Introduction

The method of successive snow pits is commonly used in the study of snow metamorphosing processes. In stable-isotopic studies, it is used to examine evaporation (Reference StichlerStichler and others, 2001), melting (Reference SuzukiSuzuki, 1993; Reference Taylor, Feng, Kirchner, Osterhuber, Klaue and RenshawTaylor and others, 2001; Reference Zhou, Nakawo, Hashimoto, Sakai, Narita and IshikawaZhou and others, 2001, Reference Zhou, Nakawo, Hashimoto and Sakai2008b; Reference Unnikrishna, McDonnell and KendallUnnikrishna and others, 2002) and refreezing (Reference Zhou, Nakawo, Hashimoto and SakaiZhou and others, 2008a) effects on the isotopic fractionation of a snowpack. In geochemical studies, this method is adopted to investigate the solute transport mechanism (Reference SuzukiSuzuki, 1982; Reference Harrington and BalesHarrington and Bales, 1998a) and ion eluting behaviors (Brimblecome and others, 1985; Reference LiLi and others, 2006) during snowmelt, and the deposition of atmospheric aerosols in snow (Reference FerrariFerrari and others, 2005; Reference Zhao, Li, Edwards, Wang, Li and ZhuZhao and others, 2006). It has also been used in hydrological and environmental studies of glacier firn at high altitudes (Reference FujitaFujita and others, 2006; Reference Zhou, Nakawo, Sakai, Matsuda, Duan and PuZhou and others, 2007). This method is based on the assumption that a snowpack is horizontally homogeneous on a microscale. However, the reliability of this assumption has never been proven, although great variability of ion concentrations between sites, which were only 1 m apart, was reported (Reference Brimblecombe, Tranter, Abrahams, Blackwood, Davies and VincentBrimblecombe and others, 1985; Reference Tsiouris, Vincent, Davies and BrimblecombeTsiouris and others, 1985). As slight melting could occur in summer, even at the firn surface on top of very high-elevation glaciers, it could introduce some chemical redistributions and subsequent horizontal heterogeneity, and thus leads to a misunderstanding of the snow metamorphosing processes. It could also cause incorrect interpretations of the seasonality of snow-core records, although its impact may be negligible on the ice-core records on the long-term scale. This issue has become increasingly noticeable as global warming has accelerated in recent years (Reference TianTian and others, 2006). Therefore, it is necessary to clarify the extent of spatial isotopic and chemical variability of firn pack on a microscale.

Methods

The fieldwork was carried out at East Rongbuk Glacier on the northern slope of Qomolangma (Mount Everest) in the central Himalaya. On 21 April 2005, a firn pit was dug at a flat site at 6520ma.s.l. (28^˚01^'08^˚ N, 86^˚57^'48⁰⁰ E) in the upper accumulation zone of the glacier. The pit wall was 1.2 ×1.2m² and wholly sampled at intervals of 10cm in a matrix pattern, with a total of 144 samples collected by workers wearing non-particulating suits and plastic gloves. Densities of one firn column were measured. The frozen samples were transported to a freezer at the Institute of Tibetan Plateau Research, Chinese Academy of Sciences (ITPCAS).

All the samples were subjected to isotopic and chemical analyses at ITPCAS. d¹⁸O was measured using a MAT-253 spectrometer (precision ±0.2%), and ion concentrations were studied using ion chromatography. Anions (Cl^–, NO₃ ^–, SO₄ ^2– and PO₄ ^3–) were analyzed by a Dionex ICS 2500 (RFIC) with an AS11-HC column, 500 mL loop and 25 mM KOH eluent. Cations (Na⁺, K⁺, Ca²⁺, Mg²⁺ and NH₄ ⁺) were analyzed by a Dionex ICS 2000 (RFIC) with a CS12A column, 200 mL loop and 20mM methanesulfonic acid (MSA) eluent.

Results

Figure 1 shows the distributions of isotopic and ionic concentrations of the snow-pit wall. Table 1 presents the statistical data of isotopic and chemical concentrations for each (12 samples) of the twelve 10 cm thick snow layers. It includes the lowest, highest and average values as well as the coefficient of variation (CV: standard deviation divided by average value). It is seen from both Figure 1 and Table 1 that both isotopic and ionic concentrations of a layer display differences, but the extents of the differences vary. The averaged CV value of the 12 layers (CVa) for δ¹⁸O is 0.052, although the differences between the lowest and highest δ¹⁸O values of a layer (Dm) are 1.09–4.96%. This indicates that the horizontal isotopic distribution is homogeneous as a whole. However, the CVa values and Dm’s for every one of the nine ions are large. The Dm’s for each ion, except for Ca²⁺ and PO₄ ^3–, generally vary from several tens to several hundreds of mg L^–1. The Dm’s for Ca²⁺, except for one layer, reach 625.04–8244.21 mg L^–1. The Dm’s for PO₄ ^3– are limited to several tens of mg L^–1 because of its low total load in snow. The CVa’s are: Cl^–, 1.097; NO₃ ^–, 0.637; SO₄ ^2–, 0.850; PO₄ ^3–, 0.745; Na⁺, 1.202; K⁺, 1.335; NH₄ ⁺, 0.628; Ca²⁺, 1.385; Mg²⁺, 1.477. These values indicate large horizontal heterogeneity of chemical distributions, although the intensity levels vary for different ions: high for Ca²⁺, Mg²⁺ and K⁺; intermediate for Cl^– and Na⁺; and low for NO₃ ^–, SO₄ ^2–, PO₄ ^3– and NH₄ ⁺.

Fig. 1. Distributions of isotopic and ionic concentrations of the snow pit wall.

Table 1. Statistical data of the snow-pit wall, showing the lowest, highest and averaged isotopic and ionic concentrations of each layer. The coefficient of variation CV is also shown. Each layer includes 12 samples (ionic concentrations: mgL–1)

The isotopic and chemical heterogeneity/homogeneity of a snow layer would also depend on layer thickness, except for the depositional and post-depositional processes. In general, a large thickness corresponds to a low heterogeneity and vice versa. In order to know the thickness-specific isotopic and chemical heterogeneity/homogeneity, the CVs of snow layers of different thicknesses were calculated. This was done by calculating the n-layer (n: 2, 3, 4. . .12) moving mass-weighted averages of isotopic and chemical concentrations using the measured 10 cm thick layer data, including the densities. The calculations started from the snow surface and moved downwards to the bottom layer. Table 2 lists the ranges and averages of the calculated CV values. It is seen from Table 2 that, except for NH₄ ⁺ and on several other occasions (e.g. the 80 cm thick value for K⁺), the CV averages for δ ¹⁸O and all the ions decrease with increasing snow thickness, although the decreasing rates and extents are different. The CV average for δ¹⁸O decreases to 0.016 when the snow thickness is 120 cm. This value falls in the range of those for measurement precision (0.010–0.017) and thus indicates that no more heterogeneity can be detected. However, the ionic concentrations still exhibit large horizontal heterogeneity in the 120cm thick firn pack. The largest CV average is 1.116 for K⁺, and the lowest still reaches 0.183 for SO₄ ^2– at this thickness. Differing from the other ions, the CV average for NH₄ ⁺ increases slightly with increasing thickness between 20 and 100 cm, but all the values are lower than the average of the 10cm thick layers.

Table 2. Statistical data of the snow-pit wall, showing the calculated CV ranges and averages of 20–100 cm snow thicknesses for δ¹⁸O and all the ions. The data of 10cm thick layer from Table 1 are also shown for comparison (see text for details)

Discussion

Larger horizontal isotopic homogeneity corresponds with larger horizontal chemical heterogeneity, indicating that significant chemical redistributions occurred after snow deposition. These redistributions are most likely caused by snowmelt. Meteorological observations (Reference Xie, Ren, Qin and JiangXie and others, 2006) by an automatic weather station at the same site during April to July 2005 show that the daily maximum air temperatures surpassed 0˚C, and some of them approached 10˚C from late May onwards. In fact, even on days when the air temperature remains sub-freezing, subsurface melting can occur because the solar radiation penetrating into the firn pack combined with the low thermal conductivity of snow can lead to a subsurface temperature maximum (Reference Koh and JordanKoh and Jordan, 1995). In the snow pit, four ice layers were observed at depths of 49–50, 53–54, 59–60 and 88–94 cm. These ice layers indicate occurrences of snowmelt, which can introduce large microscale spatial heterogeneity, even if it is of small intensity. This is due to the fractionation process, which tells us that solute is more concentrated in the first meltwaters than in the original parent snow (Reference Johannessen and HenriksenJohannessen and Henriksen, 1978; Reference Goto-Azuma, Nakawo, Hayakawa and GoodrichGoto-Azuma, 1998). It is also due to the preferential water flow, which states that the liquid water in snow is not homogeneously distributed, but in different flow paths or pools (Reference Harrington and BalesHarrington and Bales, 1998b; Reference Feng, Kirchner, Renshaw, Osterhuber, Klaue and TaylorFeng and others, 2001). Hence, when the meltwater is refrozen in the snow, the areas of the flow paths or pools would have very high solute concentrations. Compared to the chemical impact, the effect of melting and refreezing on the isotopic composition of a snowpack is not so obvious (Reference Zhou, Nakawo, Hashimoto and SakaiZhou and others, 2008a, Reference Zhou, Nakawo, Hashimoto and Sakaib), so the microscale heterogeneity is limited.

The ion-differing CV values for a given layer indicate that the extent of horizontal heterogeneity is different for different ions. This could be due to the preferential elution that ions do not fractionate into meltwaters in the same ratios at which they existed in the parent snow, or, in other words, some ions are removed at faster rates from the parent snow than others (Reference Davies, Vincent and BrimblecombeDavies and others, 1982). However, different workers have found different elution sequences (e.g. Brimblecome and others, 1985; Reference LiLi and others, 2006). Since these elution sequences were derived either by comparing the chemical composition of meltwater with that of the parent snow or by the method of successive snow pits (Reference Goto-Azuma, Nakawo, Hayakawa and GoodrichGoto-Azuma, 1998), this study may provide an insight into this problem from another perspective. Because of the large microscale horizontal heterogeneity and its dependence on the snow thickness, as represented by the CV values shown in Table 2, the chemical snowpack observations from only one snow column and at only one given thickness may produce misleading results. This could explain the discrepancies between the elution sequences observed in the different studies.

Since the snow-pit site is at an elevation where the world’s highest glaciers develop, the large spatial chemical variability also suggests that great care must be taken when interpreting the chemical data of ice cores from high mountains. As the water equivalent of the firn-pack thickness of 120cm is 531.5 mm, which is more than the total precipitation of 480.8 mm in the year 1994 at the Pyramid Meteorological Station (5050ma.s.l.) (Reference Stravisi, Verza, Tartari, Baudo, Tartari and MunawarStravisi and others, 1998) in the Khumbu valley, Nepal, a few miles southwest of Qomolangma, the large spatial variability indicates great uncertainties in seasonal and even annual chemical variations. The temporal chemical variations of ice cores on seasonal and annual scales may no longer reflect the atmospheric and environmental records of snow deposition, although they could do so over longer timescales.

Conclusion

Microscale horizontal isotopic distribution in the firn pack is largely homogeneous. In a horizontal distance of 120 cm, the averaged CV value of the twelve 10 cm thick layers for δ¹⁸O is 0.052, and 0.016 in the whole 120 cm firn pit. The value of 0.016 is in the range of analytic precision and thus means complete homogeneity. However, the ionic distribution shows a considerable heterogeneity. The averaged CV values of the 10cm thick layers vary in the range 0.628–1.477, depending on the ions. Based on these CV values, the heterogeneity sequence is Mg²⁺ > Ca²⁺ > K⁺ > Na⁺ > Cl^– > SO₄ ^2– > PO₄ ^3– > NO₃ ^– > NH₄ ⁺. The averaged CV values for all the ions, except for NH₄ ⁺, decrease with increasing snow thickness, although the decreasing rates and extents are different. However, the CV values of different ions are still large and in the range 0.183–1.116 when the snow thickness increases to 120 cm. The heterogeneity sequence becomes K⁺ > Mg²⁺ > Ca²⁺ > NH₄ ⁺ > Na⁺ > Cl^– > PO₄ ^3– > NO₃ ^– > SO₄ ^2–. The CV average change with thickness is different for NH₄ ⁺. As a whole, it increases slightly with increasing thickness between 20 and 120 cm, but all the values are lower than the average of the 10 cm thick layers. Great care must be taken when adopting the method of successive firn pits for temporal studies of snow chemistry or interpreting the chemical data of ice cores from high mountains.