Introduction

The traditional view that chemical erosion is least intense in glacial environments, founded on the notion that low-temperatures inhibit chemical weathering processes, has been overturned by numerous studies since 1970 (see Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994; Reference Sharp, Tranter, Brown and SkidmoreSharp and others, 1995). The traditional view failed to account for the importance of acid snowmelt, and underplayed the role of moisture availability, the freshly ground character of the rock flour, and geochemically reactive trace minerals. Regarding the extent to which water acts as a reagent, catalyst and carrier in chemical weathering reactions, its abundance should be crucial to the rate of chemical weathering reactions (Reference Reynolds and JohnsonReynolds and Johnson, 1972; Reference LermanLerman, 1979). Therefore, the seasonality of runoff in glaciated catchments is of especial significance, since the rate of dissolution is inversely related to time-dependent solute concentration, and hence the meltwater flushing rate (Reference LermanLerman, 1979; Reference Brown, Tranter and SharpBrown and others, in press).

A major focus of geochemical research in active glaciated regions has been to establish removal rates of major cations in meltwaters and to define chemical denudation rates. Despite temperature-dependence effects on the rate of dissolution (Reference Drever and ZobristDrever and Zobrist, 1992; Reference VelbelVelbel, 1993), such studies indicate rates of chemical erosion in glaciated catchments which are 1.2–2.6 times the continental average (Reference Sharp, Tranter, Brown and SkidmoreSharp and others, 1995). The efficacy of chemical weathering in glacial environments is attributable to high flushing rates, turbulent meltwaters, high suspended-sediment concentrations and the low buffering capacity of dilute meltwaters (Reference Sharp, Tranter, Brown and SkidmoreSharp and others, 1995). However, few studies have attempted to disaggregate the solute load of glacial runoff into crustal, atmospheric and snowpack sources (cf. Reference Sharp, Tranter, Brown and SkidmoreSharp and others, 1995), and to identify variations in solute sources and chemical weathering mechanisms on a seasonal time-scale (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993; Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994).

This paper analyzes the provenance of solute in the proglacial stream draining Haut Glacier d’Arolla during most of the 1989 ablation season, utilising meltwater-quality data derived from twice-daily sampling. Dissolved species are partitioned into components derived from sea salt, acid aerosol, dissolution of atmospheric CO₂, and lithogenic sources, namely carbonates, sulphides and aluminosilicates. Variations in solute provenance are then related to the seasonal evolution of the subglacial hydrological system.

Field Site

Sampling was undertaken at Haut Glacier d’Arolla, the most southerly glacier in the Val d’Hérens, Valais, Switzerland. The glacier has an altitudinal range of ~2560–3500 m, and occupies approximately 6.3 km² of a 12 km² catchment. The maximum length of the glacier is approximately 4.2 km. A number of portals (2–5) contribute meltwater to the bulk runoff (Reference SharpSharp and others, 1993). The bedrock geology is varied and consists of metamorphic and igneous rocks of the Arolla series of the Dent Blanche nappe, the highest tectonic unit of the Valais Alps (Reference Dal Piaz, Vecchi and HunzikerDal Piaz and others, 1977; Reference MazurekMazurek, 1986). Geochemically reactive minerals, such as pyrite and calcite (Table 1), have been identified by microscopy and are present in trace amounts in many of the rocks throughout the catchment (Reference BrownBrown, 1991). In addition, XRD data from the main lithological units indicate that the main feldspars are albite, anorthite, microcline and sanidine, the main olivines are diopside, enstatite and spodumene, Actinolite, muscovite, corderite, hematite, hydrobasaluminite, quartz and talc are also present (personal communication from D. Webb, cited in Reference TranterTranter and others, in press).

Table 1. Mineralogical composition (%) of fine material (<5 mm) from sampling locations A–I on the medial moraines of Haut Glacier d’Arolla (data from Reference BrownBrown, 1991)

Methods and Techniques

Bulk meltwaters were sampled ~100 m from the glacier snout twice daily at 1000 and 1700 h local time (approximating to minimum and maximum diurnal discharge, respectively), from 1 June (Julian day 152) to 31 August 1989 (Julian day 243). The samples were immediately vacuum filtered through 0.45 µm cellulose nitrate membranes, and stored in pre-cleaned plastic bottles. Total alkalinity was determined colorimetrically in a field laboratory to an end-point of pH 4.5 using BDH mixed indicator solution and 1 mmol HCl. Precision was ±2%. The concentration of HCO₃ ⁻ was determined from the total alkalinity by correcting for acid needed to acidify a volume of deionised water equal to the volume of the aliquot and the titre. The major cations (Ca²⁺, Mg²⁺, Na⁺, K⁺) were determined by AAS, using an air-acetylene flame. Spectrochemical buffers, La(NO₃)₃ and CsCl, were added to overcome chemical interferences and ionisation, respectively. Accuracy was ±5%. Major anions (Cl⁻, NO₃, SO₄ ²⁻) were determined by ion chromatography on a Dionex 4000i. Accuracy was ±3%.

Bulk meltwater dissolved-ion concentrations were separated into crustal, sea-salt and snowpack derived components, as detailed in Reference Sharp, Tranter, Brown and SkidmoreSharp and others (1995). All NO₃ and Cl⁻ was assumed to be atmospherically derived, with sea-salt contributions of Ca²⁺, Mg²⁺, Na⁺, K⁺ and SO₄ ² derived from standard sea-water ratios with Cl⁻ (Reference HollandHolland, 1978). Atmospherically derived SO₄ ²⁻ associated with acid sulphate aerosols was calculated from the average Cl⁻: SO₄ ²⁻ ratio measured in the 1992/93 winter snowpack (0.4107). Since the early stages of aluminosilicate dissolution are non-stoichiometric (sec Reference LermanLerman, 1979), atmospheric HCO₃ ⁻ derived from the carbonation of aluminosilicates was estimated from the sum of *Na⁺ and *K⁺ (where * denotes non-sea-salt sodium and potassium) rather than from a ratio with silica (although carbonation of aluminosilicates probably also contributes to the fluxes of Ca²⁺ and Mg²⁺). Therefore, this represents the lower limit for CO₂ drawdown associated with the carbonation of aluminosilicates. Coupled sulphide oxidation and carbonate dissolution (SO/CD) generates all the crustally derived SO₄ ²⁻, yielding one equivalent of HCO₃ and two equivalents of Ca²⁺ + Mg²⁺. Snowpack acidity from acid sulphate and nitrate aerosols was assumed to weather carbonate minerals to produce crustally derived HCO₃ ⁻. The residual HCO₃ was assumed to arise equally from the dissolution and dissociation of atmospheric CO₂ and the carbonation of carbonate minerals, which produced all the crustally derived non-SO/CD Ca²⁺ and Mg²⁺.

Results and Discussion

Figure 1 shows the discharge and suspended-sediment concentration records for much of the 1989 ablation season (Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994). Discharge and discharge amplitude increased as the season progressed, accompanied by an increase in suspended-sediment concentration, reflecting the seasonal evolution of the drainage system (Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994).

Fig. 1. Variability in meltwater discharge and suspended-sediment concentration in bulk runoff from Haul Glacier d’Arolla during the 1989 ablation season.

The subglacial hydrological system of Alpine glaciers appears to be adequately described by two principal flow components (Reference CollinsCollins, 1978; Reference Oerter, Behrens, Hibsch, Rauert and StichlerOerter and others, 1980; Reference Tranter and RaiswellTranter and Raiswell, 1991). Quick-flow waters are derived largely from icemelt, and pass rapidly through a channelised hydrological system of ice-walled channels and major arterial conduits (Reference SharpSharp, 1991; Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994). Conversely, delayed-flow waters are derived largely from snowmelt, and pass more slowly through a distributed hydrological system (e.g. linked, water-filled cavities (Reference WalderWalder, 1986)). These two components mix to form bulk meltwater draining the glacier, with the channelised system expanding headwards at the expense of the distributed system as the melt season progresses (Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994).

Sources of Aqueous Protons

Chemical erosion by Alpine glacial meltwaters is predominantly by the broad class of reactions known as acid hydrolysis (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993). Reference RaiswellRaiswell (1984) has suggested that the anionic content of glacial meltwaters demonstrates the acid source used in the chemical weathering of rock minerals. HCO₃ is the dominant anion where the dissolution and dissociation of atmospheric CO₂ provides protons to fuel chemical erosion (Equations (1) and (2)).

(1)

(2)

Conversely, SO₄ ² is the dominent anion where protons are derived from the oxidation of sulphide minerals (Equation (3)).

(3)

Typically, chemical reactions are coupled, such that solute acquisition in glacial meltwaters involves species in solid, liquid and aqueous phases (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993), e.g.

(4)

Therefore, the proportion of HCO₃ and SO₄ ²⁻ in the bulk meltwaters will reflect the relative dominance of the two major sources of protons driving subglacial chemical erosion (Fig. 2). Expressed as the ratio of HCO₃ ⁻ to (HCO₃ + SO₄ ²⁻), a ratio of 1.0 would signify carbonation reactions involving pure dissolution and acid hydrolysis, consuming protons derived from atmospheric CO₂ (Equations (1) and (2)). Conversely, a ratio of 0.5 suggests coupled reactions involving the weathering of carbonates by protons derived from sulphide oxidation (Equation (4)).

Fig. 2. Variability in the ratio of HCO₃ ⁻/(HCO₃ ⁻ + SO₄ ²⁻) in bulk meltwaters draining Haut Glacier d’Arolla over the 1989 ablation season.

It is clear from Figure 2 that when flow from the distributed system, which underlies the whole glacier in early June (Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994), dominates bulk discharge early in the ablation season, the ratio of HCO₃ ⁻/(HCO₃ ⁻ + SO₄ ²⁻) (hereafter the C-ratio) (~0.65) suggests that coupled reactions, involving carbonate dissolution and protons derived primarily (though not exclusively) from the oxidation of sulphide minerals, largely control bulk meltwater composition. It is unlikely that a system of linked cavities (Reference WalderWalder, 1986) is in intimate contact with a large source of atmospheric CO₂ to enhance the C-ratio to >0.5 (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993). Other possible sources of CO₂(aq) in the distributed system may include HCO₃ ⁻ in meltwaters feeding the distributed system, the release of bubbles trapped in basal ice facies (Reference Tison, Petit, Barnola and MahaneyTison and others, 1993; Reference TranterTranter and others, in press), microbial oxidation of organic carbon (Reference TranterTranter and others, in press) and the oxidation of elemental free carbon. Carbon occurs both as a combined form (mainly as carbonates of Ca²⁺ and Mg²⁺) and as a free element (e.g. diamond and graphite) in addition to that associated with CO₂ (Reference Greenwood and EarnshawGreenwood and Earnshaw, 1984). Graphite is widely distributed, associated with quartz and metamorphosed sedimentary silicate rocks such as mica-schists and gneisses all of which are present in the Haut Glacier d’Arolla catchment (Reference BrownBrown, 1991). Indeed, this recycling of carbon from sediments to fresh waters via rock weathering forms an integral part of the global carbon cycle (e.g. Reference HollandHolland, 1978; Reference Stumm and MorganStumm and Morgan, 1981). The importance of these coupled reactions in controlling meltwater composition in the delayed-flow component is also evident to a lesser extent in the C-ratio (<0.75) during the major recession flow event at the end of July (Julian day 212) (when snow fell on the glacier surface, reducing surficial meltwater inputs to the hydroglacial system, resulting in delayed flow waters forming an increasing proportion of the bulk runoff), and in the diurnal variability from mid-July onwards when the C-ratio is lower at daily minimum discharge. Conversely, when the quick-flow component dominates maximum diurnal bulk meltwater discharge from mid-July onwards (Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994), the C-ratio (0.75–0.9) suggests protons are derived primarily from the dissolution and dissociation of atmospheric CO₂.

Protons to fuel chemical erosion may also be derived from the dissolution of acid sulphate and nitrate aerosols in the seasonal snowpack (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993). Significant contributions from this source of acidity are limited to the early melt season (Fig. 3), as a result of fraction processes within the snowpack (Reference Johannessen and HenriksenJohannessen and Henriksen, 1978) and the up-glacier retreat of the seasonal snow cover as the ablation season proceeds (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993).

Fig. 3. Sources of protons (snowpack (acid nitrate and sulphate aerosols)), SO (sulphide oxidation) and CO₂ (carbonic acid) driving chemical weathering during the 1989 ablation season.

Lithogenic Solute Provenance

Contributions of solute from various rock types are often calculated by assuming that their contribution is proportional to their mass (Reference Garrels and MackenzieGarrels and Mackenzie, 1971). However, minerals weather at different rates, and examples of such weathering sequences are common in the geochemical literature (e.g. Reference GoldichGoldich, 1938; Reference Stumm and MorganStumm and Morgan, 1981). Therefore, the relative abundance and reactivity of available minerals within a catchment will affect the degree and nature of chemical erosion. Using HCO₃ as an indicator of the relative proportions of solute contributed by lithogenic (coupled SO/CD erosion, dissolution of carbonates by acidity derived from the snowpack (SP/CD) and (carbonation of carbonates (C/CD)) and atmospheric (carbonation of aluminosilicates and carbonates) sources (Fig. 4), the calculations detailed above suggest that the weathering of carbonates supplies the vastly greater part of the HCO₃ ⁻ in solution. This is in direct contrast to the actual percentage of carbonate within the bedrock, which from samples of fines (<5 mm diameter) from supraglacial moraines is <12%, and generally <5% (Table 1; Reference BrownBrown, 1991). While the total contribution from carbonate minerals is in excess of 85%, the proportion of solute acquired from carbonation of carbonates and SO/CD reactions, and the weathering mechanisms producing the solute, change as the ablation season progresses. This is clearly illustrated in Figure 4, and will be discussed in more detail in the following paragraphs, where we discuss solute provenance from the major lithogenic sources.

Fig. 4. HCO₃ ⁻ provenance in bulk runoff during the 1989 ablation season. Crustal HCO₃ ⁻ (a) is derived from coupled sulphide oxidation and carbonate dissolution (SO/CD), dissolution of carbonates by snowpach acidity (SP/CD) and carbonation of carbonate minerals (C/CD). Atmospheric HCO₃ ⁻ (b) is derived from the dissolution and dissociation of atmospheric CO₂ associated with the hydrolytic weathering of aluminosilicate and carbonate minerals.

(i) Carbonation of Carbonates

The stoichiometry of Equation (1) suggests that the proportion of the HCO₃ ⁻ load derived from lithogenic and atmospheric sources is equal in equivalence units. If it is assumed that acidity derived from the seasonal snowpack (Fig. 3) is predominantly neutralised by the weathering of carbonates during the early melt season (based on the idea that most solute appears to come from the weathering of lithogenic carbonates; Reference Sharp, Tranter, Brown and SkidmoreSharp and others, 1995), this augments the levels of early-season HCO₃ ⁻ derived from the carbonation of carbonates, resulting in a relatively constant contribution of lithogenic HCO₃ ⁻ throughout the entire ablation season (Fig. 5a). This is in direct contrast to the proportion of HCO₃ ⁻ derived from the dissolution and dissociation of atmospheric CO₂ (Fig. 5b), which shows a steady increase as the ablation season proceeds, reflecting the headwards expansion of the channelised system, the increasing proportion of quick-flow waters in runoff, and increasing suspended-sediment concentrations.

Fig. 5. HCO₃ ⁻ derived from (a) the carbonation of carbonates (C/CD + SP/CD) and (b) the dissolution and dissociation of atmospheric CO₂ during the 1989 ablation season.

(ii) Coupled Sulphide Oxidation and Carbonate Dissolution

Recent models of solute acquisition beneath Alpine glaciers suggest that chemical erosion in the distributed system is dominated by coupled sulphide oxidation and carbonate dissolution (Reference Tranter, Brown, Raiswell, Sharp and GurnellTranter and others, 1993, in press; Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994; Equation (4)). This is confirmed in the provenance calculations, which illustrate a peak in HCO₃ ⁻ (SO/CD), SO₄ ²⁻ (crustal) and Ca²⁺ + Mg²⁺ (SO/CD) during the early part of the ablation season (Julian days 152–165;), when delayed flow waters dominate runoff (Fig. 6). The role of SO/CD reactions in the distributed system is also evident to a lesser extent during the recession flow event at the end of July (Julian days 212–220).

Fig. 6. Bulk meltwater concentrations of (a) HCO₃ ⁻ (SO/CD) (crustal)), (b) SO₄ ²⁻ and (c) Ca²⁺ + Mg²⁺ associated with couplted SO/CD reactions during the 1989 ablution season.

(iii) Carbonation of Aluminosilicates

Relative to the carbonation of carbonate minerals, the weathering of aluminosilicates by carbonation reactions remains relatively constant during the ablation season (Fig. 4b), exhibiting only a slight increase as the melt season progresses. This is illustrated (Fig. 7a) by variations in atmospherically derived HCO₃ ⁻ associated with the carbonation of aluminosilicates. *Na⁺ shows a monotonic increase as the ablation season progresses, associated with increased weathering of plagioclase- and sodic-feldspars (Fig. 7b). Following an early-season peak, *K⁺ also increases, especially during August, associated with increased weathering of biotite, muscovite and potash-feldspar (Fig. 7c). This suggests that part of the increasing proportion of solute derived from carbonation reactions between dilute quick-flow waters and suspended sediment is derived from aluminosilicate minerals in suspension as the ablation season progresses (Fig. 4; Reference Brown, Sharp, Tranter, Gurnell and NienowBrown and others, 1994).

Fig. 7. Bulk meltwater concentrations of (a) HCO₃ ⁻ (atmospheric), (b) *Na⁺ and (c) *K⁺ associated with the carbenation of aluminosilicate minerals during the 1989 ablation season.

Conclusions

To date, few studies have attempted to disaggregate the solute load of glacial runoff into crustal, atmospheric and snowpack sources, and to identify variations in solute sources and weathering mechanisms on a seasonal time-scale. Therefore, snowpack-derived solute is usually incorporated in estimates of chemical denudation rates in glacial catchments (Reference Sharp, Tranter, Brown and SkidmoreSharp and others, 1995). When dissolved species are partitioned into components derived from sea salt, acid aerosol, dissolution of atmospheric CO₂ and lithogenic sources, seasonal changes in solute provenance and in the dominant chemical weathering process are observed.

Trace geochemically reactive minerais in the bedrock (e.g. carbonate and sulphide minerals) appear to contribute the major proportion of solute found in bulk runoff. The major anions are bicarbonate and sulphate, suggesting that two primary sources of protons are utilised to fuel the chemical erosion of glacial sediments: (i) sulphide oxidation, and (ii) the dissolution and dissociation of atmospheric CO₂. Solute contributions from the seasonal snowpack are limited to the early melt season, accompanied by a transient weathering of crustal material associated with snowpack acidity.

Whereas the chemical weathering of aluminosilicate minerals by carbonation reactions supplies a relatively constant proportion of bulk meltwater solutes during the ablation season, the chemical erosion of carbonates shows distinct seasonal variations in the proportion supplied, reflecting changes in the nature of the subglacial drainage system. There are two subglacial environments in which carbonates are chemically eroded, the distributed and channelised drainage systems. Chemical weathering of carbonates in the distributed system is driven by protons derived from the oxidation of sulphide minerals. This peaks early in the ablation season and is relatively constant thereafter, reflecting the reduction in areal extent of the distributed system, and hence the proportion of delayed flow waters in the bulk discharge. In contrast, carbonation of carbonates is the dominant weathering process in quick-flow waters where there is usually free access to atmospheric CO₂. The magnitude of carbonation increases as the channelised drainage system grows throughout the ablation season, accompanied by an increase in the proportion of quick-flow waters and suspended-sediment concentrations in bulk runoff. In addition, carbonation of the aluminosilicate fraction of the suspended-sediment load also increases, reflected in the proportion of atmospheric HCO₃ ⁻ derived from the carbonation of aluminosilicates, *Na⁺ and *K⁺. Therefore, the seasonal evolution of the subglacial drainage system plays an important role in determining both the source of solute and the weathering mechanism driving chemical erosion.

The chemical weathering of carbonates is enhanced during the early melt season by acidic snowmelt. The decrease in this mechanism of chemical weathering through the ablation season is offset by the increase in chemical weathering in the channelised drainage system driven by carbonation reactions. We conclude that subglacial drainage structure and the relative abundance and reactivity of available minerals within the catchment are key controls on the extent of subglacial chemical weathering.