust storm events. However,at the LHG and DML sites,in addition to the dust layers we observed dust distribution throughout the entire profile,which was most likely the result of the effects of strong percolation during the summer. In addition,significant surface melting was also observed at MSD,DKMD, and YL.

Figure 2 Physical stratigraphy and profiles of insoluble particle concentrations (×106 mL−1) for each studied snow pit, except for DKMD and Yala, which were only 20 cm in depth

Figure options

The dating of snow pit samples was mainly based on the seasonal variations of δ¹⁸O,which is dominated by distinct seasonal fluctuations in western China(Tian et al., 2007). Details regarding the snow pit dating have been reported in Zhang et al.(2012). In addition,the dust layer(s)can also be useful for dating certain snow pits. For example,at TS the snow pit could be dated for three years(2006,2007, and 2008)because one dust layer is normally deposited each year(You et al., 2006). Our dating results are presented in Table 2. Most of the snow pit samples covered an entire year.

Figure 3a presents the mass concentration of insoluble particles at each site. For the ZD and LHG sites,which were sampled twice,the presented concentrations are the mean values of these two samplings. The mass concentration for Yala was not determined due to the shortage of samples. The mean mass concentrations of insoluble particles in each snow pit were much higher at the northern sites than at the southern sites(Figure 3a). The maximum value was observed at the TS site(43,651.8 µg/kg),followed by LHG(34,712.7 µg/kg),whereas the two minimum values were observed at JMYZ(1,411.1 µg/kg) and QM(545.6 µg/kg),which were located in the Himalayas. The average insoluble particle mass concentration at TS was approximately 50 times the mean concentrations at JMYZ and QM. The mass concentrations at the inl and sites DKMD,ZSGR and GL were 9,818.0,8,423.0, and 3,447.4 µg/kg,respectively,which were 5 to 10 times lower than the results from TS and LHG.

Figure 3 Insoluble particle data profiles in each studied snow pit. (a) arithmetic mean mass concentration; (b) dust deposition flux

Figure options

High dust concentrations,however,were also detected at ZD(19,262.9 µg/kg),MS(16,237.5 µg/kg),MSD(11,685.5 µg/kg), and DML(11,600.1 µg/kg). If the snow pits are grouped into geographic sub-regions,the region surrounding the Taklimakan Desert shows the highest concentration(31,428.5 µg/kg,the average of TS,MS, and LHG),which is about 4 times higher than that inl and of the TP(7,229.5 µg/kg,the average of DKMD,ZSGR, and GL),about 30 times higher than that of the Himalayas(978.3 µg/kg,the average of JMYZ and QM), and only about 3 times higher than that southeast of the TP(11,256.8 µg/kg,the average of ZD,DML, and YL). The dust concentrations in the snow of western China showed a decreasing trend from north to south,which is consistent with the results of Wake et al.(1994),who attributed this decreasing trend to the increasing distance from dust sources from north to south.

The annual fluxes represent the sum of the fluxes for all of the samples in an annual layer(Table 2). The flux at DKMD was not determined due to a lack of density data. The results are shown in Table 2 and Figure 3b. The two maximum values were observed at ZD(1,830.3 μg/(cm²·a)) and TS(1,494.0 μg/(cm²·a)),followed by LHG(1,416.0 μg/(cm²·a)),MS(1,032.3 μg/(cm²·a)),DML(1,000.5 μg/(cm²·a)),YL(729.7 μg/(cm²·a)),ZSGR(348.2 μg/(cm²·a)),MSD(333.3 μg/(cm²·a)),JMYZ(300.2 μg/(cm²·a)), and GLDD(146.0 μg/(cm²·a)). QM showed the minimum flux(64.7 μg/(cm²·a)). For the regional average of the northern part of western China,the average of TS,MS and LHG was 1,314.1 μg/(cm²·a),whereas the average flux for the area southeast of TP was 1,186.8 μg/(cm²·a). The minimum deposition fluxes were found at the inl and sites of the TP and Himalayas at 220.4 and 182.4 μg/(cm²·a),respectively.

The average volume size distributions at each sampling site were fitted using multi-mode lognormal functions(Figure 4). Most of the sampling sites exhibited bimodal distributions peaking at approximately 5 µm and 10 µm. The size distribution at MS had three modes,including a smaller mode at 3 µm,whereas the size distribution at JMYZ consisted of only one mode at 3.6 µm. Bimodal size distributions in snow pits have been reported in previous studies in Japan(Osada et al., 2004) and at the Hispar Glacier located on Karakoram Mountain of TP(Wake et al., 1994). Regarding dust transport,Rajot et al.(2008)reported two modes of geometric means equal to 8.7 µm and 4.7 µm, and stated that the first mode was dominant at the beginning of the dust event and decreased in importance with time after transport. For our data,the two modes may represent two dust sources at different distances,that is,the larger one represents a local dust source and the smaller one represents a regional dust source. The first mode in the northern part of western China decreased from approximately 6 µm in MSD and TS to approximately 3 µm in ZSGR and GL,suggesting a longer transport distance. However,the second modes exhibited less variation,supporting the possibility of a local source.

Figure 4 The average size distributions (SD) in each snow pit sample. The size distributions were fitted with multi-mode lognormal functions using one to four lognormal functions for each measured size distribution. The mode sizes (D1 and D2) of the fitted size distributions are also presented in each figure. D1 is corresponding to small size mode and D2 corresponding to bigger size mode

Figure options

The first modes in the northern TP region were also similar to the results at certain downstream sites(e.g.,Japan and Yukon Mountain),which peaked at 4-7 µm(Osada et al., 2004; Zdanowicz et al., 2006). The size distribution at JMYZ had only one mode at 3.6 µm,which suggests only one dominant dust source at that site; this mode was close to the first mode(3.1 µm)at QM. The sites southeast of the TP exhibited relatively larger values for these two modes. For example,the two modes for DML were 8.6 µm and 13.3 µm,respectively,indicative of the importance of local dust sources.

The relative amplitude of the two modes may suggest the relative contribution of the two types of dust sources. Overall,the two modes at each site showed a very similar amplitude,suggesting that local dust may dominate the mass concentration at most sites because of its larger diameter.

The bulk insoluble particle contents at sites TS and LHG,representing the northern regions, and ZD and DML,representing the southern regions,were analyzed by the XRD method. Table 3 shows the results of the mineralogical analysis,which demonstrated that the insoluble particles,on average,consisted of mica(27%),chlorite(26%),quartz(21%), and plagioclase(13%). At the two northern sites(TS and LHG),the insoluble particles consisted of a greater amount of quartz(30% and 24% vs. 20% and 9%,respectively),whereas they were mixed with more mica(35% and 27% vs. 21% and 25%,respectively)at the southern ZD and DML sites. In addition,at DML the insoluble particles were mixed with a much greater amount of talc(20%)than at the other three sites. The higher fraction of quartz at TS and LHG indicated the important contribution of desert s and s(Jeong,2008),whereas the increased mica fraction at ZD and DML was likely transported from local debris.

Figure 5 illustrates the relative chemical elemental compositions of the insoluble particles at each site. In general,the element contributions in the snow pits did not exhibit evident spatial characteristics(Figure 5a). The elements Si,Al,Fe,K, and Mg accounted for more than 97% of the chemical composition at all of the sites,with a contribution order of Si > Al > Fe > K > Mg. At site QM there was a relatively elevated amount of Mg(5%),which contributed less than 1% to the total sample composition at all the other sites.

Figure 5 Mass percent composition of elements in each snow pit except for MSD and MS. (a) Element breakdowns for whole samples, and (b) breakdowns of smallest proportions of elements (0.99 to 1.0). The values reported represent the mean percentage of each fraction from two or three samples

Figure options

The major differences among the sites were the contributions of the trace elements(Figure 5b). The elements V and Cr comprised a higher proportion at QM compared to the other sites. ZD contained the highest diversity of elements,followed by TS and LHG,whereas YL,JMYZ, and DKMD contained the fewest elements. The sites in the southern TP(ZD,DML, and YL) and GL all contained Ni,whereas Ni was not detected at the northern sites in western China and the Himalayas.

Enrichment factors(EFs)are used to estimate whether an element in snow is of anthropogenic or natural origin. This method compares the amount of an element to a reference element assumed to be entirely from crustal sources(e.g.,Al or Fe); here,the reference element is aluminum and an EF is defined as follows:

\[EF({\rm{crust}}) = \frac{{{{(X/{\rm{Al}})}_{{\rm{snow}}}}}}{{{{(X/{\rm{Al}})}_{{\rm{crust}}}}}}\]

where(X/Al)_snow and (X/Al)_crust refer to the ratio of the concentrations of metal X to that of Al in the snow and in the average crustal material(Wedepohl,1995),respectively. An EF value near unity suggests that the source of the element is mostly crustal erosion,whereas elements with values larger than approximately 10 can be attributed to an anthropogenic source.

Here,the EF values were generally less than 5 for Mg,Al,Si,K,Ca,Ti,Mn, and Fe at all of the sites(Table 4),indicating a mainly crustal source for these elements. V was enriched at most of the sites,especially in the TS(10) and JMYZ(37)samples,indicating a significant anthropogenic source, and was relatively less enriched at the inl and sites of the TP,such as LHG(6),DKMD(6), and GL(8), and at the other sites(ZD(7) and DML(8)). Cr was also enriched in the JMYZ(37) and QM(8)samples,followed by TS(7) and LHG(6). Cu was significantly enriched at the TS(53),LHG(35), and ZD(34)sites,which were located close to anthropogenic regions. Ni was enriched at the GL(13),ZD(6),DML(5), and YL(8)sites,all of which were located in or close to the southeastern TP region. Zn was also relatively enriched at most of the sites,especially at TS(13). Overall,the TS,LHG,JMYZ, and QM sites were all located in the peripheral mountains of western China and exhibited the greatest enrichment of these trace metal elements.

The characteristics of dust deposition in western China are helpful for underst and ing the emission,diffusion, and transport of dust sources in central Asia. An underst and ing of these characteristics can also improve the underst and ing of the formation of the Quaternary Loess in the alpine areas of western China(e.g.,the TP)as well as the range of the modern dust source areas(Fang et al., 2004). The widespread snow and ice records from mountain glaciers of western China compensate,to some extent,for the conspicuous lack of modern dust deposition measurements in these remote regions. Our results show a spatial variability of dust deposition in western China,with high values occurring in the northern areas and lower values occurring inl and of the TP and in southern areas of the TP. However,considerable dust deposition has also occurred in the southeastern TP. The concentrations obtained in this study are comparable to,or slightly greater than,those obtained by Wake et al.(1994)at the same sites. For example,Wake et al.(1994)estimated the flux to be 200-600 μg/(cm²·a)at MS,which is slightly less than our calculation(1,032 μg/(cm²·a)). However,the results of these two studies were much more similar at GL(approximately 140 μg/(cm²·a)) and JMYZ and QM(30-300 μg/(cm²·a)). The dust flux derived from two shallow ice cores from the Himalayas also showed a similar result(77-103 μg/(cm²·a))(Wu et al., 2010). In addition,consistent spatial distributions,i.e.,more than a 10-fold dust deposition variation between the northern and southern regions,were all observed in these two datasets. Wang et al.(2006)found that in the past 1,000 years,the dust concentration recorded from the ice cores in the northern and southern TP showed a reverse relationship.

These results suggest that the dust deposition mechanism in the northern and southern regions of the TP are different and are influenced by the different climate conditions. Compared to the results of Wake et al.(1994),our dataset extended the sampling sites to the southeastern TP,the Tienshan Mountains, and the Altai Mountains. Although the sampling period of our study was relatively short(within 2 years vs. 8 years),our analysis methods were more consistent(the same size range for insoluble particle analysis),which would allow for more reasonable comparisons between the different regions. Finally,our dataset also included elemental and mineralogy measurements,which are beneficial for underst and ing the differences in dust content among the different regions of western China.

Compared with other results obtained from snow/glacier records,including those from the Arctic,Greenl and , and Antarctic regions that have dust deposition rates of approximately 1-50 μg/(cm²·a)(Zdanowicz et al., 1998 and references therein),the rate of dust deposition is much greater in western China. Interestingly,the results at Mount Tateyama in central Japan(approximately 770 μg/(cm²·a)from winter to spring)(Osada et al., 2004) and the San Juan Range in Colorado,USA(1,250 μg/(cm²·a))are comparable to the results from the sites in the northern part of western China and the southeastern TP.

Zhang et al.(2001)collected atmospheric aerosol samples at three sites on the TP(Lhasa and Gongga,close to the southeastern TP, and Wudaoliang,close to GLDD) and found similar dust deposition fluxes at these three sites. They estimated the dust deposition flux to be approximately 10,000 μg/(cm²·a),which is much greater than the flux determined in our study. This discrepancy may have resulted from differences in the collection methods,i.e.,differences between the use of passive and active samplers(Lawrence et al., 2010),as well as variations in the elevation of the sampling sites and the contribution of local dust sources. For example,Zhang et al.(1996)estimated that the contribution from local dust sources on the TP can be as high as 75%. Moreover,the online measurements of aerosol at the NPO-Pyramid station also suggested that local sources can be important for coarse particles(Bonasoni et al., 2008). However,our sampling sites were located in the accumulation areas of glaciers(the highest elevation range of glaciers) and at higher elevations,both of which are comparatively less influenced by local dust sources. For example,the elevation of the aerosol sampling site at Lhasa was 3,600 m a.s.l.,whereas the elevation at ZD was approximately 5,800 m a.s.l.,which is approximately 70 km from Lhasa. A comparison with previous dustfall results using the passive method in major cities of western China shows that the dust deposition fluxes on glaciers were also much lower,e.g.,28,450 μg/(cm²·a)at Urumqi,28,140 μg/(cm²·a)at Xining,17,730 μg/(cm²·a)at Lhasa, and 7,670 μg/(cm²·a)at Xi'an(Zhang et al., 2010; Cao et al., 2011). In addition to long-distance transport,the sources of dustfall in cities may also include dust generated from roads and construction,as well as particles from industry,coal burning,vehicle exhaust, and waste incineration(Zhang et al., 2002).

There is great interest in dust sources and dust transport in mountainous areas of western China with respect to paleoclimate studies,which seek to underst and of the dust record of ice cores in western China(Xu et al., 2010). Most studies on snow and glaciers have suggested that dust transport in western China involves a transport process from north to south based on atmospheric circulation and the spatial distribution of dust concentrations(Wake et al., 1994; Xiao et al., 2002; Wu et al., 2010). In recent years,Lagrangian integrated trajectory models(e.g.,HYSPLIT)have been frequently used to assess dust transport in the alpine area of western China,although these models are often limited by the fit of reanalysis data in this area(Ma et al., 2008). Most recently,satellite instruments(e.g.,CALIPSO)have used aerosol properties to retrieve large-scale dust transport data,which have been used to identify the dust transport pathway over arid and semi-arid regions of western China. Based on CALIPSO data,Huang et al.(2007)found that dust storms occur more frequently during summer than was previously believed. Liu et al.(2008),using a full year of CALIPSO data,revealed two major dust transport routes in the TP,occurring within an "airborne dust corridor" that extends from west to east and dominates the northern slope and eastern part of the Plateau; the desert dust particles(originating mostly from northwestern India and Pakistan)were found to be primarily transported along the southern side of the Himalayas.

Our results are consistent with the results of satellite observations,showing that high dust concentrations occur near the northern transport route(near the Taklimakan Desert) and low concentrations occur distant from it(inl and of the TP). For the southern transport route,the high elevation and blocking of dust transport by the Himalayas causes dust to be rarely,if ever,transported over the Himalayas to the TP. The size distribution of insoluble particles in snow samples suggests an important contribution of local/regional dust sources(the high proportion of coarse particles). The Sr-Nd-Pb isotopic results analyzed from the same samples also suggest a local/regional dust source for all the sampling sites(Xu et al., 2012; Yu et al., 2013). For the JMYZ site,only the fine mode of size distribution supported the Sr-Nd-Pb isotopic results,which showed a Sr-Nd-Pb isotopic composition similar to those of the northern sites,suggesting the long-range transport of dust from the northern TP to the Himalayas.

The relatively high dust deposition flux in the southeastern TP and the lack of a significant dust source and transport route,according to the satellite data,suggest that there could be an additional important local dust source for this region. High water equivalent accumulation(precipitation) and ablation rates characterize the glaciers in this area(Yang et al., 2008). The high ablation rates during monsoon periods result in the erosion of debris. These phenomena are easily observed at the DML site,where during monsoon seasons a considerable amount of debris scours the glacier surface with the melt of seasonal snow on the cliff around the glacier,which makes the glacier dirty. Research has shown that the Brahmaputra drainage,including the eastern Himalayan syntaxis,which is one of the most tectonically active regions on Earth,exhibits high rates of physical and chemical weathering(Hren et al., 2007). The relatively larger mode values of size distribution in the ZD and DML sites than in the other snow pit samples(>10 µm vs. <10 µm)suggest an important contribution from a local source. Additionally,the increased fractions of mica and decreased fractions of quartz at the ZD and DML sites further support the lesser influence of desert dust on these two sites. Thus,it is likely that the local environment is the main dust source for the glaciers of the southeastern TP.

Studies of the anthropogenic influence on the remote areas of western China began in the 1990s with the analyses of many types of samples,such as aerosol,water(Hu et al., 1982; Yang et al., 1994), and sediments(Hou et al., 2003; Yang et al., 2010). Generally,these results showed that most of the remote areas in western China are less influenced by anthropogenic pollutants than areas closer to urban areas. Recent studies of the surface water quality in Tibetan rivers have also reported low levels of heavy metals(e.g.,Cu,Co,Cr,Ni,Cd,Pb, and Hg)(Huang et al., 2009). The concentrations of some elements(such as Pb,Mn,As,V, and Cr)in snow collected from Mt. Qomolangma(Everest)are even comparable with the results found in other remote sites(<100 µg/L)(Kang et al., 2007). Many of the observed contaminants,such as Al,Fe,Mg,Mn,Ti, and As,are believed to have been influenced by local sources(Shen et al., 1997; Huang et al., 2008; Sheng et al., 2012). In addition,the complexity of the geology in western China possibly induces variations in certain elements,such as Ni,which has been enriched in the southeastern TP surface soil(Sheng et al., 2012). The enriched mica in the ZD and DML sites observed by XRD could be the product of weathering in the southeastern TP.

The long-distance transport of air pollutants has been frequently observed in the Himalayas because of heavy pollution in South Asia and optimal atmospheric circulation(Hindman and Upadhyay, 2002; Bonasoni et al., 2008). For example,black carbon accumulates on the southern slope of the Himalayas each year during the end of spring and early summer(Yasunari et al., 2010), and significant amounts of heavy metals and organic pollutants have been detected on the glaciers of the Himalayas(Wang et al., 2008; Kaspari et al., 2009).

Our data did not show a pronounced enrichment of heavy elements in the insoluble particle contents at the study sites; however,the enrichment of certain elements,such as V,Cr,Cu, and Zn,were observed on the peripheral mountains of western China. This was also confirmed by lead isotope results in the same snow pit samples(Yu et al., 2013). In general,the anthropogenic sources of these heavy metals are attributed to emissions from fuel oil or coal combustion,vehicle emissions, and resuspended road dust in urban areas(Horvath et al., 1996; Praharaj et al., 2002). Satellite-based measurements have indicated that the upper Ganges Valley in India has some of the highest persistently observed aerosol optical depth values, and that its aerosol layer covers a vast region that extends across the Indo-Gangetic Plain to the Bay of Bengal during the winter and early spring of each year(Di Girolamo et al., 2004). Based on the atmospheric circulation and extensive pollution in southern Asia,the enrichment of the elements Cr,V,Cu, and Zn in the QM and ZD sites can be attributed to long-range transport from this area(Xia et al., 2011). Cong et al.(2007)reported similar conclusions by analysis of aerosol samples at Nam Co Lake(close to the ZD site)using the PIXE and back-trajectory methods. The enriched elements at TS,LHG, and YL were likely transported from nearby cities(Zhao et al., 2008) and are somewhat related to the local development of tourism.

In this study,the dust deposition in snow pit samples collected from glaciers in western China between 2008 and 2010 is characterized. Our data show a distinct spatial distribution of dust deposition,with dust concentrations in the Tienshan Mountains that are nearly 50 times higher than in the Himalayas,which is consistent with the earlier study by Wake et al.(1994)in these regions. The dust deposition in the southeastern TP shows another high flux area that has not been reported previously. Differences in the mineral compositions were observed in the northern and southern parts of western China,with a greater amount of quartz at the northern sites and a more mica at the southeastern TP sites. The insoluble particle sizes in the southern and northern snow pit samples generally exhibited bimodal distributions corresponding to local and regional dust sources. In addition,the enrichment factor for metal elements suggests a primarily crustal source for all of the snow pit samples,whereas a significant influence of anthropogenic activity was detected at the peripheral mountain sites. These characteristics enhance our underst and ing of dust deposition in the mountain areas of western China.

These results may be considered preliminary because these first observational estimates must be substantiated through model simulations at a high spatial resolution. Additional observations of surface fluxes and dust parameters are needed to further evaluate the biogeochemical effects in these remote cryospheric regions of western China.