IntroductionHigh alpine catchments in heterogeneous fractured

crystalline rocks present a great challenge when assessingthe large-scale ground water flow systems within these for-mations. Regional flow systems in mountainous terrain havebeen theoretically addressed by Tóth (1963, 1984), Forsterand Smith (1988a, 1988b), and Zijl (l999), while the impactof fault zones on such systems was studied by López andSmith (1995). Among others, studies in crystalline massifs

of Switzerland and France show the influence of the degreeof fracturations and their orientation on preferential flow-paths on the massif scale (Lhomme et al. 1996; Pistre 1993).Crucial factors for understanding the nature of the flow sys-tems are the analysis of the origin of encountered groundwater and especially the recharge conditions in these topo-graphically complex terrains. Systematic studies of thechemical composition of ground water encountered in crys-talline formations of the Swiss Alps show how these para-meters can assist in delineating the origin of the encounteredwaters (Dubois 1993). Recharge conditions in these highmountainous areas are strongly influenced by seasonal vari-ations (Kattelmann and Elder 1991). Accumulated winterprecipitation in the snowpack and meltwater from glaciatedareas both contribute to ground water recharge (Martinec etal. 1982; Forster and Smith 1988a). The influence of the sur-face hydrology and, in particular, that of snowmelt andglaciation on shallow ground water systems has been inves-tigated and a characteristic signature of this recharge sourcewas shown in terms of environmental isotope content and

AbstractTo assess the contribution of accumulated winter precipitation and glacial meltwater to the recharge of deep

ground water flow systems in fracture crystalline rocks, measurements of environmental isotope ratios, hydrochemi-cal composition, and in situ parameters of ground water were performed in a deep tunnel. The measurements demon-strate the significance of these ground water recharge components for deep ground water flow systems in fracturedgranites of a high alpine catchment in the Central Alps, Switzerland. Hydrochemical and in situ parameters, as wellas δ18O in ground water samples collected in the tunnel, show only small temporal variations. The precipitation recordof δ18O shows seasonal variations of ~14‰ and a decrease of 0.23‰ ± 0.03‰ per 100 m elevation gain. δ2H andδ18O in precipitation are well correlated and plot close to the meteoric water line, as well as δ2H and δ18O in groundwater samples, reflecting the meteoric origin of the latter. The depletion of 18O in ground water compared to 18O con-tent in precipitation during the ground water recharge period indicates significant contributions from accumulateddepleted winter precipitation to ground water recharge. The hydrochemical composition of the encountered groundwater, Na-Ca-HCO3-SO4(-F), reflects an evolution of the ground water along the flowpath through the granite body.Observed tritium concentrations in ground water range from 2.6 to 16.6 TU, with the lowest values associated witha local negative temperature anomaly and anomalous depleted 18O in ground water. This demonstrates the effect oflocal ground water recharge from meltwater of submodern glacial ice. Such localized recharge from glaciated areasoccurs along preferential flowpaths within the granite body that are mainly controlled by observed hydraulic activeshear fractures and cataclastic faults.

868

Environmental Isotopes as Indicators forGround Water Recharge to Fractured Graniteby U.S. Ofterdinger1, W. Balderer2, S. Loew2, and P. Renard3

1URS Ireland Ltd., Iveagh Court, 6–8 Harcourt Rd., Dublin 2, Ire-land; +353–1–4155100; fax +353–1–4155101; ulrich_ofterdinger@urscorp.com

2ETH Zurich, Engineering Geology, ETH-Hoenggerberg,CH–8093 Zurich, Switzerland

3University of Neuchâtel, Centre of Hydrogeology, 11 RueEmile Argand, CH–2007 Neuchâtel, Switzerland

Received May 2002, accepted February 2004.Copyright © 2004 by the National Ground Water Association.

Vol. 42, No. 6—GROUND WATER—November–December 2004 (pages 868–879)

hydrodynamic response (Abbott et al. 2000; Flerchinger etal. 1992; Ward et al. 1999). However, these investigationsare mainly restricted to shallow ground water systems in thedecompressed zone of the basement rock (Cruchet 1985)with near surface weathering and stress relief.

To investigate the effects of the specific recharge con-ditions in high alpine regions on the deeper flow systemswithin fractured crystalline rocks, a high relief catchment(total area ~40 km2) in the western Gotthard Massif waschosen for this study. Results from a hydrological model(PREVAV-ETH) applied to the research area provide esti-mates of spatially and temporally distributed ground waterrecharge rates (Vitvar and Gurtz 1999), indicating highrecharge rates in the glaciated areas during ablation peri-ods. Even though only a small percentage (~14.6%) of thetotal catchment is glaciated, significant effects on theground water flow regime are anticipated, particularlyregarding the flow to a deep and unlined tunnel (BedrettoTunnel), which partly passes beneath the research area andits glaciated regions. This tunnel was used for direct sub-surface observations and monitoring of ground water flowand sampling. Analysis of the hydrochemical composition,as well as the environmental isotope content, was used toestablish the relationship between surface waters, springdischarges, and deeper ground water. Interpretation of thedata was furthermore based on the analysis of the structuralanisotropy and heterogeneity of the host rock due to frac-turation and faulting.

Geological SettingThe research area is situated in the western Gotthard

Massif of the Swiss Alps. Geologically, it consists mainlyof the late Hercynian Rotondo Granite, ~220 Ma (Jäger andNiggli 1964) and orthogneiss of early Palaeozoic age,~420 Ma (Arnold 1970). Topographic elevation of thegranite body ranges from 1800 to 3200 m a.s.l. (Figure 1).

On the northern margin of the study area, the FurkaBasetunnel passes through the granite body at an elevationof ~1490 m a.s.l. An abandoned and unlined support seg-ment to this basetunnel, the Bedretto Tunnel is furthermorepassing through the granite body in a northwest-southeastdirection. Due to the partial collapse of the tunnel, only thenorthern ~1.5 km are still accessible. Figure 1 shows thatthis latter tunnel segment is also passing beneath glaciatedareas of the study domain. It offers the opportunity fordirect subsurface observations, and was used for samplingof ground water, monitoring of discharges, and in situ para-meters. The granite body shows varying degrees of frac-turation and faulting throughout the research domain,changing both in lateral and vertical directions. Data onfracture orientation and fracture frequency were gatheredby means of scanline surveys (Priest 1993) on surface out-crops and along the unlined tunnel profile of the BedrettoTunnel. From this data, four major fracture sets can bededuced (Table 1), where, especially, sets 1 and 2 are dom-inant at the terrain surface. The northwest-southeast strik-ing fracture sets (sets 3 and 4) are more abundant along thetunnel profile and are believed to be influenced by the tun-nel construction, i.e., induced or reactivated by local stressamplifications.

In addition to these steeply dipping sets, further shal-low dipping fractures can be identified at the terrain sur-face. These latter fractures generally strike subparallel tothe valleys with dip directions toward the valley bottom,most probably representing quaternary stress release joints.Occasionally, concentrical jointing can also be observedwithin the granite at the terrain surface. In addition to thefracturation, steeply dipping fault zones can be identified asimportant larger features, both at the surface and along thetunnel profile. They are mostly orientated subparallel or atacute angles to the general trend of the alpine structures ofthe Gotthard Massif (east northeast–west southwest). Oneof the major fault zones intersecting the Bedretto Tunnel isindicated in Figure 1. Both preferred orientations of frac-turation (strike east northeast–west southwest) and, on alarger scale, the orientation of fault zones cause a signifi-cant structural anisotropy within the granite body.

869U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879

G

T1

T2T3

T4 T5

T6 T7

T11 S1

T9S2

3070

2710

3039

2944

3062

2913

3082

3192

3068

3023

1778

FURKA-BASETUNNEL

BEDRETTO-TUNNEL

3026

γR

γR

γR

γR

G

G

G

am

am

am

γR

γR

Saasb

a ch

(bro

ok)

Ger enwass er (bro

ok)

27600 1000 2000 meters

γR

Gam

3070Geological boundary

Rotondo-Granite

OrthogneissAmphibolite

Topographic reference withelevation in meters a.s.l.

Glaciated areaSpring discharge

elevation contour interval = 35 meter

Major fault zone (partly inferred)

gg

wg

ro

le

wggg

Witenwasseren-glacierGeren-glacier

lero

LeckihornRotondo

Figure 1. Enlarged plan view of the research area. Indicatedare the projected tunnel traces together with the location ofthe subsurface observation points, geological boundaries. andthe trace of the major fault zone intersecting the BedrettoTunnel. Locations of spring discharges and glaciated areasare also marked.

Table 1Mean Orientation of Fracture Sets and

Corresponding Mean Normal Fracture Frequencyl (Priest 1993) from Surface and Tunnel Outcrops

MeanSet Orientation λsurface λtunnelNumber (strike/dip) (m–1) (m–1)

1 049/75 SE 2.1 0.52 080/83 SE 1.5 0.23 140/86 SW 0.7 2.64 170/79 SW 0.04 0.3

HydrochemistrySamples of the encountered ground water were col-

lected from spring discharges at the terrain surface of thegranite body, as well as from inflows to the Bedretto Tun-nel. Their major ionic composition was analyzed by ion-exchange liquid chromatography, while parameters such aspH, temperature, conductivity, alkalinity, and oxygen con-tent were measured in situ during sampling with electronicprobes or field titration, respectively. Saturation indices (SI= log IAP/KT) for several minerals where calculated withWATEQ4F (Ball and Nordstrom1992).

Spring DischargesFew spring discharges from the granite body can be

identified in the research area. They are mostly associatedwith zones of denser fracturation with northeast-southweststriking, steeply dipping fractures (Figure 1). Table 2 givesan overview of the average in situ parameters measured atthese spring discharges during three successive summerfield campaigns (1997–1999). In terms of decreasingmEq% of cations and anions, respectively, the encounteredwaters are characterized as dilute, slightly alkaline Ca-Na-HCO3(-SO4) waters. Their mean Ca/Na ratio, deducedfrom mEq/L concentrations, lies at 2.58 (σ = 0.9). Allencountered waters are undersaturated with regard to cal-cium-bearing minerals such as calcite or sulfate mineralssuch as gypsum or anhydrite (mean value for SICalcite= 3.417, σ = 0.122; SIGypsum = –4.793, σ = 0.151; SIAnhydrite= –4.541, σ = 0.151). Spring discharge from the northernorthogneiss shows a different hydrochemical compositionand can be characterized as Ca-HCO3-(SO4) waters with amean Ca/Na ratio of 8.93 (σ = 1.26). Mean in situ param-eters are stated in Table 2. Compared to the spring watersencountered from the granite body, the discharge from theorthogneiss is less, but still undersaturated with regard tocalcium-bearing minerals or sulfate minerals (mean valuefor SICalcite = –1.756, σ = 0.276; SIGypsum = –3.691,σ = 0.046; SIAnhydrite = –3.948, σ = 0.045).

Tunnel InflowsAlong the ~1.5 km accessible section of the unlined

Bedretto Tunnel, 11 tunnel inflows were monitored andsampled (Figure 1). Hydrochemical sampling was carriedout from September 1998 until November 1999 (15 months)on a monthly or bimonthly basis, while in situ parameterswere measured from September 1998 until August 2000.Ground water samples were collected directly at the rock-face from discrete features such as fractures and fault zones.Measurements of discharge rates at the individual samplinglocations were carried out manually with calibrated vessels.For measurements of integral inflow rates along particulartunnel sections, depth-integrated flow velocity measure-ments with an anemometer were carried out along crosssections in the basal drain of the tunnel, which cumula-tively captures the tunnel inflows. Discharge was then cal-culated according to ISO748:1997(E) (ISO 1997). At thespecific observation points S1 and S2 (Figure 1), triangu-lar-notch thin-plate weirs, in accordance withISO1438/1:1980(E) (ISO 1980), equipped with float levelgauges, were installed for continuous measurement of thecollected discharge at these locations. Measurements were

recorded from January 1999 until August 2000. However,the records prior to mid-June 1999 were partly obscured bytechnical problems and damages caused by rock bursts inthe gallery. Discharge rates were calculated from therecords applying the Kindsvater-Shen formula (ISO 1980).

Similar to the surface observations, inflows are mainlyassociated with northeast-southwest striking fracturesand/or fault zones. This indicates that the observed domi-nant fracture orientations not only cause a strong structuralanisotropy within the granite body, but are furthermorecausing a hydraulic anisotropy and heterogeneity favoringpreferential flow in the plane of these features. The obser-vation points T9 and S1 are associated with a major steeplydipping fault zone, striking ~75° (Figure 1). This ~120 mwide zone consists of several discrete cataclastic faults,partly with clay mineral infilling (gouge), and neighboringzones of dense fracturation. Along this zone, the most sig-nificant inflows to the tunnel occur. Discrete inflows in thiszone are abundant, but often obscured by tunnel supportsystems.

The discharge rates at the individual sampling loca-tions along the tunnel profile range from drop-inflows withmean rates of 0.56 to 0.85 L/min to more significantinflows with mean discharge rates up to 33.8 L/min. Eventhough small variations from the mean value of discharge

U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879870

Table 2Mean In Situ Parameters

Temper-Q ature � TDS O2

Location (L/min) (°C) (µS/cm) pH (meq/L) (mg/L)

T1 0.6 19.3 83.1 8.9 1.64 2.7(0.11) (0.2) (1.4) (0.2) (0.07) (0.2)

T2 26.9 19.7 88.1 9.0 1.73 2.3(1.22) (0.1) (1.4) (0.2) (0.07) (0.2)

T3 0.6 19.9 71.8 9.1 1.37 3.0(0.03) (0.2) (1.6) (0.1) (0.10) (0.1)

T4 0.8 19.3 94.0 8.9 1.89 1.1(0.07) (0.2) (1.6) (0.1) (0.12) (0.3)

T5 2.1 19.7 89.1 9.1 1.71 2.4(0.23) (0.1) (1.5) (0.1) (0.10) (0.3)

T6 1.1 19.6 89.4 9.1 1.74 1.9(0.10) (0.1) (1.4) (0.1) (0.05) (0.2)

T7 6.6 19.0 104.0 9.1 1.99 1.8(0.85) (0.1) (1.7) (0.1) (0.14) (0.3)

T9 10.3 15.4 91.5 9.2 1.76 1.2(0.65) (0.27) (1.9) (0.1) (0.12) (0.2)

T11 3.0 19.6 115.1 9.3 2.14 0.3(0.27) (0.05) (1.7) (0.1) (0.14) (0.1)

S1 18.4 17.7 105.5 9.2 2.10 2.0(0.47) (0.1) (2.2) (0.1) (0.10) (0.2)

S2 30.5 18.9 79.2 8.9 1.54 2.9(1.91) (0.04) (1.4) (0.1) (0.10) (0.3)

Q (�R) 7.9 16.0 7.2 0.35 8.1(1.6) (1.5) (0.3) (0.05) (0.8)

Q (G) 6.5 64.7 7.3 1.27 6.7(0.6) (2.5) (0.3) (0.05) (0.2)

Mean in situ parameters (discharge Q, temperature, conductivity �, pH, totaldissolved solids [TDS] and dissolved oxygen [O2] at individual sampling loca-tions along the tunnel, as well as mean in situ parameters for spring dischargesfrom the Rotondo Granite (γR) and the neighboring orthogneis (G) (Figure 1).Standard deviation in parentheses as indicator for variability over time.

over time can be observed in the manual measurements atthese locations, the automated stations at S1 and S2 showno significant small-scale temporal variations of discharge.Figure 2 depicts the discharge rates at these stations for therecent period of 1 yr together with the daily precipitation atthe meteorological station Grimsel/Hospiz (1980 m a.s.l.),located in the northern vicinity of the research area. Bothdischarge records show a steadily increasing trend, but nodirect response to recent precipitation in this period. Fluc-tuations are only minor with respect to the registration res-olution of the gauges at S1 and S2 (0.7 and 0.9 L/min,respectively). However, the small, but consistent, increaseof discharge might be linked to the large-scale annual fluc-tuations of total precipitation. Figure 3 shows the annualprecipitation at Grimsel/Hospiz for the period from July1990 to July 2000. It is apparent that the past 3 yr have beencharacterized by high total precipitation and, thus, the dis-charge rates at the stations might be showing a delayedresponse to these long-term variations in precipitation.

Total discharge along the whole ~1.5 km tunnel sec-tion amounts to a mean value of 890 L/min. The variationsover time around this average value are again small(σ = 18 L/min) and are within the range of the potentialmeasurement error (± 24 L/min).

Table 2 gives an overview of the manually measuredin situ parameters for the individual observation pointsalong the Bedretto Tunnel. The temperature measurementsshow only little variation over time at the individual sam-pling locations and along the tunnel profile, except for asignificant negative temperature anomaly at locations T9and, to a lesser degree, S1. Temporal variability at T9 is thehighest (σ = 0.27°C). These two sampling locations areassociated with the large fault zone encountered along thegallery (Figure 1), where ground water inflow to the tunneloccurs at high discharge rates (along the whole fault zonesection ~450 L/min). Figure 4 illustrates this anomalybelow the glaciated area and indicates the potential contri-bution of cold glacial waters to the ground water recharge.

The hydrochemical composition of the ground waterencountered along the tunnel profile can mainly bedescribed as dilute, moderately alkaline (pH 8.9 to 9.3)(Table 2) Na-Ca-HCO3-SO4(-F) waters with a mean Ca/Naratio of 0.64 (σ = 0.21, deduced from mEq/L concentra-tions).

Saturation indices show undersaturation with regardto calcite or sulfate minerals as well as for fluorite for allsamples taken from the tunnel inflows (mean value forSICalcite = –0.184, σ = 0.051; SIGypsum = –3.284, σ = 0.159;SIAnhydrite = –3.535, σ = 0.156; SIFluorite = –0.729,σ = 0.205). These waters have most probably evolved fromthe ground water composition as measured at the surfacedischarges mainly through the process of feldspar (plagio-clase) hydrolysis along the flowpaths through the granitebody. Petrographic studies of the Rotondo Granite (Hafner1958) state a composition of the rock matrix of 25% to 35%quartz, 20% to 40% potassium feldspar, 20% to 35% albite,2% to 7% anorthite, and 5% to 10% biotite with minoramounts of garnet, chlorite, epidote, and apatite. Local sig-nificant deposits of pyrite have been reported during the tun-nel construction. During silicate weathering, sodium isreleased from the albite-rich granite. This process results inan alkalinity increase and calcium is most probably lost tothe solid phase by ion-exchange processes on clay mineralscontained in fillings and linings of fractures and/or faultzones. Precipitation of calcite is unlikely to occur as bothspring samples and tunnel inflows are undersaturated withrespect to calcite.

At the northern end of the Bedretto Tunnel, Ca-Na-HCO3-SO4(-F) waters with a mean Ca/Na ratio of 1.19(σ = 0.07) are encountered. Several factors contribute tothis changed Ca/Na ratio. Hafner (1958) reports a decreas-ing albite content within the granite composition toward themargins of the intrusive body. Therefore, these waters closeto the northern boundary of the granite body may reflectthis changing geochemical composition of the host rock.Additionally, the degree of fracturation and especially thenumber of large fault zones with significant clay infilling(gouge) is smaller in this part of the tunnel. Less abundanceof these features would imply less potential for ion-exchange with clay minerals to remove calcium from

871U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879

Jul99 Oct99 Jan00 Apr00 Jul005

10

15

20

25

30

35

40

45

Jul99 Oct99 Jan00 Apr00 Jul000

10

20

30

40Daily precipitaion atGrimsel/Hospiz

Date [mmyy]

Date [mmyy]

Pre

cipi

tatio

n [

mm

]D

isch

arge

[l/

min

]

S1

S2

Figure 2. Record of automated discharge measurements S1and S2 in the Bedretto Tunnel together with daily precipita-tion recorded at the meteorological station Grimsel/Hospiz(1980 m a.S.l).

0

500

1000

1500

2000

2500

3000

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Year

Pre

cipi

tatio

n [m

m]

Figure 3. Record of annual precipitation (1990–2000 summedfor period July–July) at the meteorological station Grim-sel/Hospiz (1980 m a.s.l.).

solution. Variations in the relative abundance of calcitewithin the granite body may also influence the range ofobserved Ca/Na ratio. Finally, influences from potentiallycalcium-richer ground waters originating in the neighbor-ing orthogneiss and, in particular, in the amphibolitewedges within the gneiss (Figure 1) are other potential cal-cium sources as indicated by the analysis of spring dis-charges. Water from the amphibolite wedges could not besampled; however, studies in the central Swiss Alps showthat similar calcium-rich compositions can be expected tooriginate from the amphibolites (Marechal 1998). Becauseof the observed strong anisotropy of the gneiss, withsteeply dipping schistosities, mainly aligned subparallel tothe main orientation of the Gotthard Massif (east north-east–west southwest), a pronounced hydraulic anisotropywithin these rocks seems plausible. This would imply pref-erential flowpaths along the strike of the schistosity planes

and, together with the preferred orientation of observedhydraulic active fractures and fault zones within the granitebody (east northeast–west southwest), potential contribu-tion from these neighboring formations would most proba-bly originate in the northeast direction of the granite body(Leckihorn area) (Figure 1).

Environmental IsotopesTo investigate potential sources of ground water

recharge and flowpaths, as well as the age of the encoun-tered waters, the contents in environmental isotopes ofwater samples from tunnel inflows were measured, i.e.,contents of 18O, 2H, and tritium (3H). Additionally, 18O and2H content in precipitation were analyzed. 18O and 2H con-tent, the latter after zinc reduction to H2, in ground waterand precipitation samples were determined by isotope ratiomass spectrometry. Tritium values of ground water sampleswere measured by means of an enriched liquid scintillationcounter. Isotope composition is reported as d value of theheavy isotope = (RSample/RStandard –1) × 1000 in units of ‰,where R = (heavy isotope/light isotope). Analytical errorsin the previous δ18O and δ2H determination are ± 0.1‰.Tritium determinations are associated with an error of ± 0.8TU (all expressed as 2σ).

The conventional standard for δ2H and δ18O values isVienna Standard Mean Ocean Water. Tritium values (3H) arereported as absolute concentrations in TU (1 TU correspondsto 1 tritium atom per 1018 hydrogen atoms or 0.118 Bq/L).

All isotope measurements were carried out at the GSFInstitute of Hydrology in Neuherberg, Germany.

TritiumFor the analysis of the tritium content of the tunnel

inflows, two sampling series were carried out and analyzed(December 1998 and March 1999). Differences in the tri-tium content at the individual sampling points between thetwo series do not exceed the accuracy of the measurementof ± 0.8 TU. The mean tritium concentrations observed atthe sampling points along the tunnel profile (Figure 4)show values of 7.1 to 15.1 TU at the northwestern end ofthe tunnel. Values of 2.8 to 3.5 TU are observed toward thesoutheastern end of the profile below the glacier. At thesoutheastern end, the mean tritium content lies at 10.8 TU.

Qualitatively, for continental regions, tritium values of5 to 15 TU are attributed to recent recharge (< 5 to 10 yr)while tritium contents of 0.8 to ~4 TU would represent amixture between submodern (recharged prior to 1952) andrecent recharge (Clark and Fritz 1997). Two possible sce-narios might elucidate the presence of the low tritium val-ues and the origin of the submodern component. Thesampling sites showing these low tritium concentrations(T9, T11, and S1) are situated below the valley bottom andmight be influenced by upwelling of old ground water fromwithin the granite body. A plot of tritium content vs.sodium content (Figure 5) shows the negative correlation oftritium values with sodium content. This correlation is typ-ically observed with the presence of older ground watercomponents characterized by higher mineralization, i.e.,sodium content. However, the sodium concentration of thelow tritium samples is still small (< 0.7 meq/L). Other

U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879872

1480

2400

2600

2800

3000

3200 0 500 m

T9 T11 T2S1Tunnel collapsed

Alti

tude

[m

a.s

.l]

SE NW

Glacier

sampled inflow

b

a

c

0

4

8

12

16

Triti

um

[TU

] δ

18O

[ %

]

-15.2

-15.0

-14.8

-14.6

12.5

15.0

17.5

20.0

Tem

pera

ture

[o C

] 22.5

Figure 4. Cross section along the profile of the Bedretto Tun-nel with locations of sampled inflows: (a) mean temperature,(b) mean tritium concentration, and (c) mean �18O content.Analytical error is indicated.

studies in the central Swiss Alps show that for a causal rela-tionship between sodium and tritium content with respectto a source of submodern ground water, sodium concentra-tions of at least > 1 meq/L are common (Marechal 1998).With regard to the petrographic analysis of the RotondoGranite (Hafner 1958), the resulting correlation betweenlow tritium values and small, but increased, sodium con-centrations might merely be caused by coincidence with theobserved spatial variation of albite content of the rock and,thus, the increased availability of sodium to go into solutionaway from the margins of the granite body. As the tunnelprofile represents a transect from the margin of the bodytoward its center, this trend is reproduced in the water com-position sampled at the specific locations.

Another scenario with regard to the influence of theglaciated areas above the observed inflows to the tunnel isalso plausible. Meltwater from the glaciated areas isderived from a distinct reservoir, which preferentially accu-mulates winter precipitation with low tritium contentsaccording to the observed seasonal variation in precipita-tion. However, the observed mean tritium concentrations inrecent winter precipitation range from 8.4 to 15.8 TU(observations during the 1990s at meteorological stationsGuttannen, Grimsel, and Meiringen in the northern vicinityof the research area). Therefore, ground water infiltrationfrom this accumulated recent winter precipitation can onlyexplain the observed data with lower values down to 2.8TU, if the residence time of this water is assumed to exceedthe half-life cycle of tritium (12.43 yr).

Another variant of this latter scenario is regarded to bemost plausible. Assuming that not only recent accumulatedwinter precipitation with seasonal low tritium concentra-tions contributes to the ground water recharge, but alsomeltwater from old submodern glacial ice, a source ofground water recharge is present, which can explain themeasured low tritium content. This source is tapped by thelarge fault zone associated with the observation points T9and S1. On a plan view projection (Figure 1), it can be

shown that the inferred trace of this structure passesbeneath a large portion of the glaciated region that reachesup to altitudes of 3000 m a.s.l. Ground water rechargedalong the strike of this structure is anticipated to move bytopography-driven flow and enhanced by the drainage tothe subsurface gallery along this preferential flowpathtoward the tunnel, thereby preserving the characteristic lowtritium values of recharge from the glacial meltwater of thesubglacial drainage system (Tranter et al. 1996). A similarinfluence of meltwater from tritium-free glacial ice on theoverall tritium content of glacial discharge is also reportedby Ambach et al. (1971) and, for the Grimsel area in thenorthern vicinity of the study area, by Keppler (1996).

The tritium values of 7.1 to 15.1 TU are in the range ofthe tritium concentrations in recent winter precipitation.For comparison, the tritium concentrations from recentsummer precipitation are generally higher and range from14.7 to 25.3 TU (observations during the 1990s at the mete-orological stations Guttannen, Grimsel, and Meiringen inthe northern vicinity of the research area). Thus, theobserved tritium contents reflect the selectivity (Gat 1981)of ground water recharge mainly from recent accumulatedwinter precipitation. Snow cover in the study area is gener-ally substantial and apart from the high availability of snowmeltwater during the melting period; Rodhe (1998)describes the meltwater input to ground water recharge asvery efficient. The snow temperature cannot exceed 0°Ceven if the atmosphere may be warm (and moist). Warm airwill therefore give a small or negative vapor pressure dif-ference between snow and the atmosphere and, thus, asmall or negative evaporation (i.e., condensation). Further-more, evaporation from water under the snowpack is negli-gible due to low temperatures and lack of turbulentexchange with the atmosphere.

The spread of the encountered tritium data can beattributed to varying flow velocities (reflected by the dif-ferent tritium contents) within separate hydraulic featuressuch as fractures and small-scale faults. In this sense, theground water system might be viewed as an assemblage ofdiscrete water parcels that move along defined flowpaths(Gat 1981). Similar observations were made by Fontes etal. (1978) in a comprehensive study in the Mont-BlancTunnel, where rapid recharge was indicated by the tritiumdata with flow velocities of ≥ 150 m/yr.

2H/18O in Precipitation, Glacial Meltwater,and Ground Water

As a further means to evaluate the ground waterrecharge conditions in the study area, 2H and 18O contentsin ground waters samples were analyzed in comparison tod2H and δ18O values in precipitation and glacial meltwater.Qualitative observations were reproduced through a simplemathematical transport model using a convolution/lumped-parameter approach (Maloszewski and Zuber 1996, 1998)to investigate recharge conditions and, in particular, thecontribution of glacial meltwater to the flow system. In thisapproach, an observed output function, e.g., δ18O contentover time in the ground water, is related to an observedinput function, e.g., δ18O content over time in precipitation,by the convolution integral. The type of transport model ischosen by defining a transit time distribution function and

873U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879

0 5 10 15 20Tritium [TU]

0.0

0.2

0.4

0.6

0.8

1.0

Sod

ium

[m

eq/l]

r2 = 0.80

limit of 95%-confidence interval

slope = -0.03

Figure 5. Plot of tritium vs. sodium content for samples fromthe Bedretto Tunnel.

parameterizing this latter function in order to calibrate thetheoretical output of the convolution integral to theobserved δ18O record.

2H/18O in PrecipitationData on the δ18O and d2H content of precipitation were

obtained from three meteorological stations in Haslital inthe north of the study area (Guttannen, Meiringen, andGrimsel/Hospiz 1970–1999) and from five additional sta-tions in the eastern, western, and southern vicinity of theresearch area (Gütsch, Andermatt, Oberwald, Binn, andRobiei 1998–1999). These stations were chosen to coverthe trajectories of potential precipitation to the study area.From this data, the local meteoric water line for the studyarea was derived following

δ2H = (7.65 ± 0.05) × δ18O + (4.7 ± 0.77) (1)

Figure 6 depicts this local meteoric water line and indi-cates its relationship to the meteoric water line defined byCraig (1961) as the locus of the isotope composition ofworldwide fresh water sources. From the same givendataset, the δ18O altitude (z) relationship was derived as

δ18O = (–0.0023 ± 0.0003) × z – (9.4 ± 0.5) (2)

This calculated δ18O altitude correlation is in goodagreement with published data (Pearson et al. 1991) giving acorrelation for the northern central Swiss Alps of (Figure 7)

δ18O = (–0.0021 ± 0.0006) × z – (10.6 ± 0.7) (3)

The δ2H altitude (z) relationship was derived as

δ2H = (–0.0146 ± 0.0025) × z – (71.46 ± 3.86) (4)

The mean annual deuterium excess d = δ2H–8 × δ18Oin precipitation at the aforementioned meteorological sta-tions ranges from 8.3‰ at Oberwald in the northern vicin-ity of the study area to 13.3‰ at Robiei in the south of thestudy area. Subtle seasonal variations in d can be observed,with deuterium excess values during the summer of 7.2‰to 12.1‰ and during winter of 9‰ to 14.2‰, respectively.In the latter monthly maxima, up to 21‰ are observed. Ageneral explanation for increased d values in winter precip-itation lies in the type of precipitation during this period.Solid precipitation such as snow during the wintertimeshows higher d values due to nonequilibrium condensationduring the growth of the ice particles (Gat 2000; Jouzel andMerlivat 1984). Additionally, Schotterer et al. (1995) reportthat winter precipitation in the Alps originating from theMediterranean Sea often shows increased d values.

2H/18O in Glacial MeltwaterGlacial meltwater is characterized by a complex signal

owing to varying fractions of snow- and icemelt contribut-ing to the overall meltwater composition throughout themelting period. Furthermore, firnification of the accumu-lated snow cover, percolating meltwater, and liquid precip-itation through the glacial body, sublimation andcondensation, as well as melting and recrystallization,cause additional fractionations in the δ18O of the ice andsnow (Stichler and Schotterer 2000). Unfortunately, in thescope of this study, no sufficient data were available toclearly constrain these relationships. To investigate the sig-nificance of the contribution of glacial meltwater to groundwater recharge, a strong simplification was made for thepurpose of the succeeding mathematical model (Mal-oszewski and Zuber 1996) of δ18O values.

Glacial meltwater is regarded as a combination ofsnow- and icemelt. At the onset of the ablation period, thecontribution from snowmelt dominates while, at later times,melting of glacial ice becomes increasingly important

U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879874

0 1000 2000 3000

Altitude [m a.s.l.]

-16

-15

-14

-13

-12

-11

-10

δ 18

O [

]

0 00

1 Meiringen2 Guttannen3 Binn4 Andermatt5 Oberwald6 Grimsel7 Guetsch8 Robiei

r2=0.916

1

2 345

6

7

8

limit of 95%-confidence interval

Figure 7. Relationship between altitude and mean annual�18O content in precipitation at meteorological stations nearthe research area.

Figure 6. Local meteoric water line based on �2H/�18O datafrom meteorological stations together with the meteoricwater line as defined by Craig (1961). Indicated are also�2H/�18O data from ground water samples collected in theBedretto Tunnel.

(Malard et al. 1999). As each of these two components canbe considered as dominant for an equally long time duringthe total ablation period, the overall δ18O concentration ofglacial meltwater (cβ) was derived from the arithmetic meanof δ18O values in snow and glacial ice, respectively (cβsnowand cβice). From balance equations, Ambach et al. (1971)show that the fraction of ice meltwater in total glacial dis-charge may even reach up to 60%. The δ18O concentrationof the snowmelt (cβsnow) was approximated from the meanannual value of winter precipitation (1970–1999) extrapo-lated to the mean altitude of the glacier. Values for the con-centration in the glacial ice (cβice) were measured in five icesamples taken from the glacier, with δ18O values rangingfrom –13.38‰ to –13.81‰ and d2H values varying between–88.3‰ and –101.1‰, respectively. These few ice samplescannot be regarded as representative for the heterogeneousand complex signal originating from the glacial ice (Stichlerand Schotterer 2000). Rather, it should be considered as anestimate for the potential contribution of glacial ice to theoverall glacial melt component. The calculated estimate forthe δ18O concentration in the glacial meltwater amounts tocβ = –16.2‰.

2H/18O in Ground WaterGround water samples for 2H and 18O determination

were collected on a monthly or bimonthly basis for a periodof 1 yr (September 1998 until September 1999) along the~1.5 km accessible section of the unlined Bedretto Tunnel.Eleven locations along the tunnel profile were sampled(Figure 1). The δ18O values observed in the tunnel showonly small variations. The meteoric origin of the groundwater is shown by its alignment to the global meteoricwater line (Figure 6). Figure 4 shows the mean δ18O valuesmeasured at the individual sampling points along the tunnelprofile. Mean deuterium excess d of the sampled groundwater range from 13.3‰ to 14.5‰. Compared to the d val-ues in recent precipitation at the meteorological stations,this may indicate a significant contribution of accumulatedwinter precipitation with relative higher d values to groundwater recharge.

Generally, encountered 18O values seem to reproducethe altitude effect, i.e., increase of overburden impliespotentially higher recharge altitudes implying depleted 18Oin precipitation. From Equations 2 and 4, estimated meanrecharge altitudes for the tunnel inflows range from 2600 to2800 m a.s.l.

Although Figure 1 shows that a straightforward delin-eation of potential recharge areas is not simple due to thecomplex and steep topography, it is anticipated thatrecharge and flow through the granite body is preferentiallyassociated to the east northeast–west southwest strikingdiscontinuities with a strong vertical component toward thesubsurface gallery. This, in turn, implies that when advanc-ing along the tunnel profile in the southeast direction, i.e.,from below the northeast-southwest trending ridge towardbelow the Gerental Valley bottom, the tunnel inflows arerecharged from decreasing altitude zones. This trend is thenreflected by the subsurface samples.

An intriguing difference from this general trend can beobserved at the sampling points T9, T11, and S1. Here,more anomalously negative δ18O values are observed than

would otherwise be predicted by the decrease in overbur-den and, thus, in potential recharge altitude. The δ18Odepleted values may be attributed to two factors: (1) theadditional component from the glacial meltwater in theground water recharge via the fault zone to the gallery, aswell as an additional altitude effect as indicated by numer-ical simulations of the ground water flow (Ofterdinger et al.submitted); and (2) the δ18O composition of the sampledground water in the northern part of the tunnel can beassumed to be derived by recharge from accumulated win-ter precipitation and actual summer precipitation during therecharge period. The ground water samples taken along thefault zone, however, show the imprint of recharge by thesetwo components, plus the additional component of glacialice. Even though this additional component is enriched in18O due to fractionation processes in the transition fromsnow to ice (Stichler and Schotterer 2000), the resultingδ18O content is still isotopically lighter than actual summerprecipitation. This additional glacial component will thusshift the resulting overall δ18O concentration in the groundwater recharge from the glaciated areas toward more nega-tive values.

Additionally, a numerical ground water flow model ofthe study area (Ofterdinger et al. submitted), in which themajor fault zone was included with discrete higher conduc-tive elements, shows that recharge via the fault zone to thetunnel occurs mainly along the upper slopes of the glaciatedareas between Geren- and Muttenglacier, plus to the southalong the upper region of the north-facing slope of theRotondo Summit. This higher recharge altitude, comparedto the neighboring sample locations in the north of the tun-nel, is also contributing to the depleted δ18O values encoun-tered along the fault zone section of the tunnel.

Figure 8 shows the temporal variations in δ18O valuesfor ground water sampled at observation point T9, as wellas the mean signal from the remaining observation points.The temporal variations at the individual sampling loca-tions are small (≤ 0.2‰; highest along the fault zone sec-tion), indicating a lack of correlation to small-scaletemporal variations in δ18O of precipitation.

Lumped-Parameter ModelTo analyze the relationship between precipitation input

and glacial meltwater in ground water recharge, theobserved δ18O values were introduced in a simplified trans-port model using a convolution/lumped-parameter approach(Maloszewski and Zuber 1996, 1998). Even though theδ18O seasonal variations at the tunnel inflows show only ahighly damped response (Figure 8), investigation by meansof model calculations seem beneficial to assess the role ofprecipitation and glacial meltwater in ground waterrecharge.

In this investigation, we applied two approaches toreproduce the observed δ18O values in ground water sam-ples taken along the major fault zone in the tunnel. First, weintroduced the precipitation input into the model to investi-gate whether observed δ18O values in the tunnel can bereproduced solely by this precipitation-derived input func-tion. In a second approach, we modified the precipitationinput function by adding a glacial meltwater component.

875U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879

The δ18O input function for precipitation (1970–1999)was derived from the stations Guttannen, Meiringen, andGrimsel (1970–1999), as well as from additionally ana-lyzed data from the stations Gütsch, Andermatt, Oberwald,Binn, and Robiei (1998–1999). The extrapolation to themean recharge altitude of the massif above the tunnel wascarried out following Equation 2 for a series of potentialrecharge altitudes of the massif above the tunnel, rangingfrom minimum elevation of 2200 m a.s.l. to maximum ele-vation of 3000 m a.s.l. The resulting δ18O signal in precip-itation shows large seasonal variations of ~14‰ with δ18Ovalues ranging from ~–6‰ during summer down to~–20‰ during winter months. Finally, weighing of theinput function by the mean monthly ground water rechargerates, derived from the hydrological model (Vitvar andGurtz 1999), was performed.

The applied code (Maloszewski and Zuber 1996)solves the direct problem, calculating the theoretical outputconcentration to a given input concentration (input func-tion) from the convolution integral for a chosen transit timedistribution (parameterized by the mean residence time)under constant flow assumption. The inverse problem issolved by a trial-and-error procedure with the possibility tochoose between six model types for the transit time distrib-ution (exponential, piston-flow, linear, dispersion, expo-nential-piston-flow, and linear-piston-flow). The best fitbetween the calculated output curve and observation data isjudged by an error criterion (standard deviation σ).

During the modeling process, it could be observed that,regardless of model type and model parameters, the theo-retical output curves calculated from the precipitation inputfunction showed enriched δ18O values compared to theobserved data. Figure 8 illustrates this behavior. This obser-vation holds even when the extrapolation of the input func-tion is carried out to the maximum potential rechargealtitude (with the potentially maximum depleted δ18O sig-nal). This implies that the input function derived from pre-cipitation data alone cannot explain the observed δ18Odepleted data. This discrepancy indicates that rechargefrom a δ18O depleted source such as accumulated recentwinter precipitation and/or glacial meltwater has a signifi-cant share in the overall recharge. Two main scenarios needto be considered to address this discrepancy.

First, recent depleted winter precipitation is accumu-lated and stored in the snow cover during the winter andenters the flow system in large amounts during the meltingseason in spring and summer, when most of the recharge tothe ground water flow system occurs. As weighing of theinput function is done by the calibrated recharge rates (Vit-var and Gurtz 1999), this latter observation is included inthe analysis. The δ18O concentrations in the input functionand, thus, in recharge, are derived from precipitation. How-ever, this precipitation-derived input function does notproperly represent the input to the flow system during themelting period. Second, an additional source of depletedδ18O values is present in parts of the recharge area in theform of glaciers. This additional source of depleted δ18O isnot accounted for in the precipitation-derived input func-tion.

Considering these scenarios, it seems plausible tomodify the input function to accommodate the δ18O con-

centrations of the glacial meltwaters. Therefore, in a secondapproach, we modified the input function to the model bycombining the monthly precipitation data with the glacialmeltwater component. In the numerical ground water flowmodel (Ofterdinger et al. submitted), the ground waterrecharge area to the fault zone section along the tunnel (andthus to sampling point T9) was delineated through particletracking. As previously mentioned, this recharge area com-prises parts of the Geren- and Witenwasserenglacier, aswell as parts of the north-facing slope of the Rotondo Sum-mit. The calibrated hydrological model (Vitvar and Gurtz1999) resolves apart from the monthly ground waterrecharge rate in this area, as well as the fraction of meltwa-ter and actual precipitation participating in the correspond-ing ground water recharge. The δ18O signal of the modifiedinput function was derived by combining the δ18O signal in

U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879876

Figure 8. Observed �18O data at station T9 and calculatedmean signal of remaining ground water sampling stations(GWmean). Indicated is the output function for lumped-parameter modeling representing the best fit to the observeddata when the precipitation-derived input function is modi-fied by an additional meltwater component (line 3). Line 1illustrates the output function for solely precipitation-derivedinput function for identical model parameters of the best fitmodel. Even for potential large residence times (line 2), theoutput function from the precipitation-derived input functionremains enriched compared to the observed data (mrt =mean residence time).

Jul98 Sep98 Nov98 Jan99 Mar99 May99 Jul99 Sep99-15.2

-15.0

-14.8

-13.8

-13.6

-13.4

-13.2

-13.0

-12.8

-12.6

-12.4 T9

Precipitation only;mrt = 14 months

Precipitation only;mrt = 72 months

Precipitation and glacialmeltwater; mrt = 14 months

GWmean

1

2

3

precipitation and the δ18O signal in the meltwater (cβ)according to their contribution to the monthly ground waterrecharge.

From the available model types, a best fit in terms ofshape of the output function, as well as in error criterion,was achieved by applying the exponential model combinedwith a piston-flow component (EPM). The applied ratio oftotal volume to the volume with the exponential distribu-tion of transit times, η, was 1.2. The necessity to apply acombined model in order to achieve a good fit as in thiscase is most probably linked to the nature of the graniticrocks. These provide, due to zones of dense fracturationand faults as well as more intact massive rock, differentflow components with different transit time distributions.As the modeled δ18O signal (sampling location T9) is asso-ciated with the major fault zone with dense fracturation, itis not surprising that a combined model leads to the best fit.

The best fit of the EPM model to the observed δ18Odata (Figure 8) yields a mean residence time of ~14 monthswith an error criterion of σ = 0.0096‰. The lack of strongvariations in the observed δ18O signal in the ground water,despite the short residence time, is mainly attributable tothe large fraction of glacial meltwater contributing to theground water recharge (~70% of annual recharge) com-pared to liquid summer precipitation.

Discussion and ConclusionsSeveral approaches have been shown to give informa-

tion on the origin of the ground water encountered in theRotondo Granite and especially on the conditions ofrecharge, in particular, the influence of the glaciatedregions in the study area. While the interpretations of theindividual investigated parameters offer potential answersto the previous questions, a more compelling scenario canbe deduced by combining the results of the individualinvestigation.

In general, all investigated parameters show that, at thetunnel elevation, only little temporal variations can beobserved. The measured δ2H/δ18O signal in ground watershows its meteoric origin (Figure 6). Hydrochemical analy-sis of the tunnel inflows shows that their composition canbe explained as an evolution along the flowpath within thegranite body. Flow is hereby assumed to be preferentiallyorientated along the steeply dipping fractures and faults,following an overall east northeast–west southwest strike.

Tritium, temperature, and δ18O content measuredalong the tunnel profile (Figure 4) all show anomalousbehavior below the glaciated areas. As this section is at thesame time situated approximately below the valley bottom,two potential explanations for these observations need to beaddressed.

Contribution of old ground water componentsupwelling below the valley bottom may be associated withlow tritium content. This would be in agreement with theobservation. The tritium/sodium correlation observed is nocompelling indicator for the contribution of older groundwater components because sodium concentrations are stillvery small and the correlation is most probably only a spa-tial coincidence and no causal relationship. The conductiv-ity measurements and hydrochemical analysis of the

ground water give also no evidence of old ground watercomponents with potentially higher mineralization. Fur-thermore, the negative temperature anomaly is notexplained by this scenario.

A second explanation for the observed data would be asignificant contribution of cold submodern glacial meltwa-ter from the overlaying glaciers with low tritium concentra-tions and depleted δ18O values. Model calculationsunderline the contribution of this δ18O depleted source toground water recharge and even though the absolute figuresof this model should be treated with care because of thesimplified model assumption, the significance of this con-tribution is demonstrated. The large fault zone associatedwith the inflows where this anomaly is observed is herebyproviding a preferential flowpath. Due to the subverticaleast northeast–west southwest strike of this structure, apotentially large glaciated area with topographic elevationup to 3000 m a.s.l. (Figure 1) is tapped. Ground waterrecharged along the strike of this structure in the glaciatedareas is believed to move by topography-driven flow andenhanced by drainage to the subsurface gallery along thispreferential flowpath toward the tunnel, whereby preserv-ing the glacial signature.

On the whole, these investigations suggest that theground water flow system within the Rotondo Granite isrecharged by meteoric waters with significant contributionof glacial meltwater, as well as meltwater from accumu-lated recent winter precipitation. Recharge contributionsfrom the glaciated areas is therefore enhanced by the orien-tation of the water-conducting fracture sets and fault zones.

Temporal variations in the observed parameters of theground water are small. Cross flow from neighboring for-mations is minor and is, if at all present, mainly restrictedto the northern boundary of the granite body. The groundwater becomes moderately mineralized along the flowpathwithin the granite, mainly through the process of silicateweathering (plagioclase hydrolysis). Model calculationsapproximate the mean residence times of the encounteredground water along the fault zone section in the gallery of~1 to 1.5 yr and underline the significant contribution ofglacial meltwater to ground water recharge. Even thoughthe absolute figure of the mean residence time should beregarded as an approximation due to the involved simplifi-cations, it indicates rather rapid recharge from the glaciatedareas via the fault zone to the gallery.

In general, the findings of this study indicate that in ahigh alpine catchment, accumulated winter precipitationand glacial meltwater may contribute significantly torecharge of deep ground water flow systems. Permeablefault zones may therefore yield preferential pathways forrapid recharge from these components to deep groundwater flow systems.

AcknowledgmentsPrecipitation samples from the meteorological stations

at Gütsch, Andermatt, Oberwald, Binn, and Robiei werekindly taken by the station supervisors. Data from the sta-tions Guttannen, Meiringen, and Grimsel were obtainedfrom the Schweizerische Meteorologische Anstalt-SMA,Zürich and the Landeshydrologie, Bern, as well as

877U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879

extracted from the Global Network for Isotopes in Precipi-tation database provided by the International AtomicEnergy Agency, Vienna. Access to the Furka Basetunneland the Bedretto Tunnel, as well as logistic help, waskindly provided by the Furka-Oberalp-Bahn. This researchwas funded by the ETH research fund. The authors alsoexpress their thanks to Ian Clark, C.K. Keller, and ananonymous reviewer for helpful reviews of the manuscript.

ReferencesAbbott, M., A. Lini, and P. Bierman. 2000. δ18O and 3H mea-

surements constrain groundwater recharge patterns in anupland fractured bedrock aquifer, Vermont, USA. Journalof Hydrology 228, 101–112.

Ambach, W., H. Eisner, R. Haefeli, and M. Zobl. 1971. Bestim-mung von Firnrücklagen am Eisschild Jungfraujoch durchMessung der Gesamtbetaaktivität von Firnproben. Zeitschriftfür Gletscherkunde und Glazialgeologie 7, 57–62.

Ambach, W., H. Eisner, H. Moser, W. Rauert, and W. Stichler.1971. Ergebnisse von Isotopenmessungen am Gletscherbachdes Kesselwanferners (Ötztaler Alpen). Annalen der Meteo-rologie 5, 209–218.

Arnold, A. 1970. On the history of the Gotthard-massif (CentralAlps, Switzerland). Eclogae Geologicae Helvetiae 63,29–30.

Ball, J.W., and D.K. Nordstrom. 1992. User’s manual forWATEQ4F, with revised thermodynamic data base and testcases for calculating speciation of major, trace, and redoxelements in natural waters. U.S. Geological Survey Techni-cal Open-File Report 91–183.

Clark, I.D., and P. Fritz. 1997. Environmental Isotopes in Hydro-geology. Boca Raton, Florida: Lewis Publishers.

Craig, H. 1961. Isotopic variations in meteoric waters. Science133, 1702–1703.

Cruchet, M. 1985. Influence de la decompression sur le com-portement hydrogeologique des massifs cristallins en basseMaurienne (Savoie, France). Géologie Alpine 61, 65–73.

Dubois, J.-D. 1993. Typologie des aquifères du cristallin: Exem-ple des Massifs des Aiguilles Rouges et Du Mont-Blanc(France, Italie et Suisse). Ph.D. thesis, Département deGénie Civil, GEOLEP, Ecole polytechnique fédérale deLausanne, France.

Flerchinger, G.N., K.R. Cooley, and D.R. Ralston. 1992.Groundwater response to snowmelt in a mountainous water-shed. Journal of Hydology 133, 293–311.

Fontes, J.-C., G.C. Bortolami, and G.M. Zuppi. 1978. Hydrolo-gie isotopique du massif du Mont Blanc. In Proceedings ofSymposium, Isotope Hydrology, June 19–23, Neuherberg,vol. 1, 411–440. Vienna: International Atomic EnergyAgency.

Forster, C., and L. Smith. 1988a. Groundwater flow systems inmountainous terrain, 1. Numerical modeling technique.Water Resources Research 24, no. 7: 999–1010.

Forster, C., and L. Smith. 1988b. Groundwater flow systems inmountainous terrain, 2. Controlling factors. Water ResourcesResearch 24, no. 7: 1011–1023.

Gat, J. 1981. Groundwater. In Stable Isotope Hydrology: Deu-terium and Oxygen-18 in the Water Cycle, IAEA TechnicalReport Series No. 210, ed. J. Gat and R. Gonfiantini,223–240.

Gat, J. 2000. Atmospheric water balance—The isotopic perspec-tive. Hydrological Processes 14, 1357–1369.

Hafner, S. 1958. Petrographie des südwestlichen Gotthardmas-sivs. Schweizerische Mineralogische und PetrographischeMitteilungen 38, 255–362.

Hoefs, J. 1997. Stable Isotope Geochemistry. Berlin: Springer.ISO. 1997. Measurement of liquid flow in open channels—

Velocity-area methods. International Organization for Stan-dardization Technical Report ISO 748: 1997(E).

ISO. 1980. Water flow measurements in open channel usingweirs and venturi flumes, Part 1. Thin-plate weirs. Interna-tional Organization for Standardization Technical ReportISO 1438/1–1980(E).

Jäger, E., and E. Niggli. 1964. Rubidium-Strontium-Isotopen-analysen an Mineralien und Gesteinen des Rotondogranitsund ihre geologische Interpretation. Schweizerische Miner-alogische und Petrographische Mitteilungen 44, no. 1: 61–81.

Jouzel, J., and L. Merlivat. 1984. Deuterium and oxygen 18 inprecipitation: Modeling of the isotopic effects during snowformation. Journal of Geophysical Research 89, no. D7:11749–11757.

Kattelmann, R., and K. Elder. 1991. Hydrologic characteristicsand water balance of an alpine basin in the Sierra Nevada.Water Resources Research 27, no. 7: 1553–1562.

Keppler, A. 1996. Hydrogeologische, hydrochemische und iso-topenhydrologische Untersuchungen an den Oberflächen-und Kluftwässern im Grimselgebiet, Schweiz. GSF-Bericht4/96. Neuherberg: GSF-Forschungszentrum für Umweltund Gesundheit.

Lhomme, D., M. Dzikowski, G. Nicoud, B. Payraud, S. Fudral,and P.-L. Guillot. 1996. Les circulations actives des eauxsouterraines des massifs cristallins alpins: Exemple desAiguilles Rouges (Haute-Savoie, France). Compte renduAcadémie des Sciences (série IIa) 323, 681–688.

López, D.L., and L. Smith. 1995. Fluid flow in fault zones:Analysis of the interplay of convective circulation and topo-graphically driven groundwater flow. Water ResourcesResearch 31, no. 6: 1489–1503.

Malard, F., K. Tockner, and J. Ward. 1999. Shifting dominanceof subcatchment water sources and flow paths in a glacialfloodplain, Val Roseg, Switzerland. Arctic, Antarctic andAlpine Research 31, no. 2: 135–150.

Maloszewski, P., and A. Zuber. 1996. Manual on mathematicalmodels in isotope hydrology. International Atomic EnergyAgency Technical Report TECDOC–910.

Maloszewski, P., and A. Zuber. 1998. A general lumped parame-ter model for the interpretation of tracer data and transit timecalculation in hydrologic systems: Comments. Journal ofHydrology 204, 297–300.

Marechal, J.-C. 1998. Les circulations d’eau dans les massifscristallins alpins et leurs relations avec les ouvrages souter-rains. Ph.D. thesis, Département de Génie Civil, GEOLEP,Ecole polytechnique fédérale de Lausanne, France.

Martinec, J., H. Oeschger, U. Schotterer, and U. Siegenthaler.1982. Snowmelt and groundwater storage in an alpine basin.In Hydrological Aspects of Alpine and High MountainAreas, 169–175. Wallingford, United Kingdom: IAHSPress.

Ofterdinger, U.S., P. Renard, and S. Loew. Submitted. Numeri-cal modeling of the ground water flow systems in theRotondo Granite. Hydrogeology Journal.

Pearson, F., W. Balderer, H. Loosli, B. Lehmann, A. Matter, T.Peters, H. Schmassmann, and A. Gautschi. 1991. Appliedisotope hydrogeology—A case study in northern Switzer-land. National Cooperative for the Disposal of RadioactiveWaste Technical Report 88–01.

Pistre, S. 1993. Rôle de la fracturation dans las circulationssouterraines du massif granitique de Millas (Pyrénées-Ori-entales). Compte rendu Académie des Sciences (série II)317, 1417–1424.

Priest, S.D. 1993. Discontinuity Analysis for Rock Engineering.London: Chapman and Hall.

Rodhe, A. 1998. Snowmelt-dominated systems. In Isotope Trac-ers in Catchment Hydrology, ed. C. Kendall and J.J.McDonnell, 391–433. Amsterdam: Elsevier.

Schotterer, U., T. Stocker, J. Hunziker, P. Buttet, and J.-P. Tripet.1995. Isotope im Wasserkreislauf—Ein neues eidgenössis-ches Messnetz. Gas Wasser Abwasser 9, 1–8. SonderdruckNr. 1348 des Schweizerischen Vereins des Gas- und Wasserfaches (SVGW), Zürich.

U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879878

Stichler, W., and U. Schotterer. 2000. From accumulation to dis-charge: Modification of stable isotopes during glacial andpost-glacial processes. Hydrological Processes 14, 1423–1438.

Tóth, J. 1963. A theoretical analysis of groundwater flow in smalldrainage basins. Journal of Geophysical Research 68, no.16: 4795–4812.

Tóth, J. 1984. The role of regional gravity flow in the chemicaland thermal evolution of ground water. In Proceedings ofthe First Canadian/American Conference on Hydrogeol-ogy: Practical Applications of Ground Water Geochemistry,June 22-26, Banff, Alberta, Canada, ed. B. Hitchon and E.Wallick, 3–39. Worthington, Ohio: National Water WellAssociation.

Tranter, M., G.H. Brown, A.J. Hodson, and A.M. Gurnell. 1996.Hydrochemistry as an indicator of subglacial drainage sys-tem structure: A comparison of alpine and sub-polar envi-ronments. Hydrological Processes 10, 541–556.

Vitvar, T., and J. Gurtz. 1999. Spatially distributed hydrologicalmodeling in the Rotondo area, western Gothard-Massif.Technical Report 3465/16, ETH-Zürich, Engineering Geol-ogy, ETH-Hönggerberg.

Ward, J., F. Malard, K. Tockner, and U. Uehlinger. 1999. Influ-ence of ground water on surface water conditions in a gal-cial flood plain of the Swiss Alps. Hydrological Processes13, 277–293.

Zijl, W. 1999. Scale aspects of groundwater flow and transportsystems. Hydrogeology Journal 7, 139–150.

879U.S. Ofterdinger et al. GROUND WATER 42, no. 6: 868–879