Determining long time-scale hyporheic zone flow paths in Antarctic streams

HYDROLOGICAL PROCESSESHydrol. Process. 17, 1691–1710 (2003)Published online 5 March 2003 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hyp.1210

Determining long time-scale hyporheic zone flow paths inAntarctic streams

Michael N. Gooseff,1* Diane M. McKnight,1 Robert L. Runkel2 and Bruce H. Vaughn1

1 Institute of Arctic and Alpine Research, University of Colorado, Boulder, CO, 80309-0450, USA2 U. S. Geological Survey, Mail Stop 415, Denver Federal Center, Denver, CO, 80225, USA

Abstract:

In the McMurdo Dry Valleys of Antarctica, glaciers are the source of meltwater during the austral summer, and thestreams and adjacent hyporheic zones constitute the entire physical watershed; there are no hillslope processes inthese systems. Hyporheic zones can extend several metres from each side of the stream, and are up to 70 cm deep,corresponding to a lateral cross-section as large as 12 m2, and water resides in the subsurface year around. In this study,we differentiate between the near-stream hyporheic zone, which can be characterized with stream tracer experiments,and the extended hyporheic zone, which has a longer time-scale of exchange. We sampled stream water from GreenCreek and from the adjacent saturated alluvium for stable isotopes of D and 18O to assess the significance and extentof stream-water exchange between the streams and extended hyporheic zones over long time-scales (days to weeks).Our results show that water residing in the extended hyporheic zone is much more isotopically enriched (up to 11‰D and 2Ð2‰ 18O) than stream water. This result suggests a long residence time within the extended hyporheic zone,during which fractionation has occurred owing to summer evaporation and winter sublimation of hyporheic water. Wefound less enriched water in the extended hyporheic zone later in the flow season, suggesting that stream water may beexchanged into and out of this zone, on the time-scale of weeks to months. The transient storage model OTIS was usedto characterize the exchange of stream water with the extended hyporheic zone. Model results yield exchange rates(˛) generally an order magnitude lower (10�5 s�1) than those determined using stream-tracer techniques on the samestream. In light of previous studies in these streams, these results suggest that the hyporheic zones in Antarctic streamshave near-stream zones of rapid stream-water exchange, where ‘fast’ biogeochemical reactions may influence waterchemistry, and extended hyporheic zones, in which slower biogeochemical reaction rates may affect stream-waterchemistry at longer time-scales. Copyright 2003 John Wiley & Sons, Ltd.

KEY WORDS isotope transport; OTIS; Dry Valleys; hyporheic zone

INTRODUCTION

Fluvial hydrological flow paths, and in particular hyporheic exchange processes, have been identified asimportant for understanding aquatic biogeochemical cycling (Harvey and Bencala, 1993; Mulholland et al.,1997; Gooseff et al., in press). White (1993) noted that there is no single definition or method of delineationof the hyporheic zone that applies for all hyporheic zone studies, although Harvey and Bencala (1993)developed the following hydrological definition: a subsurface flow path parallel to stream flow in which waterthat was recently in the stream mixes with subsurface water and will shortly return to the stream. As Bencala(2000) points out, ‘hyporheic zones influence the biogeochemistry of stream ecosystems by increasing soluteresidence times.’ The extent of the actively exchanging hyporheic zone within the adjacent wetted zoneof a typical stream cannot be measured directly (i.e. vertical extent versus lateral extent). Current methodsfor characterizing hyporheic zone influences on stream processes generally require a stream-tracer injectionexperiment and subsequent inverse modelling to fit model output to observed data using the transient storage

* Correspondence to: Michael N. Gooseff, Department of Aquatic Watershed and Earth Resources, Utah State University, 5210 Old MainHill, Logan, UT 84322-5210, USA. E-mail: [email protected]

Received 22 February 2002Copyright 2003 John Wiley & Sons, Ltd. Accepted 6 August 2002

1692 M. N. GOOSEFF ET AL.

(TS) model developed by Bencala and Walters (1983). Harvey et al. (1996) have discussed the reliability ofthe stream-tracer approach to characterize hyporheic exchange. Their results show that the method is reliablefor characterizing ‘near-stream’ flow paths and storage zones, that is, flow paths that exchange on the time-scale of the experiment (generally hours to a few days), and that the technique most reliably characterizesthis exchange at low baseflow conditions. The tracer method is not sensitive to slower exchange with storagezones more distant from the stream, because the time-scale of exchange into and out of ‘distant’ hyporheiczones is greater than the duration of the stream-tracer experiment. In this study, we define the hyporheic zoneas the area of saturated alluvium adjacent to the stream that exchanges water and solutes, on any time-scale,with the stream. Further, we will distinguish between two hyporheic zones, (i) the near-stream hyporheic zone,which is readily characterized by the stream-tracer technique, as described by Harvey et al. (1996), and hastime-scales of exchange on the order of hours, and (ii) the extended hyporheic zone, in which exchange withthe stream occurs on the time-scale of days to weeks or longer. We propose that the extended hyporheic zoneis tenuously connected to the stream, and we expect that over a long time (weeks to months), the influenceof the stream on the extended hyporheic zone, and vice versa is detectable.

The streams of the Antarctic Dry Valleys provide a unique environment in which to study fluvial processes.The streams are fed by glacial meltwater during the austral summer, a period of constant sunlight. Becausethere is no vegetation, the extent of the wetted sediment adjacent to the stream, which comprises the hyporheiczone, can be measured directly, unlike temperate streams. There is no regional groundwater system in contactwith the streams (Chinn, 1993), thus all water within the adjacent wetted zone comes from the stream. Theexchange of stream water between the stream and the near-stream hyporheic zone has been characterized intwo Dry Valley streams in previous tracer experiments: Huey Creek (Runkel et al., 1998) and Green Creek(McKnight et al., 1999). Both studies found that the porous alluvium of the streambeds allowed for rapid ratesof near-stream hyporheic exchange. Consequences of rapid exchange between the stream and the near-streamhyporheic zone are increased weathering rates of streambed material (Lyons et al., 1997; Maurice et al., 2002;Gooseff et al., in press) compared with temperate watersheds, and increased nutrient uptake as stream waterflows toward the closed-basin lakes on the valley floors (McKnight et al., 1999). Physical cross-sectionalwetted areas adjacent to streams are much larger than the storage zone size identified by previous tracerexperiments in Dry Valley streams (Conovitz, 2000). We hypothesize that all of this extensive wetted cross-sectional area comprises an actively exchanging hyporheic zone, and that previous tracer methods have notidentified long timescale extended hyporheic exchange.

Many recent studies have focused on the discretization of water flow paths and water residence times in theenvironment using stable isotopes (Stout, 1967; Hooper and Shoemaker, 1986; Kennedy et al., 1986; Turneret al., 1987; Maule and Stein, 1990; McDonnell et al., 1991; Cooper et al., 1993; Kendall and Caldwell,1998). Other studies have shown that the flow path water takes and its residence time in the environment (e.g.in the subsurface of a hillslope) strongly influence the quality of water observed in streams (Kennedy, 1971;Pilgrim et al., 1979; Cirmo and McDonnell, 1997). The most commonly used stable isotopes as hydrologicaltracers are deuterium (D) and 18O. Stable isotopes can be used in hydrograph separation, to discern subsurfaceflow from overland flow as well as to discern spring snowmelt contribution to stream flow (Cooper et al.,1991, 1993; McDonnell et al., 1991; Stewart and McDonnell, 1991; McNamara et al., 1997). Stable isotopescan also be used to quantify hydrological fluctuations in glacially dominated catchments (Epstein and Sharp,1959; Theakstone and Knudsen, 1996), and as an indicator of permafrost and active layer contributions tolocal hydrology (McLean et al., 1999).

Modelling isotope transport and fractionation in groundwater systems (Krabbenhoft et al., 1990; Johnsonand DePaolo, 1997) has provided a framework for similar studies in other hydrological systems. Studiesof stable isotopes with respect to hydrological and atmospheric interactions in temperate streams have beenreported. McKenna et al. (1992) investigated bank storage of enriched Lake Tahoe water in the Truckee River,USA. Gremillion and Wanielista (2000) reported evaporative υ18O enrichment by roughly 1‰ of water in theEconlockhatchee River, USA, compared with other watershed sources (groundwater, precipitation). Simpson

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1693

and Herczeg (1991) also studied stream-water enrichment owing to evaporation on the River Murray insoutheast Australia. They found that υD enrichment was C0Ð62‰ for 1% loss of water volume.

In this study we use stable isotopes of stream and substream water as a natural hydrological tracer to discernthe hydrological flow path and the residence times of water in Dry Valley stream systems. We hypothesize thatisotopic signatures of the water residing in the extended hyporheic zone will be enriched compared with theisotopic signature of flowing stream water. We further propose that this enrichment in the extended hyporheiczone is probably the result of evaporation (during the summer) and sublimation (during the winter) of thewater through the porous sediments. Enrichment of the water from evaporation and sublimation suggeststhat extended hyporheic water has a longer and slower flow path than water in the stream, allowing ampletime for these processes to occur. This hypothesis is tested using the TS model to simulate stable isotopeobservations (D and 18O) from the stream and the extended hyporheic zone. Resulting exchange parametervalues provide order-of-magnitude estimates of water residence time within extended hyporheic zones, andthe time-scale of its influence on stream biogeochemistry. The magnitude of isotopic signature differencebetween the stream and subsurface water, as well as an evaporation fractionation constraint using simpleevaporation pan experiments, suggest that stream-water exchange occurring with the extended hyporheic zoneinfluences the isotopic signature of the stream. To the best of our knowledge, this is the first study in whichstable isotope data are used to model hyporheic exchange.

Site description

The McMurdo Dry Valleys of Antarctica are made up of many alpine, piedmont and terminal glaciers,permanently ice-covered closed basin lakes, bare ice-free soils and stream channels. The climate is very coldand dry. The average annual air temperature is �20 °C, and annual lake-ice-cover ablation rates are of theorder of 0Ð35 m (Clow et al., 1988; Chinn, 1993). Our study focuses on Delta Stream and Green Creek, inthe Lake Fryxell basin in Taylor Valley (Figure 1). Delta Stream originates near the Howard Glacier, alongthe southern border of the valley. It is the longest stream in Taylor Valley, at 11Ð2 km, and is gauged about100 m upstream of its mouth. Green Creek is approximately 1 km long and originates at the south-easterntongue of the Canada Glacier, and is gauged halfway down its length. Continuous discharge was computedfrom 15-min pressure-transducer stage data and a stage–discharge relationship was developed for each streamgauging station (Von Guerard et al., 1995). At the height of the austral summer, meltwater is produced atthe glaciers and flows down sandy alluvial stream channels. Stream discharge ranges from extended periodsof zero flow to peaks of the order of 100 L s�1 (Figure 2). The flow season generally lasts 6 to 12 weeks.Stream flow is the main source of water and nutrients to the closed basin lakes on the valley floors. The mainloss of water from these lakes is ablation of their 4 to 6 m thick ice covers (Chinn, 1993).

The streambed alluvium is very porous, and becomes saturated as stream flow progresses down the channelwhen glacial melt begins. At the beginning of the summer, the active layer is still frozen with stream waterfrom the previous flow season, and the amount of interstitial space within streambed alluvium is at a mini-mum (Figure 3A). Stream flow generally becomes continuous in late November or early December, and thestreambed is continually wetted. Obvious lateral wetted zones are established up to several metres from theedge of the stream over several weeks (Conovitz, 2000). As the summer progresses, the active layer thawsand the depth to permafrost is generally from 30 to 70 cm (Figure 3B). Permafrost acts as a vertical barrierto water penetration underneath the stream (Conovitz et al., 1998, Conovitz, 2000). Storage of stream waterin the streambed material represents a large reservoir of water, as much as 11 088 m3 in Delta Stream and990 m3 in Green Creek. During colder, low-flow years, the amount of water in the hyporheic zone can beseveral times the amount of water in the stream (Bomblies, 1998; McKnight et al., 1999).

The cross-sectional wetted area (i.e. the physical size of the combined near-stream and extended hyporheiczones) is measured directly by driving a thin, rigid steel rod into the observed wetted streambed alluviumat 1 m intervals across the stream until permafrost or an obstruction was reached, following the method ofConovitz (2000). This method is minimally invasive to the streambed and hyporheic zone, and it is accurate

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1694 M. N. GOOSEFF ET AL.

Canada Glacier

CommonwealthGlacier

Crescent GlacierHowardGlacier

Gre

enCre

ek

Del

taSt

ream

Harnish

Stream

VonGuerardStream

Aiken Creek

Huey Creek

CanadaStream

LostSeal

Stream

McKnight C

reek

Cre

scen

tSt

ream

ManyGlaciersPond

Scale in meters

1000 20000

N

TaylorValley

LAKEFRYXELL

Figure 1. Map of the Lake Fryxell basin, lower Taylor Valley, Antarctica. Dots represent locations of stream gauges. McMurdo Sound isapproximately 8 km to the east of this map area

within 1 m2. Cross-sectional areas have been measured as large as 12 m2 in Dry Valley streambeds. Porosityof the alluvium is of the order of 0Ð33 (Conovitz, 2000), corresponding to c. 4 m2 of porous space in thewetted streambed. Alternatively, stream-tracer techniques have identified cross-sectional hyporheic storageareas (defined later as AS) generally less than 0Ð5 m2, with the exception of one reach in Huey Creek, whichwas 3Ð07 m2 (Table I) (Runkel et al., 1998).

METHODS

Sample collection

One synoptic survey on Delta Stream was carried over 7+ km of the stream, on 18 January 1994. Fourteenstream-water samples were collected in 30-mL polyethylene bottles as liquid water and taken back to camp

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1695

05-Jan-9412-Jan-94

19-Jan-9426-Jan-94

02-Feb-94

09-Feb-94

Dis

char

ge R

ate

(L s

-1)

0

20

40

60

80

100

120

140

160

180

08-Dec-99

15-Dec-99

22-Dec-99

29-Dec-99

05-Jan-0012-Jan-00

19-Jan-0026-Jan-00

02-Feb-00

Dis

char

ge R

ate

(L s

-1)

0

20

40

60

80

100

120

(a)

(b)

Figure 2. Hydrographs from (A) Delta Stream for the 1993–1994 season, and (B) Green Creek for the 1999–2000 flow season. Arrowsdenote the dates of synoptic sampling campaigns

to be frozen. All samples were then shipped back to Boulder, CO, for analysis of υD and υ18O at the StableIsotope Laboratory at the Institute of Arctic and Alpine Research (INSTAAR).

A comprehensive sampling of Green Creek was conducted in the 1999–2000 flow season. Four synopticsampling campaigns were completed on 7 December 1999, 21 December 1999, 7 January 2000 and 23 January2000. Water samples for isotopic analysis of υD and υ18O were collected from three stream sites: 0 m, 59 mand the mouth of the stream, and 508 m, where 0 m corresponds to the injection point of a previous stream-tracer experiment. Three sample transects were established at 161 m (transect 1), 257 m (transect 2), and357 m (transect 3) (Figure 4), each of which had four sampling points: (1) the stream, (2) an in-stream well

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1696 M. N. GOOSEFF ET AL.

active layer

(a)

(b)

active layer

Wetted zone

permafrost

40 – 70 cm

3 – 5 m

10 – 12 m

Frozen infiltration fromprevious season

permafrost

Figure 3. Schematic lateral cross-sectional diagram of streambed alluvium (A) prior to austral summer stream flow and (B) when the streamsare flowing. Figures not drawn to scale

Table I. Transient storage model parameter values found in previous stream tracerstudies of conservative solutes in Dry Valley streams

Experimental Parameterreach

D (m2 s�1) A (m2) ˛ (s�1) AS (m2)

Huey Creek, 1993a

0–213 m 0Ð5 0Ð11–0Ð18 1Ð07 ð 10�3 0Ð20213–457 m 0Ð5 0Ð10–0Ð17 5Ð43 ð 10�4 0Ð25457–762 m 0Ð5 0Ð10–0Ð18 1Ð62 ð 10�2 0Ð14762–1052 m 0Ð5 0Ð19–0Ð01 4Ð67 ð 10�4 3Ð07

Green Creek, 19950–50 m 0Ð5 0Ð02–0Ð07 3Ð5 ð 10�5 0Ð0550–226 m 0Ð5 0Ð02–0Ð07 1Ð9 ð 10�4 0Ð40226–327 m 0Ð5 0Ð02–0Ð07 2Ð7 ð 10�4 0Ð39327–497 m 0Ð5 0Ð02–0Ð07 1Ð1 ð 10�4 0Ð07

a From Runkel et al. (1998).

in the thalwag of the stream, and (3 and 4) two lateral wells placed 2 m and 4 m away from the edge of thestream (Figure 5). The wells were constructed of 5Ð1-cm diameter PVC pipe, capped at one end and slottedwith a saw. In-stream well slots spanned a distance of 5 cm and were screened with 1 mm mesh screen,and effective sample depth was approximately 15–20 cm. Lateral wells were slotted over their entire length,sampling a range of subsurface depths, but roughly 20 cm at most. Water was pumped from each well with

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1697

#

#

$T

#Y

#Y

#Y

#Y

#Y

Algal Transect

Sampling point ortransect

Stream Gauge

Approximately 50 m

N

Transect 1Transect 2

Transect 3

Flow Direction

Figure 4. Map of sampling transect locations on Green Creek. Contour intervals represent 1 m of elevation change

Left Hand Bank

4 m

2 m

Instream Well Well A Well B

Figure 5. Well placement near Green Creek at three transects, each to a depth of roughly 20 cm below ground surface or streambed

a Nalgene hand pump and poured into 30-mL polyethylene bottles. Wells were not purged beforehand owingto very slow refill rates and limitations of field logistics. Duplicate sample sets were made, when the samplevolume was sufficient (>30 mL). The first set of samples were frozen the same day and remained frozenuntil being thawed just before isotopic analysis. The second set was filtered the same day and refrigerated formajor ion analysis. The lateral wells at transect 2 yielded water only during the second and third synoptics.During the fourth synoptic, lateral wells at transects 1 and 2 were dry, and no sample was collected.

On 11 January 2000, a batch evaporation pan experiment was conducted at the Lake Hoare field camp(on the west side of the Canada Glacier). At 0925 hours, 4Ð0 L of stream water was collected from nearby

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1698 M. N. GOOSEFF ET AL.

Table II. Well samples from four sequentialpurges of a lateral well on Green Creek on

10 January 2001

Time υD υ18O D excess

1245 �244 �29Ð8 �5Ð81430 �245 �29Ð9 �5Ð71530 �245 �30Ð0 �5Ð41630 �245 �30Ð0 �5Ð2

Andersen Stream, placed in a pan (water depth was c. 5 cm, surface area was c. 0Ð09 m2) on top of anactivated stirring plate to simulate stream-water movement. The water was sampled initially for baselinevalues and then repeatedly sampled on a 1-h interval until 1630 hours, when the experiment ended. Samplesof approximately 75 mL of pan water were collected manually; 30 mL for isotope analysis, 45 mL for waterquality analysis. Evaporation was calculated from evapoconcentration rate of Cl and SO4 samples from theevaporation pan.

Wells were not purged during the 1999–2000 sampling. During the 2000–2001 flow season a lateral wellwas re-established on the left-hand bank of Green Creek. The well was sampled about a week after installation,purged four times, each purging was sampled for isotopic analysis (Table II). The mean air temperaturewas �4Ð4 °C, main relative humidity was 78%, and the mean wind speed was 3Ð51 m s�1 for the date ofadditional well purging, as recorded at the Lake Fryxell meteorological station (http://huey.colorado.edu). Thedata presented in Table II suggest that the wells did not act as evaporative conduits, and that well samplesare representative of subsurface water.

Sample analysis

All water sample D and 18O analysis was carried out at the Stable Isotope Laboratory at INSTAAR. For18O, a CO2 –H2O equilibration method was used. Isotope ratios were determined on a Micromass (formerlyVG) SIRA Series II mass spectrometer (the use of trade names is for identification purposes only and does notconstitute endorsement by the U.S. Geological Survey). Results are reported relative to V-SMOW standard,to a precision of š0Ð10‰ (typically less). For D analyses, an automated flow-through uranium reductionfurnace was used to analyse up to 56 samples in an automated run. Isotope ratios are determined on adedicated Micromass (formerly VG) SIRA Series II mass spectrometer. Results are reported relative to V-SMOW standard, to a precision of š1Ð0‰ (typically š0Ð5‰). This analysis works best for relatively ‘clean’water with low or normal conductivity (Vaughn et al., 1998).

Water samples taken for major ion analysis were filtered immediately with 0Ð45 µm GFC filters. Soluteconcentrations were determined by ion chromatography at the Crary Laboratory, McMurdo, Antarctica (Welchet al., 1996). Reactive silicate was determined using a standard colorimetric method (Mullin and Riley,1955) at the Byrd Polar Research Center, Columbus, OH. All ion analyses are accurate to within 5% of theconcentration.

Transient storage modelling

To characterize the exchange between the stream and an extensive hyporheic zone, we used a one-dimensional stream solute transport model, OTIS (Runkel, 1998), which was modified to account forevaporation fractionation

�in-stream�∂υR

∂tD �Q

A

∂υR

∂xC 1

A

∂

∂x

(AD

∂υR

∂x

)C ˛�υS � υR� � υEV qEV

V�1�

�in-storagezone�∂υS

∂tD ˛

A

AS�υR � υS� �2�

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1699

where D is the dispersion coefficient (m2 s�1), A is the cross-sectional area of the stream (m2), AS is thecross-sectional area of the storage zone (m2), ˛ is the storage zone exchange coefficient between the streamand the storage zone (s�1), υR is the main channel isotope abundance (‰), υS is the storage zone isotopeabundance (‰), x is the distance downstream (m), t is time (s), Q is stream flow rate (m3 s�1), V is thevolume of stream water in a control volume (m3), qEV is the evaporation rate from a single control volume(m3 s�1) and υEV is the isotopic abundance of evaporating water (‰).

The isotopic abundance of evaporating water was characterized using the equilibrium model by Craig andGordon (1965)

υEV D ˛ŁυR � hυA � ε

1 � h � �ε10�3��3�

where ˛* is the equilibrium fractionation factor, h is the relative humidity of the air above the stream, as adecimal of 1Ð0, ε is the kinetic fractionation factor, defined as [1000 (1 � ˛*)], ε is the total fractionationfactor, assumed to be 4Ð64 for D and 5Ð27 for 18O, and υA is the isotope abundance of atmosphericmoisture (‰).

UCODE, a universal inverse model (Poeter and Hill, 1998), was used to regress optimal values of ˛ andυEV. UCODE incorporates analysis of model output sensitivities of all observations to parameter perturbationinto a modified Gauss–Newton optimization approach. Equation (3) was then used to calculate values of υA,from the regressed values of υEV. The υA abundances should be more similar to precipitation υD and υ18Oabundances than υEV abundances.

Our application of the TS model is slightly different than the typical application using stream-tracerexperiment techniques, in which the TS model is used to fit stream solute breakthrough curves from aconservative tracer. The typical approach results in optimized values of D, A, ˛ and AS, to characterizeadvection and hyporheic exchange processes. In this study, parameter values for the above equations (except˛ and υEV, which are regressed) are supplied by measurements and reasonable values from previous tracerexperiments in Dry Valley streams (Table I). Additionally, the TS model was modified so that initial conditionscould be defined for the stream and hyporheic zone.

The period between 21 December 1999 and 07 January 2000 (384 h), was modelled as a transient simulationfor the stream reach from 59 m to 508 m. Lateral well observations were averaged for each transect. Asteady stream flow of 0Ð017 m3 s�1, the average flow for this time period, was used for the modelling.Parameterization of the TS model was accomplished using representative hydrological parameters fromprevious tracer experiments on Green Creek, as well as field measurements from Green Creek. The dispersioncoefficient (D) was fixed at 0Ð5 m2 s�1, and A was fixed at 0Ð06 m2 for all reaches, based on stream-flowmeasurement records. The evaporation rate derived from the pan evaporation experiment (see Results section)was applied to the entire stream reach, assuming a surface area of 5 m2 for each control volume. The initialconditions in the stream and in the extended hyporheic zone are defined by the data collected in the secondsynoptic (21 December 1999), spatially linearly interpolated between sample sites.

The upstream boundary condition is also defined by the collected data. We used the upstream isotopeabundances linearly interpolated through time between the dates of the two synoptic sampling campaigns.The isotopic abundance of the source of the water, glacier ice, changes, as does the transport time fromthe glacier to the headwater pond and the channel throughout the season. There is also no precipitationcontribution to stream flow. Therefore, no part of the stream is acting as a constant end-member, as is ideallyassumed in typical isotope hydrology studies.

Values of AS in these simulations were fixed at 2Ð98 m2, 2Ð36 m2, 3Ð95 m2 and 3Ð00 m2 from upstreamto downstream, based on wetted zone measurements corrected for porosity (0Ð33), accounting for 10 cm ofunsaturated sediment in the lateral wetted zone. These values represent large single storage zones in the TSmodel, in this case simulating a single extended hyporheic zone for each stream control volume and includethe physical areas of both the near-stream and extended hyporheic zones. These values are larger than AS

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1700 M. N. GOOSEFF ET AL.

values previously published for temperate streams, the largest of which was reported as 2Ð0 m2 in reach 2 ofLittle Lost Man Creek, CA (Bencala, 1984; Runkel, 2002).

Because the hyporheic zone is in general quite heterogeneous, a sample at any one point at any one timemay be unique to that point and time. To overcome this limitation, we assumed that the average of thetwo lateral well-sample values at each transect, for each sampling date was representative of the extendedhyporheic zone. If one were to assume that a hyporheic sample at any one point is representative of all waterthat is the same distance away from the stream (in a defined band surrounding the stream), any number ofnested storage zones could be modelled, each likely to have a unique time-scale of exchange, and uniqueAS and ˛ values. The representation presented here intentionally forces long residence times to be producedbecause large storage zones are defined. Thus, our results are indicative only of the time-scale of exchangebetween the stream and the extended hyporheic zone.

RESULTS

Evaporation pan experiment

During the 1999–2000 flow season, a batch evaporation pan experiment was run to constrain the rateof evaporation and the effects of evaporation on isotopic enrichment. Figure 6A presents the change ofwater volume in the evaporation pan, including evaporation and sampling, and the evapoconcentration ofconservative solutes Cl and SO4. The pan volume data assume a constant evaporation rate during thetime of the experiment, found to be 7Ð14 ð 10�8 m s�1 (6Ð17 mm day�1), as determined from modellingevapoconcentration of Cl and SO4. From the data presented in Figure 6B, an evaporation enrichment rate canbe calculated as 1‰ h�1 for υD and 0Ð2‰ h�1 for υ18O. As expected, the D excess (Figure 6C), computedas [υD � 8*υ18O], decreased with time, owing to the evaporation enrichment.

Green Creek synoptic sampling

The solute data from the stream and lateral wells in Green Creek indicate generally higher concentrationsof major ions in the wells compared with the stream (Figure 7) for two of the three transects. Transect 3exhibits a different pattern because the alluvium at the base of the lateral wells was flooded at higher flows.For the other transect locations, the increased concentrations of all major ions in the extended hyporheic zone,compared with the stream, suggest that the water sampled from the lateral wells had been in storage a longertime than the time-scale of advective transport in the stream. Thus, the solute differences between the streamand hyporheic waters are evidence of increased residence time and biogeochemical differences between thestream and the extended hyporheic zone. This is further supported by the depleted isotopic abundances of thelateral well samples on 07 January 2000, compared with stream water at transect 3 (Figure 8).

The data from the first Green Creek synoptic (Figure 8A) shows a strong downstream enrichment of 8‰υD and 1Ð3‰ υ18O. A comparison of enrichment rates (based on travel times for flow conditions at the timeof sampling) reported in Table III indicates that stream enrichment is probably not the result of evaporationalone. Depleted stream water mixing with enriched waters in the hyporheic zone is probably the only otherprocess that could contribute. Assuming the evaporation enrichment rates from the pan experiment apply,evaporation would account for only 1‰ υD enrichment and 0Ð2‰ υ18O enrichment.

Further synoptic sampling after the wells were established showed smaller enrichments along the streamlength (Figures 8B–D), about 2‰ υD and less than 0Ð5‰ υ18O on 21 December 1999, and about 3‰ υD andabout 0Ð5‰ υ18O on 07 January 2000. Stream flow on 21 December 1999 was 42Ð6 L s�1, corresponding toa travel time of roughly 33 min, and the flow rate on 07 January 2000 was 7Ð2 L s�1 (Figure 2). The watersampled from the lateral wells was enriched by as much as 11‰ υD and 2Ð2‰ υ18O compared with streamwater for both 21 December 1999 and 07 January 2000. At the same time, water in the in-stream wells wasalso slightly enriched with respect to stream water. With the exception of 23 January 2000, the downstreamenrichment appears to result from both evaporation fractionation and mixing with enriched hyporheic waters.

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1701

Solu

te C

once

ntra

tion

(mg

L-1

)

4.0

4.3

4.6

4.9

5.2

5.5(a)

Vol

ume

(L)

3.2

3.4

3.6

3.8

4.0

Col 5 vs Col 6 Col 5 vs Col 7 Col 2 vs Col 3

dD (

‰)

-2333

-232

-231

-230

-229(b)

d18O

(‰

)

-29.0

-28.5

-28.0

Hour of Evaporation Experiment

D E

xces

s (‰

)

-6

-5

-4

-3

-2

-1

0(c)

D18O

0 1 2 3 4 5 6

ClSO4water in pan

Figure 6. Evaporation pan experiment results: (A) volume and conservative solute data, (B) υD and υ18O data, and (C) D excess. Note,steep drops in pan water volume represent sample acquisition

The decreasing isotope abundances of lateral well waters in the downstream direction (Figure 8B and C)do not represent a significant trend. The lateral wells at transect 3 were flooded in high flow events, so theisotope abundances at that location more closely resemble stream-water isotope abundances. The lateral wellsat transects 1 and 2 more likely represent hyporheic zone heterogeneity. The lateral well locations in transect2 may be in better contact with the stream than the lateral well locations in transect 1. However, there isno physical evidence to suggest that connections between the stream and the extended hyporheic zones areenhanced at downstream locations, compared with upstream locations.

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1702 M. N. GOOSEFF ET AL.

0

2

4

6

8 streamlateral wells

05

1015202530354045

Cl Na CaSi MgK SO4

Transect 1, 161 m

Solu

te C

once

ntra

tion

(mg/

L)

0

2

4

6

8

Solu

te C

once

ntra

tion

(mg/

L)

05

1015202530354045

Transect 2, 257 m

0

2

4

6

8

05

1015202530354045

Transect 3, 357 m

Figure 7. Solute concentration averages from Green Creek stream and lateral well locations for all synoptics from five to six stream samplesper transect, and three to five lateral well samples per transect. Error bars represent standard deviations

δD (

‰)

-258

-255

-252

-249

-246

-243

0 100 200 300 400 500

δ18O

(‰

)

-33

-32

-31

-30

-29

Distance Downstream (m)

0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500

(a) (b) (c) (d)

streamin-stream welllateral well Alateral well B

streamin-stream welllateral well Alateral well B

Figure 8. Observed isotopic abundances of D and 18O from the four synoptic surveys on Green Creek, (A) 07 December 1999, (B) 21December 1999, (C) 07 January 2000, and (D) 23 January 2000

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1703

Table III. Fractionation rates for stream synoptic data and evaporation pan experiment

Traveltime (h)

Flow rate(L s�1)

D fractionationrate (‰ h�1)

18O fractionationrate (‰ h�1)

Evaporation 6Ð5a 0 C1 C0Ð2pan experiment

Delta Stream 24 3Ð0 C1 C0Ð118 January 1994

Green Creek 4Ð9 0Ð4 C2 C0Ð37 December 1999

Green Creek 0Ð5 42Ð6 C4 C0Ð721 December 1999

Green Creek 2Ð3 7Ð2 C1 C0Ð27 January 2000

Green Creek 5Ð3 0Ð4 C0b C0Ð0b

23 January 2000

a Evaporation pan experiment lasted for 6Ð5 h; and therefore, is not a true travel time.b Fractionation rates found to be less than analytical precision of isotope abundances.

Delta stream synoptic sampling

The results of the Delta Stream synoptic study are presented in Figure 9. Over the more than 7 km ofstream length from glacier to lake, there is a substantial overall enrichment of stream water, with respectto both D and 18O. The enrichment was not a smooth increase downstream. At roughly 1Ð5 km from theglacier, a strong increase in both υD and υ18O was observed, just downstream from a small pond. The pondincreased the residence time of the water, allowing more time for evaporation fractionation before stream watercontinues down the channel. The variability in the reach from about 2 km to 5Ð5 km probably resulted fromboth evaporation (causing enrichment) and mixing with recently stored waters from the hyporheic zone, whichmay have been more depleted. The D-excess decreasing trend presented in Figure 9B suggests that some ofthis enrichment was the result of evaporation, particularly the large drop after the pond, around 1Ð5 km.

Transient storage modelling results

The TS model simulations for both D and 18O resulted in excellent fits with respect to the streamobservations, but compromised fits to the hyporheic zone observations (Figure 10). This is a result of using thesame ˛ and υEV values for each reach in Equation (1). The modelling of D and 18O transport resulted in optimal˛ values of 1Ð18 ð 10�5 s�1 and 1Ð36 ð 10�5 s�1, respectively. These values are independent measures ofexchange, and generally are an order of magnitude lower than those reported in other Dry Valley hyporheicinvestigations (Table I). Optimized values of υEV were �271‰ for υD, with 95% confidence intervals of�38‰ and �504‰ and �43Ð0‰ for υ18O, with 95% confidence intervals of �4Ð9‰ and �81Ð2‰. TheseυEV values correspond to reasonable υA values of �126‰ for υD, and �8Ð9‰ for υ18O. The UCODE-producedcomposite scaled sensitivity values for ˛ were an order of magnitude greater in both simulations than for υEV,suggesting that the data set contained more information for parameterizing ˛.

Additional model runs were made for D transport for two cases: (i) no evaporation, and therefore noevaporation fractionation, and (ii) no hyporheic exchange, therefore no lateral mixing (Figure 11). It is clearfrom the comparison of these alternative scenarios that the stream isotope abundances are greatly dependentupon the evaporation fractionation, as there is little difference between the optimized model fit (which is thesame as that presented in Figure 10A) and the no exchange scenario. In comparison with the fractionationrates calculated in Table III, it appears that stream-water evaporation is responsible for more than 1‰ h�1

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1704 M. N. GOOSEFF ET AL.

δ18O

(‰

)

-27.0

-26.5

-26.0

-25.5

-25.0

-24.5

-24.0

δD (

‰)

-220

-218

-216

-214

-212

-210

-208(a)

Distance Downstream (m)

0 1000 2000 3000 4000 5000 6000 7000

D e

xces

s (‰

)

-18

-16

-14

-12

-10

-8

-6

-4(b)

D18O

Figure 9. Data from Delta Stream synoptic, 18 January 1994: (A) isotopic abundances and (B) D excess

for D enrichment, as a simple interpretation of the evaporation pan experiment data would suggest. On theother hand, the extended hyporheic zone isotope abundances are greatly dependent on the exchange, as thereis no change in the simulated extended hyporheic zone abundances for the no exchange case. The extendedhyporheic zone is also sensitive to the evaporation fractionation, as the resulting extended hyporheic zoneabundances are depleted, compared with the optimized results.

DISCUSSION

Conceptual model

We propose that the entire extensive wetted cross-sectional areas adjacent to Dry Valley streams are activelyexchanging hyporheic zones, comprised of a near-stream hyporheic zone and an extended hyporheic zone, eachexchanging over different time-scales. The near-stream hyporheic zone is well connected to the stream andexchanges relatively quickly with stream-water, on the scale of hours (Runkel et al., 1998). This near-streamzone experiences stream-water exchange on the time-scale of stream-water travel time from the glacier tothe lake, which is quantifiable with stream-tracer injection techniques. The second storage zone, the extendedhyporheic zone, represents the farther reaches of the wetted zone in which there is slow exchange with thestream. The connection between the stream and the extended hyporheic zone is characterized by longer flowpaths and longer time-scales of exchange. In this study, near-stream and extended hyporheic zones have beenlumped together as represented in the TS modelling, mainly because the data set did not include sufficientspatial resolution to test a multiple storage zone model in which these two zones could be distinguished.

Other studies have considered multiple storage zones with respect to stream solute transport. Castro andHornberger (1991) found evidence of multiple parallel flow paths, each with a unique time-scale of interaction

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1705

δD (

‰)

-258

-255

-252

-249

-246

-243(a)Col 20 vs Col 22 Col 20 vs Col 22 Col 28 vs Col 30 Col 28 vs Col 30 Col 28 vs Col 30 Col 28 vs Col 30

Distance Downstream (m)

0 100 200 300 400 500

δ18O

(‰

)

-33

-32

-31

-30

-29(b) Col 20 vs Col 23 Col 28 vs Col 23 Col 28 vs Col 31 Col 28 vs Col 31 Col 28 vs Col 31 Col 28 vs Col 31

initial stream observationinitial hyporheic observationfinal stream observationfinal hyporheic observationstream model outputhyporheic model output

initial stream observationinitial hyporheic observationfinal stream observationfinal hyporheic observationstream model outputhyporheic model output

Figure 10. Model input, goal and simulated abundances for (A) υD and (B) υ18O for a 384 h simulation. Storage zone start and endabundances are averages of observations from lateral wells A and B from each transect

with the stream. In an exercise to test multiple competing storage zones, Choi et al. (2000) found that multiplecompeting storage zones were difficult to distinguish unless they were associated with markedly different solutefluxes. The model represented here considers a single storage zone and identifies only long-term exchange.Several more intermediate storage zones may exist, and a model that incorporates a variety of time-scales ofexchange may be a more accurate representation of the real system. Such a model must be supported withappropriate data and the appropriate tracer. Stable isotopes proved to be an effective long-term tracer in thisstudy. The heterogeneity of the hyporheic zone is one significant challenge to be considered in designing astudy that would provide appropriate data to support a multiple nested storage zone model.

An alternative conceptual model with several nested storage zones would suggest some interaction betweenthe near-stream and extended hyporheic zones. This interaction probably would result in a non-linear

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1706 M. N. GOOSEFF ET AL.

Distance Downstream (m)0 100 200 300 400 500

δD (

‰)

-258

-255

-252

-249

-246 optimizedno evaporationno exchangeHyporheic abundances

Stream abundances

Figure 11. Alternative model outputs for assuming no evaporation and no hyporheic exchange (˛ D 0)

relationship between the stream and any of possibly several nested storage zones. The influence of the streamon a distant nested storage zone, with several intermediates in between, is complicated by the intermediatezones, whose interaction between each other would be difficult to detect.

Extended hyporheic zone interaction

Evidence of interaction between the stream and the extended hyporheic zone is provided by a comparisonof the synoptic isotope abundance ratios (Figure 8). From 21 December 1999 to 07 January 2000, there is adepletion of the υD and υ18O signals in the extended hyporheic zone. Evaporation of the extended hyporheicwaters would leave the remaining water more enriched, rather than depleted. Therefore, there must be someinflux of depleted stream water into the extended hyporheic zone, and assuming there is some limited mixingwithin the porous medium, the overall isotopic signal of extended hyporheic waters would be more depleted.We have modelled a steady flow condition, which was not observed (Figure 2) during the 1999–2000 flowseason on Green Creek. Hyporheic exchange is known to be driven by hydraulic head gradients (Harvey andBencala, 1993; Wondzell and Swanson, 1996), and it is likely that the high discharges in Green Creek (around28 December 1999) forced greater and more extensive hyporheic exchange than lower discharges. Our datado not have adequate temporal resolution to resolve such an effect.

The solute and isotope abundance data from the extended hyporheic zone suggest a long residence timecompared with the time-scale of stream-water advection. Extended hyporheic zone water is characterized byhigher dissolved solute concentrations (Figure 7) and much more enriched isotopic abundances (Figure 8).Longer flow paths that require longer residence times for stream water exchange provide enough time andsubstrate contact to allow dissolution of minerals and salts, and allow more time for evaporation to fractionatethe water in the extended hyporheic zone. We expect that hyporheic water enriches as a result of evaporationin the summer and sublimation during the winter, as well as from the seasonal freeze–thaw cycles that controlthe state of water within the hyporheic zone.

As expected, the large extended hyporheic zones that were defined in the TS modelling, exchanged withthe stream at rates lower than those found in either of the two stream-tracer experiments, and at generallylower rates than any previously published in hyporheic research literature, with the exception of Uvas Creek,reach number 4 (Bencala, 1984).

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1707

Re-assessed Uvas Creek model results using parameter estimation techniques found that an ˛ value of2Ð5 ð 10�5 s�1 for reach 4 resulted in a better simulation fit to the data (D. Scott, Landcare New Zealand,personal communication 2002). The results presented here yield values of the average time a moleculeremains in extended hyporheic storage �D AS/˛A) (Thackston and Schnelle, 1970) that range from 33Ð5 daysto 64Ð6 days, two orders of magnitude higher than 0Ð2 days, the greatest storage found in a stream-tracerexperiment in Green Creek in 1995, and generally larger than those reported by Runkel (2002) in temperatestreams. Exchange fluxes into and out of the extended hyporheic zone �D ˛A� (Harvey et al., 1996) rangefrom 7Ð08 ð 10�7 to 8Ð16 ð 10�7 m2 s�1, two orders of magnitude lower than those reported by Harveyet al. (1996) for St Kevin Gulch, and slightly higher than 7 ð 10�7 m2 s�1, the lowest possible exchange fluxcomputed for Green Creek (Table I).

Additionally, the results show that the fractionation resulting from evaporation is 10 times greater(1Ð5 ð 10�3‰ s�1) than the enrichment influence from exchange (1Ð0 ð 10�4‰ s�1), by computing therespective dυR/dt components of Equation (1). This is in contrast to the results presented in Table III, whichare based on stream transport times at the time of sampling, assuming a steady-state condition along thelength of the stream. The stream fractionation rates in Table III represent a very short time-scale, probablynot representative of the long-term influence of extended hyporheic zone interactions. Near-stream hyporheicexchange and variable climate conditions may be influencing the relative influence of mixing and evaporationon downstream isotopic enrichment on such short time-scales. Over the 17 day simulation, the stream-waterfractionation dependence also is obvious from Figure 11, as simulations with no evaporation kept streamisotope abundances much more depleted than those observed on 7 January 2000. There appears to be littlestream-water isotope abundance dependence on the extended hyporheic exchange as there is very little changein the results for the simulation without exchange. On the other hand, the extended hyporheic zone is obviouslygreatly dependent upon the exchange. For the simulation with no exchange, there is no change in the extendedhyporheic zone isotope abundances; yet for the optimized and no evaporation simulations, the extendedhyporheic zone isotope abundances react strongly, becoming more depleted, closer to the abundances of thestream water.

Given that the TS simulations are more sensitive to evaporation than exchange, it is important to note thatthe evaporation rate found here, 6Ð2 mm day�1 by evapoconcentration calculations, is larger than evaporationrates found by Simpson and Herczeg (1991) of 4Ð3 mm day�1 in the River Murray, Australia, 2Ð7 mm day�1

reported by Gibson et al. (1998) for pan experiments in Yukon, Canada, and 5 mm day�1 for a high Arcticlake (Woo, 1980). The pan evaporation rate found here also is lower than the potential evaporation rate of7Ð0 mm day�1, calculated using the energy balance method (Chow et al., 1988) for the period of modelling(21 December 1999 to 07 January 2000), assuming 45 W m�2 sensible heat flux and ground heat flux. Further,a slight change in the evaporation would not greatly change the resulting exchange parameter value in thisstudy. The conclusion would be similar—there is long-timescale exchange in the extended hyporheic zonesof these streams of the order of weeks or more.

Regressed values of υEV are generally similar to the values of υR for both D and 18O, consistent with ourestimates of stream evaporation. Values of υA for either D or 18O could not be found in the literature forthis area. It is expected that υA values should be similar to precipitation isotope abundance values. The meanisotope abundances of eight fresh snow samples from the Canada Glacier are �220‰ for D, and �27Ð9‰ for18O. The υA values found here are not very close to these means. An alternative assumption is that atmosphericmoisture is derived from the ocean, given the proximity to the ocean. Those abundances would be expectedto be much more enriched, probably closer to 0‰ for both D and 18O. In addition, Equation (3) assumesthat equilibrium conditions exist between the evaporating water and the atmosphere, which is not necessarilyappropriate for this field setting. Given these limitations, the TS model results do not change with respectto exchange. For the large extended hyporheic zone, the time-scale of exchange is much longer than thatcharacterized by solute tracer experiments.

One of the limitations of our modelling approach is the lack of temporal parameter change with respectto changes in stream flow. A definitive quantification of changes in TS parameters, dependent on flow rate,

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1708 M. N. GOOSEFF ET AL.

has not been found, although Harvey et al. (1996) note that low baseflow conditions generally allow for thebest characterization of hyporheic exchange parameters because in-stream storage is restricted. Therefore,time-varying flow over the transient simulation is not supported. Consistent with that approach, evaporationwas considered to be invariant with time, and the upstream boundary condition was linearly interpolatedbetween the start and end time observations. These limitations do not appear to confound our results, as longtime-scales of exchange and very low ˛ values were found.

Biogeochemical implications of extended hyporheic zone exchange

Understanding hyporheic and wetted zone dynamics in Dry Valley streams is essential to understandingthe biogeochemical reactions that define stream ecosystem function, and the extent to which these reactionsmay be taking place in transient storage zones. The extent to which the near-stream hyporheic zone isestablished allows for a better understanding of the time-scale of reactions that more immediately affect streamwater quality, such as nitrification, denitrification, biotic assimilation (McKnight et al., 1999) and chemicalweathering (Nezat et al., 2001; Gooseff et al., in press). Outer regions of the hyporheic zone presumably alsohost important biogeochemical reactions, such as chemical weathering, denitrification and DOC mineralization,which could come closer to a chemical equilibrium state than chemical reactions in the near-stream hyporheiczone, based on residence-time differences. The solute data show that the extended hyporheic zone waterhas higher concentrations of inorganic solutes than stream water (Figure 7). Evapoconcentration of extendedhyporheic waters is also a likely cause of increased solute concentrations. Extended hyporheic zones may nothave a great impact on stream chemistry as the hydrological fluxes are small, suggesting that the solute fluxesfrom the extended hyporheic zone also will be small.

This work is also significant with respect to the spatial discretization of the hyporheic zone into rapidlyexchanging near-stream zones that are in direct connection with the stream, and slower exchanging extendedzones, which are not in direct contact with the stream. If cross-sectional areas of the hyporheic zone derivedfrom typical stream-tracer experiment modelling are only a fraction of the total hyporheic zone (the measuredwetted zone, in this case) then erroneous conclusions of faster chemical reaction rates might be taking placein deceptively small areas. Higher rates of biogeochemical processes would help explain the ability of thisresilient aquatic ecosystem to proliferate despite low temperatures and short flow seasons.

CONCLUSIONS

The results presented here show that there is υD and υ18O enrichment of stream water as it flows from glacier tolake in the Dry Valleys, Antarctica. Evaporation of stream water plays an important role in this fractionation butis not solely responsible for the enrichment. Mixing with heavier water stored in the extended hyporheic zonealso contributes to the stream water isotopic signature, and the infiltration of lighter stream water is responsiblefor decreasing isotopic abundances in extended hyporheic waters over the flow season. In a novel approach, TSmodelling was used to derive characteristic extended hyporheic zone time-scales of exchange, using physicalmeasurements as parameter values. Computed hydrological hyporheic exchange fluxes were small, suggestingthat accompanying solute fluxes from extended hyporheic zones also may be small, despite generally higherconcentrations of solutes. The assumptions within our modelling approach allowed us to discern long time-scales of hyporheic exchange, which could not be detected with the conventional stream-tracer technique.

This work focused on the hyporheic exchange processes in streams of the Dry Valleys, a cold desertenvironment. Dry Valley streams provide a unique opportunity to study the use of natural tracers in streamsand extended hyporheic zones owing to the absence of terrestrial vegetation, precipitation and connectionwith a groundwater flow system. The tracing of stream-water flow paths both in and outside of the streamare important to many systems. A similar application might be made to relatively simple warm desert oralpine systems. Stable isotope tracing provides an alternative watershed tracing method to the introduction ofchemical tracers in sensitive watersheds.

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

LONG TIME-SCALE HYPORHEIC FLOW PATHS 1709

ACKNOWLEDGEMENTS

We are very grateful to Harry House, Ethan Chatfield, Jon Mason, Chris Jaros and Peter Conovitz for assistancein the field. Special thanks also goes to Kathy Welch for assistance with sample analysis. This work greatlybenefited from discussions with Berry Lyons, Mark Williams, Jim White, Gayle Dana and Ken Bencala,colleague reviews by Roy Haggerty and Don Campbell, and three anonymous reviewers. This work wassupported by NSF grants OPP-9211773 and OPP-9810219.

REFERENCES

Bencala KE. 1984. Interactions of solutes and streambed sediment 2. A dynamic analysis of coupled hydrologic and chemical processes thatdetermine solute transport. Water Resources Research 20(12): 1804–1814.

Bencala KE. 2000. Hyporheic zone hydrological processes. Hydrological Processes 14: 2797–2798.Bencala KE, Walters RA. 1983. Simulation of solute transport in a mountain pool-and-riffle stream: a transient storage model. Water

Resources Research 19(3): 718–724.Bomblies A. 1998. Climatic controls on streamflow generation from Antarctic glaciers. Master’s thesis, University of Colorado: Boulder,

Colorado.Castro NM, Hornberger GM. 1991. Surface–subsurface water interactions in an alluviated mountain stream channel. Water Resources

Research 27(7): 1613–1621.Chinn TJ. 1993. Physical hydrology of the dry valley lakes. In Physical and Biogeochemical Processes in Antarctic Lakes, Antarctic Research

Series, Vol. 59, Green WJ, Friedman EI (eds). American Geophysical Union: Washington, DC; 1–51.Choi J, Harvey JW, Conklin MH. 2000. Characterizing multiple timescales of stream and storage zone interaction that affect solute fate and

transport in streams. Water Resources Research 36(6): 1511–1518.Chow VT, Maidment DR, Mays LW. 1988. Applied Hydrology . McGraw-Hill: New York.Cirmo CP, McDonnell JJ. 1997. Linking the hydrologic and biogeochemical controls of nitrogen transport in near-stream zones of temperate-

forested catchments: a review. Journal of Hydrology 199: 88–120.Clow GD, McKay CP, Simmons GM, Wharton Jr RA. 1988. Climatological observations and predicted sublimation rates at Lake Hoare,

Antarctica. Journal of Climate 1(7): 715–728.Conovitz PA. 2000. Active layer dynamics and hyporheic zone storage in three streams in the McMurdo Dry Valleys, Antarctica. Master’s

thesis, Colorado State University: Fort Collins, CO.Conovitz PA, McKnight DM, MacDonald LH, Fountain AG, House HR. 1998. Hydrological processes influencing streamflow variation in

Fryxell Basin, Antarctica. In Ecosystem Dynamics in a Polar Desert: the McMurdo Dry Valleys, Antarctica, Antarctic Research Series,Vol. 73, Priscu JC (ed). American Geophysical Union: Washington, DC; 93–108.

Cooper LW, Olsen CR, Solomon DK, Larsen IL, Cook RB, Grebmeier JM. 1991. Stable isotopes of oxygen and natural and falloutradionuclides used for tracing runoff during snowmelt in an Arctic watershed. Water Resources Research 27(9): 2171–2179.

Cooper LW, Solis C, Kane DL, Hinzman LD. 1993. Application of oxygen-18 tracer techniques to Arctic hydrological processes. Arcticand Alpine Research 25(3): 247–255.

Craig H, Gordon LI. 1965. Deuterium and oxygen 18 variations in the ocean and marine atmosphere. Proceedings of the Conference onStable Isotopes in Oceanographic Studies and Paleotemperatures , Spoleto, Consiglio Nazionale delle Richerche, Piza: 9–130.

Epstein S, Sharp RP. 1959. Oxygen-isotope variations in the Malaspina and Saskatchewan Glaciers. Journal of Geology 67: 88–102.Germillion P, Wanielista M. 2000. Effects of evaporative enrichment on the stable isotope hydrology of a central Florida (USA) river.

Hydrological Processes 14: 1465–1484.Gibson JJ, Reid R, Spence C. 1998. A six-year isotopic record of lake evaporation at a mine site in the Canadian subarctic: results and

validation. Hydrological Processes 12: 1779–1792.Gooseff MN, Lyons WB, Blum AE, McKnight DM. In press. Weathering reactions and hyporheic exchange controlling stream water

chemistry in a glacial meltwater stream in the McMurdo Dry Valleys. Water Resources Research .Harvey JW, Bencala KE. 1993. The effect of streambed topography on surface–subsurface water exchange in mountain catchments. Water

Resources Research 29(1): 89–98.Harvey JW, Wagner BJ, Bencala KE. 1996. Evaluating the reliability of the stream tracer approach to characterize stream–subsurface water

exchange. Water Resources Research 32: 2441–2451.Hooper RP, Shoemaker CA. 1986. A comparison of chemical and isotopic hydrograph separation. Water Resources Research 22: 1444–1454.Johnson TM, DePaolo DJ. 1997. Rapid exchange effects on isotope ratios in groundwater systems 1. Development of a transport–dissolution–

exchange model. Water Resources Research 33: 187–195.Kendall C, Caldwell EA. 1998. Fundamentals of isotope geochemistry. In Isotope Tracers in Catchment Hydrology , Kendall C, McDonnell JJ

(eds). Elsevier: New York; 51–84.Kennedy VC. 1971. Silica variation in stream water with time and discharge. In Nonequilibrium Systems in Natural Water Chemistry .

Advances in Chemistry Series No. 106, American Chemical Society: Washington, DC; 94–130.Kennedy VC, Kendall C, Zellwger GW, Wyerman TA, Avanzino RJ. 1986. Determination of the components of stormflow using water

chemistry and environmental isotopes, Mattole River basin, California. Journal of Hydrology 84: 107–140.Krabbenhoft DP, Bowser CJ, Anderson MP, Valley JW. 1990. Estimating groundwater exchange with lakes 1. The stable isotope mass

balance method. Water Resources Research 26: 2445–2453.

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)

1710 M. N. GOOSEFF ET AL.

Lyons WB, Welch KA, Nezat CA, Crick K, Toxey JK, Mastrine JA, McKnight DM. 1997. Chemical weathering rates and reactions inthe Lake Fryxell Basin, Taylor Valley: comparison to temperate river basins. In Ecosystem Processes in Antarctic Ice-free Landscapes ,Lyons WB, Howard-Williams C, Hawes I (eds). Balkema Press: Rotterdam; 147–154.

Maule CP, Stein J. 1990. Hydrologic flow path definition and partitioning of spring meltwater. Water Resources Research 26(12): 2959–2970.Maurice PA, McKnight DM, Leff L, Fulghum JE, Gooseff M. 2002. Direct observations of aluminosilicate weathering in the hyporheic zone

of an Antarctic Dry Valley stream. Geochimica et Cosmochimica Acta 66(8): 1335–1347.McDonnell JJ, Stewart MK, Owens IF. 1991. Effect of catchment-scale subsurface mixing on stream isotopic response. Water Resources

Research 27(12): 3065–3073.McKenna SA, Ingraham NL, Jacobson RL, Cochran GF. 1992. A stable isotope study of bank storage mechanisms in the Truckee river

basin. Journal of Hydrology 134: 203–219.McKnight DM, Niyogi DK, Alger AS, Bomblies A, Conovitz PA, Tate CM. 1999. Dry Valley streams in Antarctica: ecosystems waiting

for water. Bioscience 49(12): 985–995.McLean R, Oswood MW, Irons JG, McDowell WH. 1999. The effect of permafrost on stream biogeochemistry: a case study of two streams

in the Alaskan (USA) taiga. Biogeochemistry 47(3): 239–267.McNamara JP, Kane DL, Hinzman LD. 1997. Hydrograph separations in an Arctic watershed using mixing model and graphical techniques.

Water Resources Research 33(7): 1707–1719.Mulholland PJ, Marzolf ER, Webster JR, Hart DR, Hendricks SP. 1997. Evidence that hyporheic zones increase heterotrophic metabolism

and phosphorus uptake in forest streams. Limnology and Oceanography 42: 443–451.Mullin JB, Riley JP. 1955. The colorimetric determination of silicate with special reference to sea and natural waters. Analytica Chimica

Acta 12: 162–176.Nezat CA, Lyons WB, Welch KA. 2001. Chemical weathering in streams of a polar desert (Taylor Valley, Antarctica). Geological Society

of America Bulletin 113(11): 1401–1408.Pilgrim DH, Huff DD, Steele TD. 1979. Use of specific conductance and contact time relations for separating flow components in storm

runoff. Water Resources Research 15(2): 329–339.Poeter EP, Hill MC. 1998. Documentation of UCODE, a Computer Code for Universal Inverse Modeling. Water Resources Investigation

Report 98–4080, U.S. Geological Survey: Denver, CO.Runkel RL. 1998. One-dimensional Transport with Inflow and Storage (OTIS): a Solute Transport Model for Streams and Rivers. Water

Resources Investigation Report 98–4018, U.S. Geological Survey: Denver, CO.Runkel RL. 2002. A new metric for determining the importance of transient storage. Journal of the North American Benthological Society

21(4): 529–543.Runkel RL, McKnight DM, Andrews ED. 1998. Analysis of transient storage subject to unsteady flow: diel flow variation in an Antarctic

stream. Journal of the North American Benthological Society 17: 143–154.Simpson HJ, Herczeg AL. 1991. Stable isotopes as an indicator of evaporation in the River Murray, Australia. Water Resources Research

27(8): 1925–1935.Stewart MK, McDonnell JJ. 1991. Modeling base flow soil water residence times from Deuterium concentrations. Water Resources Research

27(10): 2681–2693.Stout GE (ed.). 1967. Isotope Techniques in the Hydrologic Cycle. Geophysical Monograph Series, Number 11, American Geophysical

Union: Washington, DC; 199.Thackston EL, Schnelle KB. 1970. Predicting the effects of dead zones on stream mixing. Journal of the Sanitary Engineering Division,

American Society of Civil Engineers 96: 319–331.Theakstone WH, Knudsen NT. 1996. Isotopic and ionic variations in glacier river water during three contrasting ablation seasons.

Hydrological Processes 10: 523–539.Turner JV, Macpherson DK, Stokes RA. 1987. The mechanisms of catchment flow processes using natural variations in deuterium and

oxygen-18. Journal of Hydrology 94: 143–162.Vaughn BH, White JWC, Delmotte M, Trolier M, Cattani O, Stievenard M. 1998. An automated system for hydrogen isotope analysis of

water. Chemical Geology, Isotope Geoscience Section 152: 309–319.Von Guerard P, McKnight DM, Harnish RA, Gartner JW, Andrews ED. 1995. Streamflow, water-temperature, and specific-conductance

data for selected streams draining into Lake Fryxell, Lower Taylor Valley, Victoria Land, Antarctica, 1990–92. U.S. Geological Survey,Open-File Report 94–545, U.S. Geological Survey: Denver, CO.

Welch KA, Lyons WB, Graham E, Neumann K, Thomas JM, Mikesell D. 1996. Determination of major element chemistry in terrestrialwaters from Antarctica by ion chromatography. Journal of Chromatography A 739: 257–263.

White DS. 1993. Perspectives on defining and delineating hyporheic zones. Journal of the North American Benthological Society 12(1):61–69.

Wondzell SM, Swanson FJ. 1996. Seasonal and storm dynamics of the hyporheic zone of a 4th-order mountain stream I: hydrologic processes.Journal of the North American Benthological Society 15(1): 3–19.

Woo M-K. 1980. Hydrology of a small lake in the Canadian high arctic. Arctic and Alpine Research 12(2): 227–235.

Copyright 2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1691–1710 (2003)