Introduction
Pioneering work on Alpine glaciers between 1967 and 1970 indicated that the isotope content of glacier-river water is subject to significant annual variations and, in summer, to daily fluctuations (Reference BehrensBehrens and others, 1971). Studies in Austria revealed that the deuterium content of various sources of water (snow, ice, ground water) differed (Reference ArnbachArnbach and others, 1976; Reference Stichler and HerrmannStichler and Hermann, 1977). The proportions of these sources present in glacier-river water vary with time and, whilst mixing within a glacier’s drainage systems tends to cause homogenization (Reference ArnbachArnbach and others, 1976), it does not eliminate all systematic variations of isotopic composition. In addition to reflecting the changing influences of diverse water sources, these variations are likely to be influenced by the physical characteristics of the drainage system and the flow-routing through them. Consequently, isotopic analyses of glacier-river water should provide information about the temporal and spatial variability of water storage within and beneath a glacier.
Hydrological investigations at the north Norwegian glacier Austre Okstindbreen between 1976 and 1979 indicated that major differences of water outflow occurred in different summers. It was expected thatthe composition of the water leaving the glacier would be influenced by year-to-year changes of the glacier’s hydrological systems. In both 1976 and 1977, a glacier-dammed lake drained through Austre Okstindbreen during the summer field season (Reference KnudsenKnudsen, 1978; Reference TheakstoneTheakstone, 1978). Because lake outbursts affect glacier-drainage systems, it was anticipated that future events would result in variations of glacier river-water quality. Accordingly, a study of the isotope content of glacier-river water was planned, to supplement discharge, conductivity, and water-chemistry observations. Sampling for subsequent isotope analysis has been undertaken each summer since 1980. The results indicate that the river water is isotopically lighter than most of the water sources at the upper surface of this high-latitude glacier, that it is subject to diurnal variations of isotope content in periods of fine weather, and that temporal patterns of isotope variation may differ from one summer to another.
The Study Area
Austre Okstindbreen, the largest glacier of the Okstindan area, covers about 14 km2 (Fig. 1). Its accumulation area (11.4 km3), although largely below 1500 m, rises to more than 1700 m above sea-level. Over much of the area, the depth of winter snow accumulation exceeds 2.5 m; at the highest parts, it can be more than 7 m (Reference KnudsenKnudsen, 1987). The mean height of the equilibrium line is close to 1200 m.
The glacier terminates in a small lake, at 740 m. In most summers, two rivers issue from the tongue. The discharge of the northern river generally exceeds that of the southern one (Table I). However, in 1978 and 1981 the southern river was the larger throughout the summer, and in 1985, until mid-July, its discharge again was greater than that of the northern one. On 22 July 1985, the southern river ceased to flow during a storm, and the size of the northern river increased considerably. In 1987, water did not issue from the southern side of the glacier.
Continuous recording of glacier-river water-level variations during successive summer field seasons, monitoring of electrical conductivity, dye-tracer testing, observations of the drainage of the glacier-dammed lake, and sampling of river water for chemical and isotopic analyses (Reference TheakstoneTheakstone, 1986; Reference Theakstone and DurbinTheakstone and Durbin, 1986; Reimann, unpublished; Svendsen, unpublished) have provided much information about the glacier’s hydrology. Whilst melt water generated at the surface of the ablation area flows rapidly through the glacier from some points of entry (Reference Knighton and TheakstoneKnighton and Theakstone, 1978), temporary storage of water commonly occurs within the glacier (Reference Theakstone and KnudsenTheakstone and Knudsen, 1981). Discrete glacier-drainage systems exist. These systems may be disrupted before outbursts from the glacier-dammed lake are registered at the portal through which the river emerges (Reference Knudsen and TheakstoneKnudsen and Theakstone, 1988). The isotopic composition of the river water leaving the glacier in summer varies. These variations, which include both long-term and short-term trends, are discussed in this paper.
Access to Austre Okstindbreen during the winter is difficult, and it has not been possible to obtain samples until towards the end of that season. Snow melt is slight until late May, but much melting normally occurs in June, before the start of summer field seasons at the glacier. At the beginning of the melt season, the principal source of river water is snow which has accumulated in the previous winter. Subsequently, as the altitude of the transient equilibrium line increases, melting ice begins to make a significant contribution to discharge. Rainfall has a marked effect on the glacier-river hydrograph in summer. The basin of the glacier-dammed lake, Kalvtjörna, fills as a result of melting of the winter’s snow in May-June. Outbursts from the lake, which have been known to occur in June, July, and August, disrupt the normal pattern of river discharge and have significant consequences for the glacier’s drainage systems (Reference Knudsen and TheakstoneKnudsen and Theakstone, 1988). The ablation season normally ends in late August or early September.
Methods
Water samples were collected from the northern river issuing from the tongue of Austre Okstindbreen. In 1980 and 1981, only a small number of samples was collected by hand. Between 1982 and 1987, samples were collected at pre-set intervals (Table II) with the aid of an automatic liquid sampler. In 1982 and 1985, water samples were also collected by hand from the southern glacier river. Samples were transferred to randomly selected numbered polyethylene vials and transported to the Geophysical Isotope Laboratory of Copenhagen University for analysis. Samples of glacier ice were collected along 1.6 km longitudinal traverses from the snout to the lower part of the ice fall in 1985 and 1987, and a few samples were collected at randomly selected sites in 1986. Samples of old snow were obtained from the walls of pits excavated in the glacier’s accumulation area in 1983, 1986, and 1987; in 1983 and 1987, sampling was extended into firn below the previous summer’s surface. Additional samples of old snow were collected at various sites between c. 650 and 1650 m in 1980 and 1985, and from the accumulation-area surface in 1987. In May 1987, when the entire glacier still was snow-covered, snow samples were collected from pits at 1240 and 800 m. Samples of snow, firn, and ice were allowed to melt in sealed bags before being transferred to vials. Samples of water from supraglacial streams, from melting snow, and from the glacier-dammed lake also were obtained at various times during the summers. A single water sample was collected from the very small stream which was issuing from the glacier on 26 May 1987.
δ18O/16O ratios are presented as 6180 values, expressed in relation to the standard mean ratio of ocean water (SMOW):
where R is the 6180/160 ratio. In 1985 and 1986, all samples were analysed twice. The mean difference of the two δl8O/16O values of the 582 samples was 0.072% (range: 0.00–0.28%).
The δ18O values considered here are summarized in Tables II and III. Although the summary statistics of glacier river-water samples obscure within-summer trends, they are of great utility for between-year comparisons.
The Isotopic Composition Of The Components Of Bulk Discharge Of The Glacier River
The isotopic composition of glacier ice at Austre Okstindbreen shows no significant change from one year to another. Isotopic fractionation is not normally observed during melting or sublimation of glacier ice (Reference Moser, Stichler, Fritz and FontesMoser and Stichler, 1980), and it is not surprising, therefore, that the composition of water in streams flowing over the surface of the lower glacier (“supraglacial streams”, Table III) is almost identical with that of the ice.
Since the nature of the snow or rain falling at the glacier reflects the antecedent meteorological conditions and fractionation processes to which the atmospheric water masses have been subjected (Reference DansgaardDansgaard, 1964), its isotopic composition may differ both within and between events (Reference Gat, Fritz and FontesGat, 1980). The range of the δ8O values of rainfall collected at and close to Austre Okstindbreen at various times between 1980 and 1987 is very large, some samples being richer in the heavier isotope than is glacier ice, others being depleted (Table III). δ18O values of samples of rainfall collected at the field-work base camp (535 m) on five occasions within the period 24–31 July 1987 differed significantly (Fig. 2). Because air temperature tends to decrease with increasing altitude, and heavy isotope depletion is likely to be greater when the temperature at the time of precipitation is lower, the decrease of δ18O values with altitude on 25 July 1987 was not surprising. A similar trend was displayed by samples collected at two of the stations on the previous day, and at altitudes of 740 and 1350 m on 27 July (Fig. 2). Samples obtained at the two lowest stations on 26 July did not conform to this trend (Fig. 2).
Because isotope separation increases with decreasing temperature, isotopically lighter snow may accumulate on the higher parts of the glacier. Reference HambreyHambrey (1974) reported that, at the glacier Charles Rabots Bre1 adjacent to Austre Okstindbreen (Fig. I), the δ18O value of newly fallen snow decreased by 0.6%o per 100 m increase of altitude. However, the altitudinal effect depends on the local climate and topography: initial spatial patterns of isotope distribution are likely to be modified when winds cause drifting. Sampling of snow in late winter has provided some of the lowest δ18O values obtained at Austre Okstindbreen, as well as a few unusually high values (Table III). Natural processes cause changes of isotopic composition of fallen snow. In summer, evaporation and melting, occurring at variable rates, lead to near-surface enrichment (Reference Moser and StichlerMoser and Stichler, 1975), whilst melt-water percolation within accumulated snow, and changes caused by partial recrystallization (Reference BúasonBuason, 1972), result in a tendency towards isotopic homogenization (Reference Amason, Gat and GonfiantiniArnarson, 1981). Overall, the speed and extent of homogenization will depend on factors controlling water percolation. Snow-pit studies indicate that, in the central part of Austre Okstindbreen’s accumulation area, it is almost complete within I year old firn (Reference TheakstoneTheakstone and Durbin1 1986): the range of δ18O values of 17 firn samples between 2.4 and 4.1 m below the 1982 summer surface was very small (Table III). However, in old snow less than 1 year old, isotope variations, probably resulting from discrete episodes of winter accumulation, are characteristic: in summer, the range of the δ18O values is rather similar to that of glacier ice (Table III). Summer sampling of snow surviving from the previous winter at various sites between 600 and 1700 m above sea-level has yielded a relatively large range of δ18O values, as has sampling of melt water leaving snow patches (Table III).
Variations Of Isotopic Composition Of Glacier-River Water: The Influence Of Weather Conditions
In general, ablation-area inputs of water to Austre Okstindbreen have relatively high δ18O values: compared with most of the water sources listed in Table III, bulk glacier-river water (Table II) is δ18O-depleted. Clearly, the tabulated sources do not account directly for the composition of the water discharging from the glacier - either they are modified by fractionation processes or they are supplemented by water sources which have not been sampled. Neither fractionation processes occurring within, or at the base of, the glacier nor water sources there can be monitored with ease.
For much of the ablation season, the base-flow component of glacier-river discharge at Austre Okstindbreen is diluted by a “rapid-transit” component, which varies diurnally. These variations reflect the generation of melt water by heat inputs at the glacier surface, principally below the altitude of the transient equilibrium line. Although solar radiation is the principal energy input responsible for melting of glacier ice, the variability of net radiation is relatively small (Bendixen, unpublished). Because sensible-heat flux and latent-heat flux contributions are temperature-related, there is a close correlation between air temperature and ablation (Reference Braithwaite and OlesenBraithwaite and Olesen, 1984).
During extended rain-free periods, melt water from the surface enters Austre Okstindbreen, passes through it and, together with the base-flow component, emerges at the point of outflow of the glacier river. The time taken in transit by the melt water is reflected in a lag between variations of those components of the surface-energy balance responsible for its generation and temporal variations of the δ18O value of bulk glacier-river water. In both 1983 and 1984, isotope variations of glacier-river water in periods of fine weather lagged 6 h behind those of the air temperature recorded 2 m above the ground surface at the point of emergence of the glacier river (Fig. 3). This lag indicates the mean speed of transit of melt water generated by diurnally varying energy sources. It is an average for the whole of the glacier area from which such melt water is derived, and it is a measure of the organization and storage characteristics of the glacier’s drainage system.
In view of the close relationship between air temperature and the isotopic composition of river water in rain-free periods established in previous years, an attempt was made in the first part of the 1985 field season to assess the value of air temperature as a predictor of river-water composition in differing weather conditions. Temperature was recorded every 10 min, using a data logger adjacent to the water-level recorder at site 2 (Fig. 1), where the river was confined within a bedrock channel. A rating curve could not be established for the site, but records from the previous summer indicated that variations of water level at site 2 relate closely to glacier river-discharge variations (Reference Knudsen and TheakstoneKnudsen and Theakstone, 1988). Whilst the influence of isotopically heavy melting ice on the composition of glacier-river water in fine weather was as apparent in 1985 as in the two preceding summers, marked perturbations resulted when atmospheric conditions changed.
On 9 and 10 July, in fine weather, marked diurnal fluctuations of air temperature were mirrored by the S18O values of water samples, with a lag of about 6 h (Fig. 4). On 1 I July, the lag was shorter and the amplitude of the δ18O cycle reduced markedly, with lower maximum values. On 12 July, with cloudless skies, the temperature rose from a minimum of 6°C at 04.00 h to 13°C at noon. The glacier-river level began to rise quickly at 04.00 h and δ18O values increased from around 10.00 h. During these first 4d of the study, departures from the prevailing fine weather conditions included (A) a brief mid-afternoon cloudy period on 10 July, in which air temperature was depressed, (B) a short-lived, but heavy, fall of convective rain at around 19.00 h on 10 July, (C) some rain during the early morning of 11 July, which ceased at 09.30 h, but caused the glacier-river level to begin to rise several hours before the usual time of occurrence of the discharge minimum, and (D) an abrupt, but rather short-lived, change in the early afternoon of 12 July, during which temperatures dropped as some convective rain fell. Reduced ice melt during event A may account for the fact that the water sample collected at 21.50 h on 10 July was more depleted of 18O than might have been expected. The brief episode B had no apparent major effect on river-water composition, but may have helped maintain discharge at a relatively high level. An anomalous river-water °18O value (−13.45 to (ȭ13.49%) at 08.55 h on 11 July, when rain had been falling for some time, cannot be explained directly by precipitation, since the rain-water samples of event C were isotopically heavy. Rainfall was sampled at the field-work base camp (530 m a.s.l.) at 09.50 h; it had a B18O value of (−8.77 to (−8.79%. Three rain-water samples collected on and close to the glacier (between 750 and 950 m) at 11.00 h had values of (−9.62 to (ȭ9.81%, (−10.15 to (−10.16%, and ȡ10.32 to ȭ10.34%. The release of water from storage within or beneath the glacier, as noted by Van de Reference Griend van de and ArwertGriend and Arwert (1983) during a rain storm at the Neves glacier, Italy, might have been responsible for the low B18O value; enhanced sliding resulting from rain water penetrating to the glacier bed may lead to the tapping of previously isolated cavities. Although the isotope record during the afternoon of 12 July lacks detail, a lower-than-expected value (−12.99 to −13.12%) was recorded for the sample collected at 16.00 h, suggesting that 8180-depleted water was leaving the glacier as a result of event D. This may have reflected a decrease of ice melt, and of its diluting effect on river water.
The weather changed early on 13 July. Rainfall soon after midnight caused a marked increase of river discharge (E, Fig. 5). (Unfortunately, the water-level recorder failed between 06.00 and 12.00 h and, because of an instrument malfunction, only two water samples were obtained between midnight and 10.00 h.) More heavy rainfall in the afternoon, accompanied by a rapid temperature decline (F, Fig. 5), reduced the rate of glacier-ice melt, but maintained river discharge at a high level. The low range of δ18O values between 11.00 and 21.00 h on 13 July suggests that well-mixed water was issuing from the glacier. During the night of 13–14 July, much more 18O-depleted water left Austre Okstindbreen than had been the case earlier in the study period, when discharge was more variable. Throughout the morning of 14 July (G, Fig. 5), the cloud base was below 800 m. The air temperature remained below 7° C for about 24 h, but a rapid improvement of the weather in the afternoon resulted in a sharp rise. Discharge hardly varied throughout the day, and the isotope record resembled that of discharge more closely than that of temperature; although it included a morning minimum and a late-afternoon maximum, the amplitude of the wave was much lower than that of air temperature. The river water was 18O-depleted. Despite reduced incident short-wave radiation and air temperatures, and a consequent reduction of ablation, glacier-river discharge remained high.
General Trends Of Isotopic Composition Of Glacier-River Water, July And August
Compared with water sources within the ablation area and with rain which falls on to the glacier in summer, the Austre Okstindbreen river is 18O-poor (Tables II and III). Diurnal variations of isotopic composition of the river water are superimposed on longer-term trends (Fig. 6). In part, these trends reflect variations of the “base-flow” component of discharge, which may include ground water, water draining from storage zones within the glacier (including water-filled crevasses and cavities), and water from the firn-water aquifer of the accumulation area, supplied by summer-melt processes (Reference Oerter, Baker, Stichler and RauertOerter and others, 1985; Reference Røthlisberger, Lang, Gurnell and ClarkRothlisberger and Lang, 1987). They also are likely to reflect the functioning of the glacier’s drainage systems. Outbursts from the glacier-dammed lake, Kalvtjörna, affect these systems (Reference Knudsen and TheakstoneKnudsen and Theakstone, 1988). The longer-term trends are apparent when the extreme, median, and quartile 8180 values of samples collected over periods of 4 or 5 d are plotted (Figs 7–11).
The annual snow melt in the Okstindan area does not begin until late May or June. When Austre Okstindbreen was visited in May 1986 and May 1987, very little water was issuing from the glacier. The δ18O value of a sample collected from the small stream at the snout on 26 May 1987 was −14.21%, and that of water in a pool at the glacier margin at 760 m was −14.19%. At the beginning of the melt season, the principal source of river water is snow which had accumulated in the previous winter. Snow on the glacier snout on 26 May 1987 had a δ18O value of −14.58%. Three samples were collected from the 1.85 m thick snow cover on the glacier at 800 m; the δ18O values were −15.70% (snow in contact with glacier ice), −13.27% (snow 0.85 m below the surface), and −10.06% (snow at the upper surface). Enrichment of near-surface material in 18O is likely to have resulted from evaporation (Reference Moser, Stichler, Fritz and FontesMoser and Stichler, 1980). As melt water from the surface infiltrated the snow-pack, the latter also became enriched.
As the ablation season proceeds, fractionation effects gradually lead to a gross increase of the δ18O value of the snow cover which remains on the glacier tongue. With the altitude of the transient equilibrium line increasing, melting glacier ice begins to make a significant contribution to discharge. Thus, the δ18O values of the river water discharging from the glacier tend to rise. However, some of the water generated at the surface, both above and below the transient equilibrium line, is delayed in transit for a few hours or days, and some is stored within the glacier for several weeks or more. The nature and extent of storage are likely to be affected by the degree to which the glacier's drainage systems have developed. This itself is influenced markedly by the time of drainage of Kalvtjörna (Reference Knudsen and TheakstoneKnudsen and Theakstone, 1988).
As the summer progresses, the increasing efficiency of the glacier’s drainage systems enables more water to move through them more quickly although, as shown by dye-tracer tests, considerable delays characterize throughflow from many points (Reference Theakstone and KnudsenTheakstone and Knudsen, 1981). As connections develop between conduits and cavities, and local systems coalesce, areas of the bed not previously connected to a drainage system are tapped, permitting the outflow of water previously stored in unconnected cavities. During storms, when much water is entering the glacier from its surface, drainage-system disruption is likely. Towards the end of the summer, as surface melting declines and river discharge decreases, the increased role of the base-flow component is seen in an overall fall of δ18O values.
In 1982, Kalvtjörna drained through Austre Okstindbreen on 19–20 July. Only five samples had been collected from the glacier river for isotope analysis before the event, and none was collected during it. Six samples were collected on 22 and 23 July. The median δ18O value of the 11 samples obtained between 15 and 23 July was −13.00%). That of the 15 samples collected between 24 and 27 July (−13.05%) was little different, but subsequently, as glacier ice was exposed increasingly to ablation, the river water became more enriched in the heavy isotope - median and quartile values became less negative (Fig. 7). On the basis of the somewhat limited sampling in 1982, it was concluded that the increasing contribution made to river discharge by glacier ice was responsible for the trend of δ18O values.
In 1983, the Kalvtjörna basin remained full throughout the period of field observations at Austre Okstindbreen. Heavy rain in the second week of July and at the beginning of August caused high river discharge, and both periods were characterized by elevated #x03B4;18O values (Fig. 8). Generally, however, the trend of #x03B4;18O variations of the river water in 1983 could be related to observed changes of snow and ice melt. In the early summer, dominated by snow melt, δ18O values rose. Subsequently, as the previous winter’s snow disappeared from the lower part of the catchment, values declined, but another rise followed as glacier-ice melt increased. When ablation declined in late summer, #x03B4;18O values fell.
In the 1984 sampling period, an initial slight decline of the median #x03B4;18O value was followed by a marked increase (Fig. 9), a trend similar to that for the same period in 1983. As in the previous summer, the relationship between general discharge and #x03B4;18O levels was apparent. The outburst of Kalvtjörna on 24 July had a significant effect on the #x03B4;18O content of water discharging from the glacier (Reference Knudsen and TheakstoneKnudsen and Theakstone, 1988).
In 1985, Kalvtjörna drained before the start of the field season, and the lake basin remained empty throughout the period of river-water sampling. For the first part of the summer, the northern glacier river was smaller than was the southern one, and δ18O values from the latter (Table IV) were similar to those obtained from the northern river in previous years. The high δ18O values of samples collected from the northern river during the first part of the summer suggested that, unlike the larger river at the other side of the glacier, its discharge was dominated by water generated by relatively local surface melting. On 22 July 1985, during a storm, the southern glacier river ceased to flow. Thereafter, the northern river was the only major outlet of water from Austre Okstindbreen, and the base flow which previously had sustained flow of the southern river contributed to its discharge. In the circumstances, it was not surprising that the summer’s overall pattern of δ18O values differed strikingly from that noted in previous years (Fig. 10). Samples collected on 23 and 24 July yielded the lowest δ18O values for river water in the 7 year period of sampling.
In 1986, the northern river again was the major outlet of water from the glacier. Kalvtjörna drained before the start of the field season. Heavy rainfall on 18 and 19 July resulted in high river discharge (Fig. 11). Sampling was interrupted in mid-July. As in the previous summer, the early-season drainage of Kalvtjörna must have left a major conduit between the lake basin and the glacier terminus, thereby facilitating drainage at a time when, in other circumstances, the systems within Austre Okstindbreen are relatively poorly developed.
In 1987, the glacier-dammed lake outburst occurred on 16 July At the beginning of the field season, when discharge was still influenced by melting of the previous winter’s snow in the lower part of the catchment, δ18O values of glacier-river water were relatively low (Fig. 12). Subsequently, a general pattern of increasing median δ18O values until 20 July, and of decreasing values thereafter, was similar to the discharge trends.
Discussion
Trends of δ18O variation of glacier-river water have been identified at two different scales - the hourly and the seasonal. The systematic patterns, and deviations from them, illuminate aspects of the operation of glacier-drainage systems.
During much of the summer, ice-melt water, relatively rich in 18O, and in transit for only a few hours, is responsible for diurnal variations ofδ18O values in periods of fine weather, as it dilutes the base-flow component of discharge. The 18O-depleted base flow is a mixture of waters with different histories of formation, storage, and transit. Some is generated by accumulation-area melting: modelling (Svendsen, unpublished) and mass-balance studies (Reference KnudsenKnudsen, 1987) suggest that melt water generated above the 1200 m contour may contribute as much as 60–75% of the total flow of the principal Austre Okstindbreen glacier river during the summer season.
During the long winter season at Austre Okstindbreen, ground water in the near-surface recharge zone, and some water in firn, crevasses, and cavities freezes, becoming 18O-depleted. When winter ends and snow-melt water begins to discharge from the glacier, δ18O values are influenced by the isotopic composition of both the water stored within and beneath the glacier and the winter precipitation, which generally has low δ18O values, reflecting the prevailing low winter air temperatures. As surface melting proceeds, melt-water infiltration may cause a degree of 18O-enrichment of the snow-pack. Disappearance of the snow cover from the lower part of the tongue, accompanied by an increasing contribution of ice-melt water to discharge, therefore tends to be accompanied by a rise of δ18O values of glacier-river water. Throughout the summer, however, δ18O variations of bulk glacier water are related to discharge fluctuations, because of the dominance of the base-flow component. Melt water generated in the accumulation area helps maintain base flow of the principal river throughout the summer. Rivers issuing from the glacier to which base flow makes a smaller contribution arc isotopically different to the principal river. It is probable that, in the accumulation area, melt water moves down through the porous firn to an aquifer of seasonally changing thickness, and becomes isotopically depleted as a result of exchange between ice and water (Reference Oerter, Baker, Stichler and RauertOerter and others, 1985).
Drainage-system disruption by early-summer outbursts from Kalvtjörna, results in patterns of δ18O variations of glacier-river water which differ from those detectable in years when outbursts occur later in the season. Since the events leave a large open conduit within, or at the base of, the glacier, water may be transmitted rapidly through the glacier over a much longer distance than is usual in early July. Furthermore, the conduit perimeter is likely to include a considerable proportion of glacier ice. Once the lake outburst has ended, water flowing through the conduit causes ice to melt, adding ice-melt water to the discharge of the glacier river.
Conclusions
In comparison with water entering the glacier as a result of surface melting, that in the principal river discharging from Austre Okstindbreen is 18O-depIeted: few sources with δ18O values as low as those of most river-water samples have been identified. Depletion results from fractionation processes occurring betwen the times of precipitation on to the glacier surface and outflow from the tongue. Run-off from the ablation area is determined by the roles of ice-melt water moving from the surface to the point of river outflow, of water released from storage, of rainfall and of ground-water discharge.
At the beginning of the ablation season, melting of the winter’s snow-pack dominates river discharge, and is normally responsible for a characteristically low δ18O value. Subsequently, an increasing contribution of melting ice causes δ18O values to rise. Ice-melt water is also responsible for diurnal variations of isotopic composition of river water in periods of fine weather, as it moves quickly through the glacier’s drainage systems.
Although the probability that seasonal trends of isotopic composition of glacier-river water will be repeated from one summer to another is high, changes of the glacier’s drainage systems, including those related to outbursts of a glacier-dammed lake, may cause modifications. When two rivers issue from Austre Okstindbreen, the smaller is richer in the heavier isotope, 18O, than is the principal one; apparently, the second river is maintained more by ablation-area surface sources than by the ls18O-depleted water which contributes the substantial base-flow component to the principal river.
Acknowledgements
Isotope analyses of samples collected at Okstindan since 1980 have been undertaken at the Copenhagen University Geophysical Isotope Laboratory, by arrangement with Professor W. Dansgaard and N. Gundestrup; I am very grateful for their contribution to the work of the Okstindan Glacier Project. The Project is a collaborative one, involving personnel from the University of Manchester, England, and the University of Aarhus, Denmark. I am grateful to all who have assisted in the collection of samples and in other field work at Okstindan, particularly Professor J.T. M�ller, Lektor N.T. Knudsen, and post-graduate students 0. Bendixen, C. Durbin, M. Jeffries, P. Reimann, and L. Svendsen. Financial support has been provided by the British Council, the U.K. Natural Environment Research Council, the Royal Society, and the University of Manchester. Cartographic advice and assistance from N. Scarle has been of great value. Comments from anonymous referees on an earlier version did much to improve the quality of this paper.