Isotopic exchange between snow and atmospheric water vapor: Estimation of the snowmelt component of groundwater recharge in the southwestern United States

Isotopic exchange between snow and atmospheric water vapor:

Estimation of the snowmelt component of groundwater recharge in the

southwestern United States

Sam Earman,1,2 Andrew R. Campbell,3 Fred M. Phillips,3 and Brent D. Newman4

Received 6 July 2005; revised 12 December 2005; accepted 26 January 2006; published 3 May 2006.

[1] The contribution of snowmelt to groundwater recharge at four sites in thesouthwestern United States was evaluated using stable isotopes of O and H. Pairedprecipitation collectors were installed at the study sites; data show that (1) there is often asignificant difference between the stable isotope composition of fresh snow and the bulkmeltwater derived from it (this suggests that using the isotope composition of high-elevation springs as a proxy for precipitation may not be sound if snow is a rechargesource) and (2) collector design can significantly influence the stable isotope compositionof collected snow. Because the isotope composition of snow from a given locationbecomes heavier (i.e., more rain-like) with increased exposure, using bulk snowmeltcompositions to calculate input to groundwater recharge results in significantly increasedestimates of snowmelt contributions to recharge (compared to estimates derived from freshsnow signatures). Snowmelt provides at least 40–70% of groundwater recharge at thestudy sites, although only 25–50% of average annual precipitation falls as snow. On thebasis of these results and presently accepted scenarios for alterations in precipitation in thewestern United States over the next 50 years (significantly decreased snowpack due toincreased atmospheric CO2), investigations of how climate change may affectgroundwater resources are needed. We also investigated the potential for snow/atmospheric water vapor isotope exchange to influence the isotope signature of snow(which has been a subject of debate); the results of a laboratory experiment suggest that itcan drive significant shifts in the isotope signature of snow, even at temperatures below0�C.Citation: Earman, S., A. R. Campbell, F. M. Phillips, and B. D. Newman (2006), Isotopic exchange between snow and atmospheric

water vapor: Estimation of the snowmelt component of groundwater recharge in the southwestern United States, J. Geophys. Res., 111,

D09302, doi:10.1029/2005JD006470.

1. Introduction

[2] Snowmelt is an important contributor to groundwaterrecharge in many areas of the world, including the south-western United States (referred to hereafter as ‘‘the South-west’’). In the Southwest, low-lying basin floors arecommonly surrounded by mountain ranges. The high eleva-tion of the mountain ranges relative to valley floors results inlower temperatures in the ranges, producing higher averageannual precipitation, increasing the percentage of total pre-cipitation falling as snow, and lowering evapotranspiration.[3] A basic examination of the potential for groundwater

recharge from snowmelt suggests that snow is likely to exert

a stronger influence on groundwater recharge than sug-gested by its volumetric contribution to average annualprecipitation. Wilson et al. [1980] provide a summary ofattributes of snow and rain that suggests snowmelt is morelikely to become recharge than rainfall. Summer stormevents in the Southwest tend to be of high intensity andshort duration, which is likely to produce a relatively highpercentage of overland flow versus infiltration. In contrast,snowmelt is a process of relatively low intensity but longduration, providing a pulse of water likely to have arelatively high ratio of infiltration to runoff. Summer rainsfall when temperatures are high and vegetation is at its mostactive, meaning evapotranspiration is at or near a maximum.Snowmelt occurs when vegetation is mostly dormant andtemperatures are low, minimizing evapotranspiration (therelatively high albedo of snow compared to bare or vege-tated ground contributes to this as well). Finally, stormsproducing snow tend to blanket fairly wide areas, in contrastto summer rains, which can have spotty coverage. Flint etal. [2004] describe recharge as taking place when enoughwater is present to exceed both the storage capacity of a soiland potential evapotranspiration; they state that the accu-

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111, D09302, doi:10.1029/2005JD006470, 2006

1Division of Hydrologic Sciences, Desert Research Institute, Reno,Nevada, USA.

2Formerly at Earth and Environmental Science Department, NewMexico Institute of Mining and Technology, Socorro, New Mexico, USA.

3Earth and Environmental Science Department, New Mexico Instituteof Mining and Technology, Socorro, New Mexico, USA.

4Earth and Environmental Sciences Division, Los Alamos NationalLaboratory, Los Alamos, New Mexico, USA.

Copyright 2006 by the American Geophysical Union.0148-0227/06/2005JD006470

D09302 1 of 18

mulation of snow over a months-long period can oftenprovide sufficient moisture to accomplish this during theperiod of snowmelt, whereas intermittent rain events oftendo not provide enough moisture to do so.[4] Population growth in the Southwest is placing in-

creasing demands on groundwater resources, so sound long-term management of these resources is essential. In turn,good aquifer management requires an understanding of howgroundwater is recharged, including the sources of input.Thus knowledge of the contribution of snow (especiallymountain snow) to the replenishment of alluvial aquifers isa key to sound water resources planning.[5] In section 2 of this paper, we examine the use of

stable isotopes to determine snowmelt contributions torecharge; in section 3, we examine some of the processesthat contribute to the alteration of the stable isotope signa-ture of snow in the period between snowfall and theconclusion of melt; in section 4, we investigate the potentialfor isotope exchange between snow and atmospheric watervapor to affect the isotope signature of snow at temperaturesbelow 0�C.

2. Snowmelt Contributions to Recharge

2.1. Background

[6] The stable isotope composition of precipitation tendsto plot on or near the global meteoric water line (GMWL),but its location on the line is a function of temperature andother related effects [Dansgaard, 1964]. On the globalscale, average annual precipitation in warm areas is isoto-pically heavier than that in cool areas. On a local scale,areas with distinct seasonality receive rain that is isotopi-cally heavier than snow [Clark and Fritz, 1997]. If theisotope compositions of rain, snow, and locally rechargedgroundwater from a given area are known, a mixing modelcan theoretically be used to determine the proportions ofrain and snow that contribute to groundwater recharge[Hershey, 1989; Maule et al., 1994]. However, care mustbe taken in defining the end-member compositions, becausethe isotope composition of snow tends to change with timeon the ground.[7] The physical character of snow changes from the time

of its fall to the conclusion of its melt. As described byColbeck [1987], the process of snow metamorphism affectsthe physical structure of snow, with snow crystals typicallybecoming larger and more rounded as time progresses.While rainwater falling on the snowpack can drive snowmetamorphism, it is not a necessary condition. This isimportant because some of the areas investigated in thisstudy typically receive little or no spring rain. Snowmetamorphism can result solely from mass transport drivenby temperature gradients in snowpacks that result fromdiffering temperatures at the snowpack surface and base[Colbeck, 1987]. In addition to temperature gradient meta-morphism, and rain-on-snow events, snowpacks can poten-tially be affected by processes such as evaporation/sublimation, partial melt and refreezing, condensation ofatmospheric water vapor on the snow surface, and isotopeexchange with atmospheric and soil water vapor.[8] Because physical changes impart fractionation in

stable isotope systems, processes such as those describedabove cause the isotope content of snow to change; these

changes tend to be ongoing during exposure and melt[Stichler, 1987]. As a result, an isotope composition deter-mined for fresh snow (or determined for a snowpack on onespecific day during its exposure) is not necessarily indica-tive of the composition of the meltwater that will be derivedfrom the snow. Similarly, a sample of meltwater collectedon a given day might not be representative of melt fromother days, nor of the mean composition of meltwaterderived from the snowpack.[9] Most of the investigations of changes in the isotope

composition of snow and snowmelt have focused either onthe snow itself [e.g., Stichler et al., 1981; Stichler, 1987;Sommerfeld et al., 1991; Unnikrishna et al., 2002] or onapplication of the phenomenon to hydrograph separation[e.g., Herrmann et al., 1981; Rodhe, 1981]. Relatively fewinvestigations have incorporated isotope effects on snow-pack into the determination of snowmelt contributions togroundwater recharge [e.g., Winograd et al., 1998; Rose,2003].[10] Because the stable isotope composition of snowmelt

varies with time, for some surface water studies, it may benecessary to collect meltwater samples on a frequent basisto get an accurate picture of the input function into thesystem of interest [Unnikrishna et al., 2002]. For ground-water studies, however, it may only be necessary to know abulk input value. Of course, this depends on the nature ofthe system. Water from springs and wells near rechargeareas or in fracture-flow or karst systems can displaysignificant variations in chemical and isotopic compositionon short timescales. In these cases, a knowledge of isotopeinput on a fine timescale may be necessary, similar tosurface water systems. In contrast, for many observationpoints, the stable isotope content of water from springs orwells remains fairly constant over time. If such an obser-vation point is used, then a ‘‘bulk input’’ value for stableisotope composition is sufficient to characterize the system,and information regarding short-term variations is notneeded (for discussion of variation in springwater stableisotope content, see Ingraham et al. [1991] and Newman etal. [2001]).

2.2. Methods

[11] This study was designed to determine the stableisotope composition of bulk snowmelt input to groundwatersystems, to determine how different this composition isfrom the fresh snow composition. Sets of precipitationcollectors of different designs were used to determine thebulk melt composition and its relation to the fresh snowcomposition.2.2.1. Determination of the Bulk MeltwaterComposition and Its Difference From theFresh Snow Composition[12] Most of the samplers currently in use for determining

the stable isotope composition of precipitation are based oncollectors described by Friedman et al. [1992], or aresimilar to one of the collector types they describe. For astudy of stable isotope composition of precipitation insoutheastern California, Friedman et al. [1992] employedthree types of collectors (Figure 1). For sites where no snowor only a moderate amount of snow was expected, one oftwo versions of a ‘‘funnel collector’’ was emplaced. In bothversions, the funnel was supported approximately 1 m off

D09302 EARMAN ET AL.: SNOWMELT IN GROUNDWATER RECHARGE

2 of 18

D09302

the ground by a stake, and connected by tubing to anunderground reservoir containing ‘‘hydrocarbon oil.’’ Forsites where no snow was expected, a glass funnel ofstandard design (60� slope, diameter between 50 and100 mm) was employed; for sites expected to receivemoderate amounts of snow in addition to rain, whitepolypropylene Buchner funnels were used (these arereferred to here as a ‘‘regular funnel collectors’’). Theporous plates from the Buchner funnels were removedand replaced with metal screens; the edges of the Buchnerfunnels extended 75 mm above the screens, but some werefitted with 150-mm ‘‘plastic’’ extension tubes, the color ofwhich is not described (these collectors are referred to here as‘‘extended funnel collectors’’). Finally, at sites where largeamounts of snow were expected in addition to rain, a regularfunnel collector was employed along with a snow collector(referred to here as a ‘‘PVC collector’’). The PVC (polyvinylchloride) collector consisted of a 19-cm-diameter plasticpipe, approximately 1.25 m high, sealed at the bottom, andburied about 0.3 m in the ground, with a layer of hydrocarbonoil in the bottom. At the dual-collector sites, the collectorused was alternated on the basis of the season. The regularfunnel collector was covered during the winter, so sampleswere only collected in the PVC collector during that time; inthe summer, the PVC collector was covered, and sampleswere collected only in the regular funnel collector.[13] The size of the holes in the metal screens in the

Buchner funnels of the collectors used at Friedman et al.’s

[1992] ‘‘moderate snow’’ sites is not described, but becausethe 150-mm extension tubes were fitted in order to ‘‘in-crease the amount of snow that they could hold beforemelting occurred,’’ it is clear that the screen providedsufficient obstruction to allow the accumulation of snowin the funnel body. As a result, snow falling into thecollector would not be protected from metamorphic pro-cesses by the hydrocarbon oil until after it had melted. Rain,on the other hand, should have been able to flow throughthe metal screen and be protected almost instantly by thehydrocarbon oil. The 19-cm diameter of the PVC collector’sopening would allow snow to enter immediately, evenduring heavy events. The low density of the oil (the densityof that used in our collectors was 0.825 g/cm3) should havecaused snow to sink below the oil before any significantpostfall isotope alteration could occur.2.2.2. A New Scheme for Deploying PairedPrecipitation Collectors at a Single Site to Determinethe Degree of Isotope Alteration in Snow andEstimate the Isotope Composition of Bulk InfiltrationDerived From Snowmelt[14] Although Friedman et al. [1992] used data from the

funnel collectors and the PVC collectors as though they wereintercomparable, it is apparent that the PVC collectorsprovided immediate protection from metamorphic processes,while the funnel collectors likely allowed metamorphismprior to protection of liquid melt by the hydrocarbon oil.Although a data set that relies on different types of collectorsat different sites is not ideal for interpreting spatial variationsof stable isotopes in fresh precipitation, using paired collec-tors at a given site could provide an easy means of determin-ing the degree of isotope metamorphism that snow undergoesand also the bulk input signal from snowmelt.[15] By installing a funnel collector (regular or extended)

and a PVC collector at a single site, and allowing both tooperate throughout the year, two sets of data can becollected. The PVC collector should maintain the stableisotope composition of fresh precipitation, regardless ofwhether it is rain or snow. The funnel collector shouldmaintain the stable isotope composition of fresh rain, but theplate or screen in the Buchner funnel will act as a partialanalog for the ground. Because water will flow into the oil-covered reservoir only after snowmelt events, the snowpacktrapped in the funnel body will be subjected to the sameatmospheric interactions as natural snowpack in the area(including sublimation of snow; evaporation of meltwater;snow/atmosphere water/atmosphere, and snow/water iso-tope exchange). Because of the collector design, input ofheat from the soil to the snowpack and snowpack/soil gasexchange are prevented. The resultant melt should provide abetter approximation of the melt that would infiltrate theground surface than samples of fresh snowpack.[16] Sets of precipitation collectors were installed at eight

sites in Arizona and New Mexico (Figure 2). A PVCcollector was installed at every site, and at least one typeof funnel collector was also installed. At some sites, bothregular and extended funnel collectors were operated inaddition to a PVC collector. In the Chiricahua Mountainsand Los Alamos, collector sets were installed at highelevations (as close to the crest of the range as reasonablypossible given installation and collection logistics; 3017 mand 2682 m above sea level, respectively) and at low

Figure 1. Schematic representation of precipitation col-lectors used in this study, modeled after Friedman et al.[1992].


3 of 18

D09302

elevations (1640 m and 2140 m, respectively). In theMagdalena Mountains, three collector sets were installed,at high (3243 m), low (2140 m), and middle (2560 m)elevations in the range. A single set of collectors wasemplaced near the crest of the Santa Catalina Mountains(2791 m). The collectors were sampled twice a year: once inthe spring, yielding integrated samples from the winterseason, and once in the fall, yielding integrated samplesfor the summer season.[17] All samples for d18O and dD analysis were preserved

in the field by means of the mineral oil in the reservoir.Water samples were separated from mineral oil in thelaboratory, and then stored in glass bottles with Poly-Seallids until analysis. d18O and dD (relative to Vienna StandardMean Ocean Water) were determined via gas source massspectrometry. All d18O results were obtained via analysis ofCO2 gas that had been equilibrated with a sample aliquot(similar to the methodology outlined by Clark and Fritz[1997] and discussed in greater detail by Socki et al.[1992]). dD values were obtained by analyzing hydrogengas formed during an oxidation-reduction reaction involv-ing the water sample and metal [Clark and Fritz, 1997].During the initial portion of the study, a sample aliquot wasreacted with zinc shavings at 450�C for 30 min [Coleman etal., 1982]; later sample aliquots were reacted with powderedchromium at 750�C for 60 s [Nelson and Dettman, 2001].The change in methodology coincided with the transition toa new mass spectrometer. Multiple samples run using theold instrument and methodology were rerun (for both d18Oand dD) on the new instrument to insure that values

measured with the new instrument and methodology wereidentical (within analytical error).

2.3. Results

[18] The types of collectors deployed at each site for eachseason are summarized in Table 1. Data from the precipi-tation collectors are shown in Table 2 and Figures 3 and 4.The data shown in Figure 3 represent two winter collectionseasons and one summer collection season; data collectedduring the winter of 2002–2003 are shown in Figure 4. Notall stations were activated at the same time, so some siteshave no data for a given time period.2.3.1. Winter 2002–2003 Data[19] For all sites, water in the reservoirs of the regular

funnel collector was enriched in dD and d18O compared tothat in the PVC collector (Figure 4). At the Los AlamosHigh site, the water in the extended funnel collector wasenriched compared to that in the regular funnel collector.The degree of isotopic enrichment observed in the extendedfunnel collector relative to the regular funnel collector wassimilar, but slightly less than the degree of enrichmentobserved between the regular funnel collector and thePVC collector. In all cases, the slope of the lines connectingthe samplers for a site is between 5.1 and 7.4. At the twolow-elevation sites and the Chiricahua High site, the degreeof isotopic enrichment was relatively small; the Los AlamosHigh and Magdalena High sites both showed relativelylarge enrichments.[20] The amount of isotopic enrichment is related to the

amount of time the snow was exposed between snowfall

Figure 2. Base map: continental United States, with New Mexico and Arizona shown in grey. Zoommap: locations of precipitation collector sets in New Mexico and Arizona, United States. Location LA(Los Alamos) has two sites: Los Alamos Low at 2140 m and Los Alamos High at 2682 m. Location Mag.(Magdalena Mountains) has three sites: Magdalena Low at 2149 m, Magdalena Mid at 2560 m, andMagdalena High at 3243 m. Location Chir. (Chiricahua Mountains) has two sites: Chiricahua Low at1640 m and Chiricahua High at 3017 m. Location Stew. (Santa Catalina Mountains) has one site(Steward) at 2791 m.


4 of 18

D09302

and snowmelt. Because temperature and amount of precip-itation are related to elevation, the low-elevation sites tendto receive less snow than the high-elevation sites, and thesnow is subjected to higher temperatures, thus speedingmelt. The Chiricahua Low site is an exception, as itexperienced nearly the same degree of isotopic enrichmentas the Chiricahua High site. The Chiricahua Mountains hadan unusually warm and dry winter, and because of itslocation, this site typically has warmer temperatures at agiven elevation than the Magdalena and Los Alamos sites.Because little snow fell in the Chiricahuas that winter andtemperatures were high, relatively little isotopic enrichmentoccurred, even at high elevation.2.3.2. All Data[21] The data from all precipitation collectors (Figure 3)

show that, as expected, the summer precipitation composi-tions are enriched compared to the winter precipitationcompositions. During the summer collection, all precipita-tion fell as rain, and, as predicted on the basis of collectordesign, most sites show only minimal differences betweencollector types. Much of the difference at these sites isattributable to the fact that the different collector types werenot at exactly the same elevation above ground surface andhad slightly differently sized openings, so collector efficiencywas not identical; analytical error may also be responsible forsome of the difference. The Magdalena High site did show asignificant difference between the PVC and regular funnelcollector samples. At the time of sampling, it was found thatthe regular funnel collector had been disturbed from itsupright orientation. The tilted position may have affectedthe isotope composition of the collected precipitation in two

Table 1. Listing of Collector Types Active at the Study Sites

During Given Time Periods

Site and Collector TypeaWinter

2002–2003Summer2003

Winter2003–2004

Chiricahua High PVC X X XChiricahua High RF X X XChiricahua Low PVC X X XChiricahua Low RF X XChiricahua Low EF (black) XLos Alamos High PVC X X XLos Alamos High RF X X XLos Alamos High EF (black) X X XLos Alamos Low PVC X X XLA Low RF X XLos Alamos Low EF (white) XMagdalena High PVC X X XMagdalena High RF X X XMagdalena High EF (black) XMagdalena High EF (white) XMagdalena Mid PVC b XMagdalena Mid RF XMagdalena Mid EF (white) XMagdalena Low PVC X XMagdalena Low RF XMagdalena Low EF (white) XSteward PVC X XSteward RF XSteward EF (white) XSteward EF (black) X

aPVC, PVC collector; RF, regular funnel collector; EF, extended funnelcollector.

bA PVC collector was in place at the Magdalena Mid site in summer2003, but the sample was not analyzed because of the presence of a deadanimal.

Table 2. The dD and d18O of Samples From Precipitation

Collectors Used in This Study

Site and Collector Typea

Winter2002–2003 Summer 2003

Winter2003–2004

d18O,%

dD,%

d18O,%

dD,%

d18O,%

dD,%

Chir High PVC �10.9 �69.9 �9.5 �62.1 �10.3 �71.7Chir High RF �10.5 �67.4 �9.3 �59.1 �10.2 �68.9Chir Low PVC �9.1 �60.5 �6.9 �51.6 �11.5 �77.5Chir Low RF �8.7 �58.3 �6.8 �51.1 X XChir Low EF (black) X X X X �12.1 �77.9LA High PVC �16.6 �117.7 �8.8 �54.25 �16.1 �110.0LA High RF �14.8 �107.0 �8.5 �53.3 �14.1 �102.9LA High EF (black) �13.5 �98.2 �8.3 �50.4 �14.1 �102.1LA Low PVC �15.0 �111.1 �8.1 �50.1 �15.5 �106.9LA Low RF �14.4 �106.6 �7.6 �49.2 X XLA Low EF (white) X X X X �15.81 �107.8Mag High PVC �13.2 �94.6 �9.2 �60.4 �12.1 �83.1Mag High RF �11.95 �85.3 �8.2 �56.4 �11.1 �74.7Mag High EF (black) X X X X �11.3 �73.3Mag High EF (white) X X X X �11.2 �74.6Mag Mid PVC X X X X �12.3 �79.3Mag Mid RF X X �9.2 �64.5 X XMag Mid EF (white) X X X X �11.9 �77.9Mag Low PVC X X �7.8 �55.7 �11.5 �75.6Mag Low RF X X �8.0 �56.7 �11.4 �75.1Mag Low EF (white) X X X X �11.4 �75.1Stew PVC X X �7.5 �56.5 �11.4 �72.8Stew RF X X �7.8 �58.5 X XStew EF (white) X X X X �10.7 �73.1Stew EF (black) X X X X �10.8 �73.5

aChir, Chiricahua; LA, Los Alamos; Mag, Magdalena; Stew, Steward.

Figure 3. Plot of stable isotope data for all samplings ofprecipitation collectors used in this study. For each station,the open symbols represent winter precipitation from 2002–2003, the grey-shaded symbols represent precipitation fromsummer 2003, and the solid symbols represent precipitationfrom winter 2003–2004. Station abbreviations used in thelegend are as follows: Chir, Chiricahua; LA, Los Alamos;Mag, Magdalena; Stew, Steward.


5 of 18

D09302

ways. First, the tilt imposed on the collector may havethinned the layer of mineral oil covering the collectedprecipitation at the highest point in the reservoir such thatevaporation could have taken place. Friedman et al. [1992]show that a layer of mineral oil in a collection reservoir willprevent evaporation as long as it is of sufficient thickness;although mineral oil was added in excess of the amountneeded to achieve this condition if the collector remainedupright, the tilting of the collector may have negated this.Second, the Magdalena High site is in a treeless area at thecrest of a large mountain range, and wind gusts in excess of160 km/h have been recorded in the area. If one or more rainevents occurred in conjunction with strong winds, the resul-tant directionality of the rain may have caused greatercollection efficiency for a given event or events by onesampler versus the other. The minimal differences betweencollectors for the summer samples at most sites show that thelarger variations observed for the winter 2002–2003 samplesare not attributable to differences between the collector types.[22] The degree of alteration observed for the winter

2003–2004 is typically less than that seen for the previouswinter (Figure 3). For the two sites that had all three collectortypes (Magdalena High and Los Alamos High), isotopicenrichment was observed for the funnel collector samplesrelative to the PVC collector samples, but there was noenrichment observed between the regular and extendedfunnel collector samples. This suggests that snow eventswere such that the extended funnel collectors did not accu-mulate more snow than the regular funnel collectors, and thusdid not offer an opportunity for increased exposure time. Twosites (Magdalena High and Steward) had both black andwhite extended funnel collectors to test the hypothesis thatmore rapid melt in the black collector would yield less

isotopic enrichment than observed in the white collector.However, the samples from these collectors were nearlyidentical isotopically. The 2003–2004 winter was unusuallywarm and dry throughout the study region, so less snow andhigher temperatures minimized the alteration at most sites. Inaddition, while precipitation collected during the winter of2002–2003 contained only minimal rain, significantamounts of rain were collected in the winter 2003–2004sampling because of unusual spring rains. Because rainfallinputs water of nearly identical isotope composition into bothcollector reservoirs, significant amounts of rain would lessenthe difference in isotope content between the collectors thatwould have been observed had only snow been collected.

2.4. Discussion

2.4.1. Groundwater Recharge Derived From Snowmelt[23] Several important concepts related to tracking re-

charge sources using stable isotopes are illustrated inFigures 3 and 4. If snow makes up a significant proportionof precipitation, different types of precipitation collectorscan yield significantly different values for the isotopiccomposition of precipitation at the same site. As a result,comparing isotope compositions from different collectortypes may not be valid. In addition, certain collector designsappear to preserve an isotope composition other than that offresh snow. If the goal of a study is to measure the isotopecomposition of precipitation, collector design should becarefully considered if snow is a significant portion of theprecipitation. Finally, even when significant isotopic enrich-ment of the fresh precipitation compositions takes place, itmay not be obvious from examination of a sample’s stableisotope signature. Because of the relatively high slopes ofthe isotopic enrichments on the dD-d18O plot of Figure 4, inno case does an enriched sample plot far enough to the rightof the global line to raise concern about its validity as afresh precipitation sample. In fact, at four of the fivelocations, the isotopically enriched sample plots closer tothe global line than does the fresh precipitation sample.[24] The isotopic enrichment slopes observed are higher

than would be expected if the enrichment process weredominated by a kinetically influenced process such asevaporation/sublimation. As evaporation/sublimation un-doubtedly affected the snow at these sites to some degree[e.g., Lawson, 1997; Leydecker and Melack, 1999], thissuggests that some other process or processes are operatingas well, and that these processes are not kinetically influ-enced. This issue will be discussed further in sections 3 and4 of this paper.[25] The data gathered for this study were used in

conjunction with other records of isotopes in precipitationand groundwaters from the study sites [Gross and Wilcox,1983; Adams et al., 1995; Blake et al., 1995; Cunninghamet al., 1998; Wright, 2001; Earman et al., 2003] to estimatesnowmelt contributions to groundwater recharge at the fourstudy sites. The results for all sites are shown in Table 3; anillustration of the methodology used is provided for the LosAlamos High site.2.4.2. Los Alamos High-Elevation Site[26] Data collected at the Los Alamos high-elevation site

in the winter of 2002–2003 are shown in Figure 5a; datafrom the winter of 2003–2004 are shown in Figure 5b. Theellipse labeled ‘‘Residual Snow’’ in Figure 5a gives the

Figure 4. Plot of stable isotope data from winter 2002–2003, shown with the global meteoric water line. For eachsite, the PVC collector is represented by a solid symbol, theregular funnel collector is represented by a grey symbol,and the only extended funnel collector operating that season(Los Alamos High site) is shown with an open symbol.


6 of 18

D09302

range of values for the small (<30 cm in diameter) patchesof snow remaining on the ground when melt was nearlyconcluded, and the vast majority of snowpack was nolonger present. Samples were collected on 11 April, at theconclusion of the main winter snowpack melt, and on 19May, at the conclusion of the melt of a late-season snow thatfell after the main snowpack had completely melted. Re-sidual snow was not collected in 2004.[27] The fresh snow from winter 2002–2003 (as sampled

in the PVC collector) is heavy compared to the centroid ofthe ellipse representing the range of winter precipitationvalues, but still within the observed range. The alteration ofthe snow in the regular and extended funnel collectorsshifted the isotope composition to a significantly heaviervalue, much closer to the mean value for groundwater thanis the fresh snow signature. The values observed for theresidual snow indicate that snow continues to become isoto-pically enriched until melt is complete, and that greater

exposure times result in increased enrichment. The continuedisotopic enrichment follows the same trend as the threeprecipitation collectors (the fit line for the collectors intersectsthe residual snow ellipse if extended in that direction). Theresidual snow is isotopically heavier than the mean ground-water, but as it makes up only a small portion of the bulkmeltwater contribution, its heavy isotope values are moder-ated by the lighter fraction that makes up the majority of melt.Although one could hypothesize that the snow is progres-sively enriched because of the snowmelt being isotopicallydepleted relative to the remaining solid snow (in which casethe time-integrated snowmelt would be equal in compositionto the original snowfall), we present results in section 3showing that bulk snowmelt in field conditions is isotopicallyheavier than the fresh snow from which it is derived.[28] Although no data were gathered that would allow

direct comparison of collector data to actual infiltration, theisotope signature from the extended funnel collector islikely more representative of infiltration water isotopecomposition than a fresh snow sample because the snowin the extended collector is subject to many of the snowmetamorphism effects that impact snow on the ground.Given that the sample from the extended funnel collectoris the best available representation of the bulk meltwaterinput, it was used as the snowmelt end-member in a mixingcalculation to determine the proportion of snowmelt andrain in the groundwater. If the isotope composition (x) of amixture is known, and the mixture was formed from twoend-members of known isotope composition (y for the firstend-member and z for the second), then the proportion ofeach end-member in the mixture (a for the first end-memberand b for the second) can be calculated (assuming y 6¼ z) bysolving the series of equations: x = ay + bz; a + b = 1. Usingthe d18O value of the meltwater from the extended funnelcollector (�13.5%) as the snowmelt end-member, and thevalue taken from the centroid of the long-term rain ellipse

Table 3. Values Calculated for Percentage Recharge Due to

Snowmelt at the Study Sites and Groundwater Data Used for the

Calculationsa

SiteGroundwaterd18O, %

Percent GW2002–2003

Percent GW2003–2004

Percent GWLong-Term Precip%

Chiricahua �9.29 - - 60 30Los Alamos �11.33 54 48 - 41Magdalena �10.59 51 69 - 49Steward �8.88 - 43 - 25

aIn conjunction with stable isotope data for precipitation provided inTable 2, with the exception of the Chiricahua site, which used long-termmonitoring means of �10.59% for snowmelt and �7.35% for rain.Groundwater d18O values are means calculated from data provided byEarman [2004], Earman et al. [2003], Blake et al. [1995], Gross andWilcox [1983], Cunningham et al. [1998], and Wright [2001]. ‘‘Precip%’’values (calculated from data available from the Western Regional ClimateCenter (http://wrcc.dri.edu)) are the proportion of snow in average annualprecipitation, assuming that snow-water equivalence is 10%.

Figure 5. Plots showing the isotope composition of samples from the three collectors at the Los AlamosHigh site and the global meteoric water line. Also shown in both figures are the ranges of values for summerand winter precipitation based on long-term monitoring [from Adams et al., 1995] and a mean value forgroundwater in the study area (calculated from data given by Blake et al. [1995]). PVC, PVC collector; reg.funnel, regular funnel collector; ext. funnel, extended funnel collector; mean GW, mean value forgroundwater. (a) Data for winter 2002–2003. Also shown is a range of values for ‘‘residual snow’’ (the lastsmall patches of snow remaining near the conclusion of melt). (b) Data for winter 2003–2004.


7 of 18

D09302

(�8.9%) as the rain end-member, the model predicts thatsnowmelt is responsible for 53% of the groundwater re-charge at the Los Alamos study area. In contrast, if a freshsnow value were used (as represented by the PVC collectorsample’s d18O value of �16.6%), the predicted proportionof snowmelt in groundwater recharge would be only 32%.[29] It is important to note that all the recharge estimates

discussed here have some level of uncertainty. In all cases,we are comparing bulk melt isotope values from a givenyear (these yearly values are likely to be relatively variablebecause of variations in fresh precipitation composition andthe amount of postfall isotope alteration of snow) withgroundwater isotope values which, because of diffusiveand dispersive properties, are likely to represent a weightedmean for many years of recharge (and thus be relatively stabletemporally). As mentioned previously, we believe the isotopevalues from the extended funnel collector samples to be abetter representation of bulk meltwater input to groundwaterthan the isotope values of fresh snow, but these collectors donot allow heat input from the ground or mass transfer betweenthe snowpack and soil gas, so they are not a completelyaccurate analog. The groundwater isotope value used for eachsite is the mean of several values with an associated range anduncertainty. Finally, all values suffer from analytical uncer-tainty. For some of these factors (e.g., temporal variability ofisotope alteration of snow), it is not possible to develop ameaningful level of uncertainty to associate with the rechargeestimates, given the available data. Although uncertaintyexists, these estimates are still valuable (1) as a generalquantification of the importance of snowmelt to groundwaterrecharge, and comparison of the relative importance at varioussites, (2) for comparing recharge estimates for a given sitebased on fresh snow and altered snow input signatures, and(3) for comparing year-to-year variations in snowmelt con-tributions to recharge at a given site.[30] The fresh snow (PVC collector) from the winter of

2003–2004 is isotopically heavier than the fresh snow fromthe previous winter. However, the magnitude of isotopicalteration observed for snow in the extended funnel collec-tor in the winter of 2003–2004 is smaller than that whichwas observed during the previous winter, meaning that theisotopic signature of the 2003–2004 bulk snowmelt used inthe end-member mixing analysis is lighter than that from theprevious winter. As a result, data from this season suggest aslightly lower proportion (48%) of the recharge is derivedfrom snowmelt (using the d18O of �14.1% from theextended funnel collector in the mixing model describedabove), and a lower difference between the proportionscalculated using the enriched and fresh snow samples(11% rather than 21% for the previous winter).2.4.3. All Sites[31] The calculated contribution of snowmelt to ground-

water recharge at the four study sites ranges from 43% to69%, depending on the site and the year of data used in thecalculations (Table 3). In all cases, the proportion of snow-melt in recharge is higher than the proportion of snow inaverage annual precipitation. The two yearly estimates of theestimated proportion of groundwater recharge derived fromsnow at the Magdalena site were 2% and 20% higher than theproportion of snow in average annual precipitation. Forthe Los Alamos site, the two yearly estimates showed thatthe proportion of recharge derived from snowmelt were 7%

and 13% higher than the proportion of snow in averageannual precipitation. For the Steward and Chiricahua sites,the calculated proportions of snowmelt in recharge are 22%and 30% higher (respectively) than the proportion of snow inaverage annual precipitation. The conditions during bothwinters of the study period were warmer and dryer than thehistorical norm. As a result, the amount of snow and numberof snowfall events were lower than typical. The low amountsof snow observed during the study period most likely causedthe proportions estimated here for percentage of rechargederived from snowmelt to be minima (i.e., lower than wouldhave resulted if observations had taken place during a yearcloser to the historical norm, with cooler, wetter conditions).[32] Although the estimates of snowmelt-derived re-

charge from the two years in which data were gatheredare comparable for the Los Alamos site, the Magdalena siteshowed a distinct difference. Unfortunately, there are nolonger-term data for stable isotopes in precipitation at theMagdalena site as there are for the other sites, so it is notknown how similar the PVC collector values observed inthis study are to average precipitation at the site. In all cases,long-term monitoring with paired collectors would likelyprovide a more accurate range of values for rechargederived from snowmelt, and also a mean value based on amore reasonable number of observations.[33] At the Magdalena site, increased temperatures

caused the fresh snow isotope signature to be heavier inthe winter of 2002–2003 than winter 2003–2004. Theisotopically heavy fresh snow produced by the warmtemperatures was enough to outweigh the decreased enrich-ment resulting from decreased exposure time, causing thebulk melt signature from 2002 to 2003 to be isotopicallyheavier than that from 2003 to 2004. Because of the climateregime in effect at the time of the study, no cool years wereobserved, so it is not known to what degree the isotopicallylight fresh snow falling in a cool year would be offset by theincreased exposure time resulting from greater amounts ofsnow and cooler temperatures.

3. Investigation of Processes Contributing toIsotope Alteration in Snow

3.1. Methods

[34] To investigate the processes responsible for thealteration of the stable isotope composition of snow, a studyarea was established in the Magdalena Mountains, near themidelevation Magdalena site (Figure 2). The study areaconsisted of two adjacent sites. At each site, a collection panwas placed on the ground. These pans were composed ofwhite plastic in order to mimic the albedo of snow, andprevent the large energy input that would be associated withdark and/or metal pans. The weight of the pans could bemeasured on site, and they were designed so that meltwaterproduced from the snow would enter a collection bottlecontaining a layer of mineral oil. One of the pans could becovered by a shade to block direct energy input from the sun(free circulation of air over the snow was allowed); the otherpan was left exposed to the natural cycle of sunlight anddarkness. The shade was emplaced over the first collectionpan as soon as feasible after the conclusion of a snow event.Therefore, although snow in the shaded pan may havereceived a short period of direct solar energy input, it was


8 of 18

D09302

kept to a minimum. The sample collection and analysiswere conducted in the same manner as those described insection 2.2.[35] The comparison of results from adjacent snows

experiencing different energy inputs was considered impor-tant because freeze-thaw cycling is recognized as being asignificant control on isotopic metamorphism of snow andice [Jouzel and Souchez, 1982; Cooper, 1998], and thevariation in energy input could also affect the amount ofevaporation/sublimation taking place.

3.2. Results

[36] Data from a snow event monitored for this study aregiven in Table 4, and shown in Figures 6, 7, and 8. Both theshaded and unshaded snow lasted from 20 January until31 January. In general, progressive isotopic enrichment ofboth the remaining snow and the meltwater took place in bothpans; the unshaded pan exhibited greater degrees of isotopicenrichment for both the melt and the remaining snow.[37] Shading affected the mode of mass loss in the snow

(Figure 6). The shaded snow lost 35% of its mass viaevaporation/sublimation (calculated by mass balance clo-sure), while the unshaded pan lost only 13% of its mass inthis manner. The unshaded snow yielded its first melt a dayprior to the shaded pan, and always had higher mass loss dueto melt on any given day. Another effect of the shading isevident in the period from 25 January to 27 January, whenextremely cold temperatures prevailed. The shaded pan lostvirtually no mass because of melt in this period, but theunshaded pan experienced melt loss because of the extraenergy derived from solar radiation. The unshaded pan

received small inputs of fresh snow during this period, butthe shaded pan did not. The inputs of snow were minorenough that the trend of isotope evolution was not altered inthe unshaded pan (see Figure 8), but this input preventedmassbalance closure, and thus prevented calculation of the mass ofsnow lost because of evaporation/sublimation. For both theshaded and the unshaded snow, the two days with the highesttemperatures (23 and 30 January) correlate with the highestslopes on the melt loss curves; the shift in slope on the totalmass loss curve on these days is more distinct for the shadedpan than the unshaded pan. It is interesting to note that therates of total mass loss for the shaded and unshaded pans aresimilar, even though the unshaded pan’s direct exposure tosunlight should have resulted in the unshaded snow receivingmore energy than the unshaded pan. The increased rate ofsublimation of the snow in the shaded pan (compared to thesnow in the unshaded pan) accounts for the similarity in totalmass loss rates in spite of the difference in melt loss in the twopans. The rate of sublimation is controlled by two factors: thedifference in vapor pressure between the atmosphere and thesurface of interest, and turbulent diffusion in the air (whichscales with wind speed) [MacClune et al., 2003]. It is possiblethat the shading apparatus (a sheet of plywood approximately1.25 cm thick placed�30 cm above the top of the snowpack,supported by four steel rods approximately 1 cm in diameter)increased the wind speed and/or air turbulence over the snowin the shaded pan, thus driving some increase in the sublima-tion rate. Another possibility is that the shaded snow receivedless solar energy input than the unshaded snow; because ofthis difference, the shaded snow surface should have had alower vapor pressure than the unshaded snow surface. As a

Table 4. Data for the Snow Pan Experiment at the Magdalena Study Sitea

DateTotalLoss, g Melt, g

SampleMass, g

NetLoss, g

E/SLoss, g

SnowdD, %

MeltdD, %

Snowd18O, %

Meltd18O, %

MeanTemp., �C

Min.Temp., �C

Max.Temp., �C

Unshaded Pan21 Jan. 2004 - - - - - �109.7 - �16.8 - �2.9 �4.6 �0.722 Jan. 2004 31 2.2 28.8 2.2 0.0 �106.5 - �16.5 - �1.1 �5.1 3.123 Jan. 2004 450 372.3 10.9 439.1 66.8 �101.8 �112.9 �16.7 �17.7 0.8 �3.3 4.324 Jan. 2004 785 719.0 10.3 774.75 55.8 �95.0 �103.6 �15.0 �16.2 �1.2 �2.8 0.825 Jan. 2004 65 157.4 7.6 - - �92.1 �103.6 �14.1 �16.1 �4.7 �12.1 �2.726 Jan. 2004 - - - - - - - - - �11.6 �15.8 �7.627 Jan. 2004 124 123.8 11.5 - - �86.3 �78.8 �13.1 �12.6 �5.2 �9.6 �1.128 Jan. 2004 - - - - - - - - - �1.7 �4.4 1.429 Jan. 2004 780 573.6 9.0 771 197.4 �82.5 �79.5 �12.3 - �1.5 �3.3 1.930 Jan. 2004 - - - - - - - - - 1.9 �2.3 6.931 Jan. 2004 850 775.2 0.0 850 74.8 - �78.6 - �11.8 �2.3 �5.0 0.3

Shaded Pan21 Jan. 2004 - - - - - �110.8 - �17.4 - �2.9 �4.6 �0.722 Jan. 2004 - 2.4 28.4 0.0 0.0 �108.4 - �17.2 - �1.1 �5.1 3.123 Jan. 2004 100 0.2 9.0 91.0 90.8 �103.2 - �15.5 - 0.8 �3.3 4.324 Jan. 2004 665 467.8 10.3 654.7 186.9 �97.0 �109.0 �14.3 �17.1 �1.2 �2.8 0.825 Jan. 2004 225 132.2 7.2 217.8 85.6 �96.5 �107.6 �14.2 �16.0 �4.7 �12.1 �2.726 Jan. 2004 - - - - - - - - - �11.6 �15.8 �7.627 Jan. 2004 170 0.9 10.5 159.5 158.6 �95.3 - �13.6 - �5.2 �9.6 �1.128 Jan. 2004 - - - - - - - - - �1.7 �4.4 1.429 Jan. 2004 460 252.8 8.7 451.3 198.5 �95.5 �90.8 �13.5 �13.1 �1.5 �3.3 1.930 Jan. 2004 - - - - - - - - - 1.9 �2.3 6.931 Jan. 2004 680 602.4 0.0 680.0 77.6 - �93.8 - �13.0 �2.3 �5.0 0.3

aTotal loss represents the mass of snow ‘‘missing’’ from the pan, relative to the previous measurement. This includes melt loss, the snow sample extractedfor stable isotope analysis, and evaporation/sublimation (E/S) loss. Because the snow sampling should not have impacted the isotope composition of thesnow, it is not included in the net mass loss, which was used to calculate the percentage mass loss values in Figures 6 and 7. The temperature data shown inthis table have been adjusted to account for the fact that the data were collected at a nearby site approximately 700 m higher in elevation. All values in thistable were raised by 4.5�C compared to the measured values (based on a single set of contemporaneous measurements from the two sites). Note that thevalues shown in Figures 6 and 7 show the uncorrected data, not the adjusted values. Additions of snow to the unshaded pan prevented mass balance closureon 25 January and 27 January, so values are not given for net loss or E/S loss. Min. temp., minimum temperature; max. temp., maximum temperature.


9 of 18

D09302

result, the snow/atmospheric vapor pressure gradient for theshaded snow should have been lower than that for theunshaded snow, resulting in a lower sublimation flux.

3.3. Discussion

[38] Sublimation of snow, being a kinetically influencedprocess, tends to drive isotope evolution on a dD-d18O slopesimilar to that for evaporation [Stichler et al., 2001]. During

periods of melt, it is not possible to distinguish the evap-oration of meltwater in the snow matrix from sublimation ofsnow, but the impact on the isotope evolution will be thesame. On the basis of the best fit regression lines for thedata, the unshaded snow and its meltwater evolve along dD-d18O slopes of 5.4 and 6.2 (respectively), while the shadedsnow and its meltwater evolve along dD-d18O slopes of 3.9and 4.4 (respectively) (Figure 8).

Figure 6. Data for the shaded snow pan from the Magdalena study site. (a) Temperature data and massloss as a function of time. For temperature, the box with a horizontal line through its center representsmean daily temperature, and the high and low points of the associated vertical bars represent maximumand minimum temperatures (respectively). Temperature data are from the nearest meteorological station,located in the same mountain range, but approximately 700 m higher in elevation. Although the absolutevalue of the temperature at the study site was almost certainly slightly warmer than the temperature at themeteorological station, the data should be representative of temperature trends at the study site. (b) ThedD of snow and melt as a function of time.

Figure 7. Data for the unshaded snow pan from the Magdalena study site. (a) Temperature data andmass loss as a function of time. For temperature, the box with a horizontal line through its centerrepresents mean daily temperature, and the high and low points of the associated vertical bars representmaximum and minimum temperatures (respectively). Temperature data are from the nearestmeteorological station, located in the same mountain range, but approximately 700 m higher inelevation. Although the absolute value of the temperature at the study site was almost certainly slightlywarmer than the temperature at the meteorological station, the data should be representative oftemperature trends at the study site. (b) The dD of snow and melt as a function of time.


10 of 18

D09302

[39] From 25 January to 27 January, the shaded snow lost160 g of mass, but this was almost exclusively due toevaporation/sublimation (less than 0.6% of the mass lost inthis period was due to melt). Because the loss due to meltwas negligible, processes such as fractionation due to partialmelting, melt/snow interaction, and meltwater/atmosphericwater vapor exchange can be ruled out as affecting theisotope composition of the snow. The mechanisms with thepotential to drive isotope evolution during this period aresublimation, snow/atmospheric water vapor exchange, andcondensation of atmospheric water vapor on the snowsurface. As we did not measure the amount of condensation(if any) on the snow surface we will assume it had anegligible effect on the isotope composition of the bulksnow.[40] The slope of the isotope evolution of the shaded

snow (on a dD-d18O plot) from 25 January to 27 January is5.3. To verify sublimation as the driving force of the isotopeevolution over this period, the observed data from25 January to 27 January were compared to values calcu-lated assuming pure sublimation. The isotopic effect ofsublimation can be calculated from the appropriate fraction-ation factors. The ice/vapor isotopic enrichment factors(calculated from equilibrium fractionation factors describedby Majoube [1971a] and Merlivat and Nief [1967]) wereadjusted for kinetic effects using the method of Gonfiantini[1986] (meteorological data from the crest of the range weremodified to account for the lower elevation of the study site,yielding a mean temperature of �7.2�C at the study sitefrom 25 January to 27 January (range from �15.8� to�1.1�C), and a mean relative humidity of 0.77 (range from0.41 to 0.96)). For the mean temperature during the studyperiod, the equilibrium ice-vapor isotopic enrichment fac-tors were �16.3 for d18O and �94.2 for dD. On the basis ofGonfiantini’s [1986] equations for modification of equilib-rium enrichment factors to account for kinetic effects (De18O (%) = 14.2 (1 � h); De 2H (%) = 12.5(1 � h); where Deis the ‘‘kinetic enrichment factor’’ (the amount of additionalenrichment due to fractionation above that which would

occur in a pure equilibrium process) and h is the relativehumidity), the adjusted isotopic enrichment factors are�97.05 for dD and �19.54 for d18O. The adjusted isotopicenrichment factors suggest that pure sublimation wouldcause the resultant isotopic enrichment to plot on a slopeof 5.0 on a dD-d18O diagram (based on the ratio of theisotopic enrichment factors), quite close to the observedslope of 5.3. The difference between the measured andpredicted slopes could be explained by two factors. First,the extrapolation of meteorological conditions at the sitemight have been slightly in error. Second, the estimatedfractionation factors could be correct, but an equilibriumprocess (such as exchange with atmospheric water vapor)could have been active in addition to sublimation, pullingthe slope higher. An experiment discussed in section 4 ofthis paper shows that the isotope composition of snow keptbelow freezing can be affected by exchange with atmo-spheric water vapor. Although the observed and predictedslopes are similar, the magnitude of observed isotopicenrichment is much smaller than the predicted enrichment(Figure 9). The predicted enrichment was calculated using aRayleigh distillation equation with the adjusted enrichmentfactors. The initial point on the ‘‘predicted’’ curve representsthe observed values for dD and d18O on 25 January; the finalpoint represents the predicted values for 27 January (as-suming sublimation was the only cause of mass loss) basedon the observed mass loss (i.e., the mass loss from25 January to 27 January was used to calculate the fractionof snow remaining (f) on 27 January). The most likelyexplanation is that both Majoube [1971a] and Merlivat andNief [1967] determined vapor/ice equilibrium fractionationfactors using experimental systems relying on the instanta-neous freezing of water vapor. As described by Moser andStichler [1980], the vapor/ice transition can be expected toyield ‘‘isotope effects close to those demanded by equilib-rium conditions,’’ but the ice/vapor transition may not,depending on the configuration of the ice crystals and theirsurface area/volume ratio.

Figure 8. The dD versus d18O for the snow and melt from adjacent unshaded (US) and shaded (S) snowpatches at the Magdalena study site. The triangles represent the weighted mean of all melt samples. Notethat the mean melt is enriched compared to the fresh snow.


11 of 18

D09302

[41] Another phenomenon of note is that for both snowpans, the melt derived from the snow is isotopically lighterthan the remaining snow during the initial period of melt, butbecomes isotopically heavier than the remaining snow duringthe latter portion. For both the shaded and unshaded snow,this transition occurred when about 70% of the snow’s initialmass had been lost. On the basis of laboratory experimentswith constant input of thermal energy,Herrmann et al. [1981]found that for constantly melting snow in the laboratory,initial melt is isotopically lighter than snow because of‘‘fractionation effects’’ (they describe their experimentalresults as ‘‘compatible with the experiments carried outby’’ Arnason [1969a, 1969b], ‘‘and with the theoreticalconsideration by’’ Buason [1972]). Herrmann et al.’s[1981] results show that the fractionation between the snowand meltwater decreases as the experiment progresses; inother words, as the fraction of snowmelted moves from 0 to 1,the dD of the meltwater becomes progressively more similarto that of the snow; they do not provide an explanation for this(they describe the temperature and humidity as being heldconstant during the experiments). Although the meltwater inthe experiment of Herrmann et al. [1981] did approach theisotopic composition of the snow as the fraction meltedneared unity, no melt isotopically heavier than the remainingsnow was observed. This suggests that this laboratory studywith constant melt was not a good approximation of fieldconditions in which melt occurs only for part of each day. Atleast two field studies have observed snowmelt isotopevalues heavier than those of the snow from which it isderived, but no interpretation of the driving force was made[Martinec et al., 1977; Hooper and Shoemaker, 1986];possible causes include evaporation of meltwater in thesnowpack, isotope exchange between meltwater and atmo-spheric water vapor, and interaction between meltwater and

isotopically enriched snow at the base of the snowpack.Herrmann et al. [1981] conducted a laboratory experimentwith alternating freeze-thaw cycles, but show data for themeltwater only, so the snow/meltwater relationship is notknown.[42] The isotope evolution of the snow is dependent upon

the interplay of a number of processes. The importance ofthe evaporation/sublimation process is dependant on theincoming solar energy. If the solar energy input is relativelyhigh, melt processes tend to drive most of the isotopeevolution, with evaporation/sublimation processes account-ing for a relatively small part of the observed isotopicenrichment. In cases of relatively low solar energy input,evaporation/sublimation processes are more important indriving the isotope evolution, but are balanced by equilib-rium processes, including isotope exchange between snowand meltwater, snow and water vapor, and meltwater andwater vapor.

4. Investigation of the Potential for IsotopeExchange Between Snow and Atmospheric WaterVapor to Affect the Isotope Evolution of Snow

4.1. Background

[43] As discussed previously, there are many mechanismsthat have the potential to alter the isotope content of snow,including melt, partial melt and refreezing, mass transportdue to temperature gradients (either individually or inconcert with other mechanisms), rain-on-snow events, iso-tope exchange between meltwater and atmospheric vapor(or soil vapor), and evaporation/sublimation [Cooper,1998]. Another possible mechanism for alteration is isotopeexchange between snow and atmospheric water vapor, butthe viability and potential influence of this process on theisotope composition of snow has never been demonstratedby experiment, and its practical significance is not univer-sally recognized or accepted.[44] In the past, it was commonly believed that isotope

exchange between snow and water vapor could not exert asignificant influence on the isotope composition of a snow-pack because of the slow rate of molecular diffusion insolids such as snow [e.g., Buason, 1972; Hage et al., 1975].The more prevalent view now is that snow/vapor exchangecan be important. Cooper [1998] states that exchange withatmospheric water vapor can have a significant impact onthe isotope content of snow, especially the upper layers ofthe snowpack, and a number of authors, including Stichler[1987], Friedman et al. [1991], Sommerfeld et al. [1991],and Rose et al. [1999], suggest that exchange with atmo-spheric water vapor could help explain alteration observedduring field studies of snowpacks. However, studies thatsuggest snow-vapor exchange is an important processappear to do so on the basis of ‘‘Occam’s razor’’-typereasoning: observed shifts in snow isotope compositionsare most easily explained if snow/vapor exchange is in-voked. To our knowledge, no experiment has given directevidence of snow/vapor exchange. Perhaps as a result ofthis lack of direct evidence, many studies appear to discountthe process. For instance, Feng et al. [2002] and Taylor etal. [2002] state that the two factors controlling isotopicevolution of snow are the rate of meltwater-ice isotopicexchange and the ratio of ice to liquid in the system, and

Figure 9. Observed isotopic enrichment of shaded snow atthe Magdalena study site due to sublimation (with possibleinfluence from exchange with atmospheric water vapor)from 25 to 27 January shown with calculated values basedon adjusted equilibrium fractionation factors.


12 of 18

D09302

Unnikrishna et al. [2002] describe a snowpack surface andits isotope composition as being affected by ‘‘precipitation,evaporation and condensation, and melt out’’ but do notmention isotope exchange.[45] If isotope exchange does occur between solid-phase

snow and atmospheric water vapor, the impact could besignificant. In many systems, isotope exchange is the mainfactor that determines the distribution of stable isotopes[Ingraham and Criss, 1993]. The most commonly knownexample is Craig et al.’s [1963] demonstration that ex-change with atmospheric water vapor dominates the isoto-pic evolution of liquid water after an initial period ofevaporation influence. Ingraham and Criss [1993] showthat two waters in a sealed container with different isotopesignatures will undergo isotope exchange with each othervia the atmospheric water vapor, eventually reaching acommon isotope composition.[46] A set of experiments was performed to test the

hypothesis that isotope exchange between solid-phase snowand atmospheric water vapor can significantly alter theisotope composition of the snow.

4.2. Methods

[47] Snow was collected from the Magdalena Mountains(Figure 2), and sealed in airtight containers, which wereplaced in an insulated cooler with ice packs for transport tothe laboratory. The snow was sampled by using the airtightcontainers as scoops, with sampling targeted to the upperportion of the snow column to minimize isotopic heteroge-neity and avoid incorporation of soil, plant matter, and otherforeign materials. The snow was collected within 24 hoursof fall, but because of experimental design issues, the snowused in the experiments detailed here had been stored in afreezer for several weeks and had undergone physicalmetamorphism (based on a change in crystal size), likelybecause of temperature fluctuations in the freezer. In thelaboratory, the airtight container was removed from thefreezer, and the snow was stirred to promote homogeniza-tion (we did not perform sampling specifically designed totest isotopic homogenization). An aliquot of the snow(approximately 60 g) was placed on a screen underlain bya pan to collect any melt that might be generated from thesnow. The snow and melt pan were kept in a sealed,temperature-controlled chamber for 10 days. Samples ofsnow were collected at the beginning and end of eachexperimental run. To determine if exchange occurs betweensnow and atmospheric water vapor, two runs were con-ducted in which the isotope composition of the atmosphericwater vapor in the experimental chamber was maintained atdifferent values.[48] One run was designed as a control. In this case, the

temperature of the experimental chamber was kept below0�C, and the initial water vapor in the chamber air wassimply that which was sealed in as part of the ambientatmosphere in the laboratory (see Figure 10).[49] For the second run, the temperature of the experi-

mental chamber was kept below 0�C, and, in addition to thesnow, a pan of water (approximately 200 g) having adistinctive isotopic composition was placed inside thechamber. The water was maintained in the liquid state bytwo means: addition of NaCl to suppress the freezing point,and bubbling air through the water with a battery-operated

pump. The water was precooled in the chamber prior to theintroduction of the snow in order to eliminate the possibilityof thermal pumping driving mass movement from the waterto the snow. The water was intended to provide a largereservoir compared to the vapor present in the chamber air,and force the isotopic composition of the water vapor to avalue significantly different than that in the control exper-

Figure 10. Schematic representation of experimental de-sign. (a and b) Cross-sectional views of the design; notechamber-in-chamber design, with inner chamber isolatedfrom the floor of the outer chamber. Figure 10a shows thedesign of the run for snow without CSBD in the innerchamber, and Figure 10b shows the design of the run for snowwith CSBD. (c) Plan view of the design of the run for snowwith CSBD; only the inner chamber, snow, and water areshown.


13 of 18

D09302

iment. The surface areas of the water (154 cm2) and thesnow (182 cm2) were similar, and the containers were inclose proximity to each other (both containers were approx-imately 14 cm in diameter/width, and the chamber was40 cm in diameter). If snow/vapor exchange is a significantprocess, the difference in chamber water vapor isotopecontent during these two runs should cause different isotopeevolution of the snowpack.[50] At the initiation and conclusion of both experimental

runs, a sample was collected that integrated the entirethickness of the snow (we did not investigate the redistri-bution of isotopes in the snow during the course of theexperiment). For the second run, aliquots of the water werealso collected at the start and end of the experiment.[51] In both cases, the experimental chamber was an

airtight plastic container that was housed inside a larger,temperature-controlled chamber. The inner chamber was notin direct contact with the outer chamber floor or walls. Thischamber-in-chamber design was intended to eliminate the‘‘plating-out’’ effect observed by Sommerfeld et al. [1991],in which the freezer walls are a sink for water vapor in thechamber because they are the coldest part of the systemduring the initial portions of the cooling cycles of thefreezer. Although not completely successful in eliminatingplating out, the chamber-in-chamber design reduced it toabout 10 to 20% of that observed in runs conducted withoutthe inner plastic chamber. We did not conduct any measure-ments of the relative humidity in the inner chamber or outerchambers during the experiments. In both runs, a loggingthermometer inside the inner chamber showed the temper-ature was held at a mean value of approximately �3�C, witha range in temperature from �4.7� to �1.5�C.[52] In order to assure that any exchange that took place

could be distinguished from other processes (e.g., evapora-tion), the water used in the second experiment was distillateproduced from Canadian Shield brine (referred to here as‘‘CSBD’’). This brine is isotopically unusual because,compared to a water that plots on the global meteoric waterline (GMWL), it is significantly enriched in dD, but not ind18O; in other words, it plots ‘‘above’’ or ‘‘to the left’’ of theGMWL. Isotope exchange with this water should thus causea shift in isotope composition of the snow that could not beaccounted for via evaporation/sublimation, because theslope of evolution on a dD-d18O plot would be too high.

4.3. Results

[53] The results of the experiments are shown in Table 5and Figure 11. Because of a delay between the twoexperimental runs, isotope metamorphism of the snowoccurred during storage, causing the initial d values of snowsamples used in the two runs to differ, but the relativechange from the initial d values can be used to compareresults from the two runs. Snow that was alone in the

chamber is referred to as ‘‘snow without CSBD’’; the snowaccompanied in the chamber by the CSBD is referred to as‘‘snow with CSBD.’’[54] In addition to the temperature data from the logging

thermometer, visual observation confirmed that no liquidwater was present in the catch chamber beneath the snow atthe conclusion of the experiments, and that none appearedto have been produced and subsequently frozen.[55] Compared to the snow with CSBD, the snow without

CSBD experienced greater isotopic enrichment (Figure 11),with the shift in isotope composition characterized by amuch lower slope (3.7 versus 11.9). The low slope of thesnow without CSBD is characteristic of a kinetically influ-enced process such as evaporation or sublimation [Stichleret al., 2001]. Indeed, this snow lost about 6.3 g to subli-mation (�10% of its initial mass). The snow with CSBDlost about 3.8 g (�7% of its initial mass) because ofsublimation. The lower mass loss in this instance is due tothe presence of the CSBD, which is an additional reservoirthat can provide water to the atmosphere of the system. TheCSBD lost 3.3 g during the run. The inner chamber wallsaccumulated 6.9 g of frost during the experiment, yielding amass balance error of 0.2 g (although no effort was made to

Table 5. Values for Isotope Composition and Mass of the CSBD, Snow With CSBD, and Snow Without CSBD

SampleInitialdD, %

FinaldD, %

Initiald18O, %

Finald18O, %

InitialMass, g

FinalMass, g

Snow with CSBD �104.3 �99.5 �15.34 �14.95 55.7 51.9CSBD �74.6 �71.5 �15.92 �15.26 193.8 190.5Snow without CSBD �104.7 �97.7 �16.78 �14.86 56.0 49.7

Figure 11. Plot showing the values for the CSBD, snowwith CSBD, and snow without CSBD during the experi-mental runs, along with the global meteoric water line(GMWL). On the scale of this diagram, the error barsrepresenting the uncertainty are the same size as thesymbols and are thus not shown.


14 of 18

D09302

account for the mass of water vapor in the chamber that hadbeen derived from the CSBD and snow with CSBD).

4.4. Discussion

[56] A Rayleigh distillation model was used to predict theshift in isotope composition of the CSBD and snow withCSBD that would have resulted had they lost mass solelybecause of evaporation and sublimation (respectively). Themass fractions remaining at the end of the experiment wereused in conjunction with the fractionation factors for theliquid-vapor [Majoube, 1971b] and ice-vapor systems[Merlivat and Nief, 1967; Majoube, 1971a] and the environ-mental conditions of the system. The results of the calcula-tions (Figure 12) show that the CSBD became less enriched indD than predicted, but more enriched in d18O. For the snowwith CSBD, the opposite occurred: it became less enriched ind18O than predicted, but more enriched in dD. This isconsistent with isotope exchange between the two reservoirsvia the atmosphere, as observed by Ingraham and Criss[1993] for a liquid-vapor-liquid system. Although the snowwith CSBD was enriched in d18O because of sublimation,concurrent exchange with the CSBD would render the snowwith CSBD’s d18O value lighter. Simple sublimation of thesnowwith CSBDwould result in a low slope of evolution, butconcurrent exchange with the CSBDwould cause the slope toincrease by greatly increasing the deuterium content of thesnow with CSBD relative to the increase in 18O content.[57] If the isotope evolution of snow below 0�C is not

influenced by exchange with atmospheric water vapor, then

both the snow with CSBD and the snow without CSBDshould be evolve along similar (if not identical withinanalytical error) low-slope paths on a dD-d18O plot, assublimation should dominate in both cases, and this shouldcause low-slope evolution, as observed by Stichler et al.[2001]. At the experimental temperature, the maximumslope that should result from sublimation of snow is 6.0(using the vapor/ice equilibrium isotopic enrichment factorsgiven by Majoube [1971a] and Merlivat and Nief [1967],and assuming no kinetic effects). If isotope exchange withatmospheric water vapor can influence the evolution ofsnow, then snow with CSBD and snow without CSBDshould evolve along different paths on a dD-d18O plot. Inthis case, the snow without CSBD would still evolve alongits sublimation-caused low-slope path, but the snow withCSBD would along a much higher slope, as pure exchangewith the CSBD would drive nearly vertical evolution on adD-d18O plot. The equilibration of temperatures prior to thestart of each run and the fact that the snow with CSBD lostmass during the experiment suggest that condensation ontothe snow with CSBD was not a significant process.[58] The CSBD became enriched because of evaporation

during the course of the experimental run. As both its initialand final isotope compositions are heavier in dD and lighterin d18O than those of the snow with CSBD, it is reasonableto infer that these relationships held throughout the run.Given the isotope compositions of the CSBD and snow withCSBD, the observed evolution of the snow with CSBDappears to have resulted from a combination of sublimationand exchange with atmospheric water vapor.[59] Experimental investigation of the evolution of snow

under controlled conditions in the laboratory shows that snowsamples in a sealed chamber exposed to distinctive atmo-spheric water vapor compositions manifested distinct changesin their isotope compositions. As the temperature of thesystem was maintained below 0�C, no evidence of melt wasobserved, and the snow and water temperatures were equal-ized prior to the start of the experiment, isotope exchangebetween the snow and the water vapor is the only factor thatcan reasonably explain the distinct shift in isotope signatures.Although the evolution of the snow with CSBD on a dD-d18Oplot is not along the near-vertical slope that would haveresulted from pure isotope exchange, the high slope of 11.9it is consistent with a combination of exchange with theCSBD via the atmosphere and sublimation. While theseresults are of interest, we note that they are based on a singleset of experimental runs. Further investigations of snow/vaporexchange should be conducted to confirm our observations.

5. Conclusions

[60] The isotope composition of snow changes betweensnowfall and the conclusion of melt. Alteration duringexposure causes the snow isotope composition to becomeheavier, and thus more similar to that of rain from the samearea. The shift in isotope composition is due to a combina-tion of factors, including evaporation/sublimation, partialmelt and refreezing, and exchange with atmospheric watervapor. The relative importance of the various processesdepends on ambient conditions such as temperature and theamount of incoming solar radiation. Accounting for this shiftcan significantly increase estimates of the proportion of

Figure 12. Plot of the observed isotope evolution of thesnow with CSBD and of CSBD (points connected by solidlines) and the evolution predicted to occur if onlyevaporation had affected the CSBD and only sublimationhad affected the snow with CSBD (points connected bydashed lines), along with the global meteoric water line(GMWL). On the scale of this diagram, the error barsrepresenting the uncertainty are the same size as thesymbols and are thus not shown.


15 of 18

D09302

snowmelt in groundwater recharge. In addition, differenttypes of precipitation collectors can yield significantlydifferent values if snow makes up a significant portion ofprecipitation. Collector design should be carefully consid-ered when commencing a study to ensure that the results areapplicable to the matter of interest (e.g., a study of isotopecomposition of meteoric water would probably call for adifferent collector design than a study to determine theisotope composition of water contributing to groundwaterrecharge). It is not uncommon for the stable isotope compo-sition of water from high-elevation springs in mountainranges to be used as a proxy for the isotope composition ofprecipitation from the area, because of the short residencetime of these waters. However, if snow is a significantcontributor to groundwater recharge, springwater isotopecomposition may differ significantly from that of localprecipitation.[61] Observations at four sites in the Southwest show that

snowmelt accounts for at least 40 to 70% of groundwaterrecharge; at these sites snow makes up 25 to 50% of theaverage annual precipitation. These high proportions ofsnowmelt contribution to groundwater recharge at the sitesmonitored for this study, in conjunction with the work ofSzecsody et al. [1983] and Winograd et al. [1998], suggestthat in the Southwest, snowmelt is often a greater compo-nent of groundwater recharge than rain, and that snowmelt’scontribution to recharge typically exceeds the proportion ofsnow in average annual precipitation.

6. Implications

[62] A number of recent studies show that under generallyaccepted climate change scenarios resulting from increasingCO2 in the atmosphere (these predict temperature increasesof �1.2� to 2�C over the next 50 years), the total amount ofprecipitation in the western United States will remain fairlyconstant, but the proportion of the total falling as snow willdecline significantly; many works also discuss the impactsof these predicted changes on surface water resources [e.g.,Hinzman and Kane, 1992; Miller and Russel, 1992; Gleickand Chalecki, 1999; Hamlet and Lettenmaier, 1999; Leungand Wigmosta, 1999; Ojima et al., 1999;Mote, 2003; Leunget al., 2004]. Despite the large impacts climate change couldhave on groundwater resources, there are relatively fewstudies dealing with this issue [Vaccaro, 1992; Loaiciga etal., 2000; Kirshen, 2002; Younger et al., 2002; Yusoff et al.,2002; Beuhler, 2003; Allen et al., 2004], and nearly all theseworks deal with areas outside the western United States,where the large snow-rain shifts are predicted.[63] If the results from our study sites are applicable to

other areas of the Southwest, the loss of snowpack couldhave a greater impact on groundwater recharge than esti-mates based only on changes in the amount of precipitationwould indicate. Because our results show that snowmeltyields more recharge per unit amount of precipitation thanrain, even if total precipitation remains constant, a shift fromsnow to rain could cause significantly decreased recharge.While the lessened amount of snowfall would be onecontributor to loss of recharge, the changed conditionscould also reduce the recharge efficiency of snow comparedto that observed today. Thinner snowpacks subjected toincreased temperatures would melt more rapidly that at

present, increasing the likelihood of the melt running offrather than infiltrating. Given the increasing burdens onSouthwestern aquifers resulting from rapid populationgrowth, these findings suggest that additional investigationsof the potential impact of climate change on groundwaterresources are needed.

[64] Acknowledgments. Funding for the field portion of this researchwas provided by the New Mexico Water Resources Research Institute andthe SAHRA Science and Technology Center of the National ScienceFoundation (NSF agreement EAR-9876800). We thank Chris Caldwellfor his assistance in the design and construction of field equipment. We aregrateful to the U.S. Forest Service, the Southwestern Research Station ofthe American Museum of Natural History, Langmuir Laboratory forAtmospheric Research (New Mexico Institute of Mining and Technology),and the Steward Observatory (University of Arizona) for their assistance inallowing the establishment of field sites and access to these sites. We alsoappreciate the input/assistance of Chris Eastoe, Steve Hunyady, DanKlingelsmith, and Louie Pope on various matters related to the study.Funding for the laboratory portion of this research was provided by theNew Mexico Water Resources Research Institute. We thank Rob Bowmanfor the use of his laboratory facilities, D. J. Bottomley for providing theCanadian Shield brine, and Karen Jacobs for performing the distillation ofthe brine and some sample analyses. We appreciate John Wilson’s helpfulcomments on an early version of this manuscript; we are also gratefulto the two anonymous reviewers, whose comments led to significantimprovements.

ReferencesAdams, A. I., F. Goff, and D. Counce (1995), Chemical and isotopic varia-tions of precipitation in the Los Alamos region, New Mexico, Los Ala-mos Natl. Lab. Rep. LA-12895-MS.

Allen, D. M., D. C. Mackie, and M. Wei (2004), Groundwater and climatechange: A sensitivity analysis for the Grand Forks aquifer, southernBritish Columbia, Canada, Hydrogeol. J., 12, 270–290.

Arnason, B. (1969a), Equilibrium constant for the fractionation of deuter-ium between ice and water, J. Phys. Chem., 73, 3491–3494.

Arnason, B. (1969b), The exchange of hydrogen isotopes between ice andwater in temperate glaciers, Earth Planet. Sci. Lett., 6, 423–430.

Beuhler, M. (2003), Potential impacts of global warming on water resourcesin southern California, Water Sci. Technol., 47, 165–168.

Blake, W. D., F. Goff, A. I. Adams, and D. Counce (1995), Environmentalgeochemistry for surface and subsurface waters in the Pajarito Plateauand outlying areas, New Mexico, Los Alamos Natl. Lab. Rep.LA-12912-MS.

Buason, T. (1972), Equation of isotope fractionation between ice and waterin a melting snow column with continuous rain and percolation, J. Gla-ciol., 11, 387–405.

Clark, I., and P. Fritz (1997), Environmental Isotopes in Hydrogeology,Lewis, Boca Raton, Fla.

Colbeck, S. C. (1987), Snow metamorphism and classification, in SeasonalSnowcovers: Physics, Chemistry, Hydrology, edited by H. G. Jones andW. J. Orville-Thomas, pp. 1–35, Springer, New York.

Coleman, M. L., T. J. Shepherd, J. J. Durham, J. E. Rouse, and G. R. Moore(1982), Reduction of water with zinc for hydrogen isotope analysis, Anal.Chem., 54, 993–995.

Cooper, L. W. (1998), Isotopic fractionation in snow cover, in IsotopeTracers in Catchment Hydrology, edited by C. Kendall and J. J.McDonnell, pp. 119–136, Elsevier, New York.

Craig, H., L. I. Gordon, and Y. Horibe (1963), Isotopic exchange effects inthe evaporation of water: 1. Low-temperature experimental results,J. Geophys. Res., 68, 5079–5087.

Cunningham, E. B., A. Long, C. Eastoe, and R. L. Bassett (1998), Mi-gration of recharge waters downgradient from the Santa Catalina Moun-tains into the Tucson Basin aquifer, Arizona, USA, Hydrogeol. J., 6,94–103.

Dansgaard, W. (1964), Stable isotopes in precipitation, Tellus, 16, 436–468.

Earman, S. (2004), Groundwater recharge and movement through moun-tain-basin systems of the Southwest: A case study in the ChiricahuaMountains-San Bernardino Valley system, Arizona and Sonora, Ph.D.dissertation, N. M. Inst. of Min. and Technol., Socorro.

Earman, S. B., J. O. L. McPherson, F. M. Phillips, S. Ralser, and J. M.Herrin (2003), An investigation of the properties of the San Bernardi-no groundwater basin, Arizona and Sonora, Mexico, technical comple-tion report, Hydrol. Program, N. M. Inst. of Min. and Technol.,Socorro.


16 of 18

D09302

Feng, X., S. Taylor, C. E. Renshaw, and J. W. Kirchner (2002), Isotopicevolution of snowmelt: 1. A physically based one-dimensional model,Water Resour. Res., 38(10), 1217, doi:10.1029/2001WR000814.

Flint, A. L., L. E. Flint, J. A. Hevesi, and J. B. Blainey (2004), Funda-mental concepts of recharge in the desert Southwest: A regional mod-eling prespective, in Groundwater Recharge in a Desert Environment:The Southwestern United States, Water Sci. Appl. Ser., vol. 9, editedby J. F. Hogan, F. M. Phillips, and B. R. Scanlon, pp. 159–184,AGU, Washington, D. C.

Friedman, I., C. Benson, and J. Gleason (1991), Isotopic changes duringsnow metamorphism, in Stable Isotope Geochemistry: A Tribute to Sa-muel Epstein, edited by H. P. Taylor, J. R. O’Neil, and I. R. Kaplan, Spec.Publ. Geochem. Soc., 3, 211–221.

Friedman, I., G. I. Smith, J. D. Gleason, A. Warden, and J. M. Harris(1992), Stable isotope composition of waters in southeastern California:1. Modern precipitation, J. Geophys. Res., 97, 5795–5812.

Gleick, P. H., and E. L. Chalecki (1999), The impacts of climatic changesfor water resources of the Colorado and Sacramento-San Joaquin Riverbasins, J. Am. Water Resour. Assoc., 35, 1429–1441.

Gonfiantini, R. (1986), Environmental isotopes in lake studies, Handbookof Environmental Isotope Geochemistry, vol. 2, The Terrestrial Environ-ment, B, edited by P. Fritz and J. C. Fontes, pp. 113–168, Elsevier, NewYork.

Gross, G. W., and R. Wilcox (1983), Groundwater circulation in theSocorro geothermal area, in Socorro Region II, New Mexico GeologicalSociety 34th Annual Field Conference Guidebook, edited by C. E. Chapinand J. F. Callender, pp. 311–318, N. M. Geol. Soc., Socorro.

Hage, K. D., J. Gray, and J. C. Linton (1975), Isotopes in precipitation innorthwestern North America, Mon. Weather Rev., 103, 958–966.

Hamlet, A. F., and D. P. Lettenmaier (1999), Effects of climate change onhydrology and water resources in the Columbia River Basin, J. Am.Water Resour. Assoc., 35, 1597–1623.

Herrmann, A., M. Lehrer, and W. Stichler (1981), Isotope input intorunoff systems from melting snow covers, Nord. Hydrol., 12, 309–318.

Hershey, R. L. (1989), Hydrogeology and hydrogeochemistry of the SpringMountains, Clark County, Nevada, M. S. thesis, Univ. of Nev., LasVegas.

Hinzman, L. D., and D. L. Kane (1992), Potential responses of an Arcticwatershed during a period of global warming, J. Geophys. Res., 97,2811–2820.

Hooper, R. P., and C. A. Shoemaker (1986), A comparison of chemicaland isotopic hydrograph separation, Water Resour. Res., 22, 1444–1454.

Ingraham, N. L., and R. E. Criss (1993), Effects of surface area and volumeon the rate of isotopic exchange between water and water vapor,J. Geophys. Res., 98, 20,547–20,553.

Ingraham, N. L., B. F. Lyles, R. L. Jacobson, and J. W. Hess (1991), Stableisotopic study of precipitation and spring discharge in southern Nevada,J. Hydrol., 125, 243–258.

Jouzel, J., and R. A. Souchez (1982), Melting-refreezing at the glacier soleand the isotopic composition of the ice, J. Glaciol., 28, 35–42.

Kirshen, P. H. (2002), Potential impacts of global warming on groundwaterin eastern Massachusetts, J. Water Resour. Plann. Manage., 128, 216–226.

Lawson, S. M. (1997), Sublimation from snow packs in Toiyabe NationalForest, Nevada and Dixie National Forest, Utah, M. S. thesis, Univ. ofNev., Las Vegas.

Leung, L. R., and M. S. Wigmosta (1999), Potential climate change impactson mountain watersheds in the Pacific Northwest, J. Am. Water Resour.Assoc., 35, 1463–1471.

Leung, L. R., Y. Qian, X. Bian, W. M. Washington, J. Han, and J. O. Roads(2004), Mid-century ensemble regional climate change scenarios for thewestern United States, Clim. Change, 62, 75–113.

Leydecker, A., and J. A. Melack (1999), Evaporation from snow in thecentral Sierra Nevada of California, Nord. Hydrol., 30, 81–108.

Loaiciga, H. A., D. R. Maidment, and J. B. Valdes (2000), Climate-changeimpacts in a regional karst aquifer, Texas, USA, J. Hydrol., 227, 173–194.

MacClune, K. L., A. G. Fountain, J. S. Kargel, and D. R. MacAyeal (2003),Glaciers of the McMurdo dry valleys: Terrestrial analog for Martianpolar sublimation, J. Geophys. Res., 108(E4), 5031, doi:10.1029/2002JE001878.

Majoube, M. (1971a), Fractionnement en 18O entre la glace et la vapeurd’eau, J. Chem. Phys., 68, 625–636.

Majoube, M. (1971b), Fractionnement en oxygene 18 et en deuterium entrel’eau et sa vapeur, J. Chem. Phys., 68, 1423–1436.

Martinec, J., H. Moser, M. R. de Quervain, W. Rauert, and W. Stichler(1977), Assessment of processes in the snowpack by parallel deuterium,tritium and oxygen-18 sampling, in Isotopes and Impurities in Snow and

Ice: Proceedings of the Grenoble Symposium, August/September 1975,IAHS Publ. 118, 220–231.

Maule, C. P., D. S. Chanasyk, and K. Muelenbachs (1994), Isotopic deter-mination of snow-water contribution to soil water and groundwater,J. Hydrol., 155, 73–91.

Merlivat, L., and G. Nief (1967), Fractionnement isotopique lors deschangements d’etat solide-vapeur et liquide-vapeur de l’eau a des tem-peratures inferieures a 0�C, Tellus, 19, 122–127.

Miller, J. R., and G. L. Russel (1992), The impact of global warming onriver runoff, J. Geophys. Res., 97, 2757–2764.

Moser, H., and W. Stichler (1980), Environmental isotopes in ice and snow,in Handbook of Environmental Isotope Geochemistry, vol. 1, The Terres-trial Environment, A, edited by P. Fritz and J. C. Fontes, pp. 141–178,Elsevier, New York.

Mote, P. W. (2003), Trends in snow water equivalent in the Pacific North-west and their climatic causes, Geophys. Res. Lett., 30(12), 1601,doi:10.1029/2003GL017258.

Nelson, S. T., and D. Dettman (2001), Improving hydrogen isotope ratiomeasurements for on-line chromium reduction systems, Rapid Commun.Mass Spectrom., 15, 2301–2306.

Newman, B. D., C. J. Duffy, and D. D. Hickmot (2001), Evaluating thehydrogeochemical response of springs using singular spectrum analysisand phase-plane plots, in New Approaches Characterizing GroundwaterFlow: Proceedings of the XXXI International Association of Hydrogeol-ogists Congress, vol. 2, edited by K.-P. Seiler and S. Wohnlich, pp. 763–777, A. A. Balkema, Brookfield, Vt.

Ojima, D., L. Garcia, E. Elgaali, K. Miller, T. G. F. Kittel, and J. Lackett(1999), Potential climate change impacts on water resources in the GreatPlains, J. Am. Water Resour. Assoc., 35, 1443–1454.

Rodhe, A. (1981), Spring flood: Meltwater or groundwater?, Nord. Hydrol.,12, 21–30.

Rose, T. P. (2003), Stable isotope investigation of precipitation and rechargeprocesses in central Nevada, in Hydrologic Resources Management Pro-gram and Underground Test Area Project FY2001–2002 Progress Re-port, Rep. UCRL-ID-154357, edited by T. P. Rose, G. F. Eaton, and A. B.Kersting, pp. 113–134, Lawrence Livermore Natl. Lab., Livermore, Ca-lif.

Rose, T. P., M. L. Davisson, R. E. Criss, and D. K. Smith (1999), Isotopicinvestigation of recharge to a regional groundwater flow system, GreatBasin, Nevada, USA, in International Symposium on Isotope Techniquesin Water Resources Development and Management, Rep. IAEA-CSP-2/C,session 2, pp. 63–72, Int. At. Energy Agency, Vienna.

Socki, R. A., H. R. Karlsson, and E. K. Gibson Jr. (1992), Extractiontechnique for the determination of oxygen-18 in water using preevacu-ated glass vials, Anal. Chem., 64, 829–831.

Sommerfeld, R. A., J. Clark, I. Friedman (1991), Isotopic changes dur-ing the formation of depth hoar in experimental snowpacks, in StableIsotope Geochemistry: A Tribute to Samuel Epstein, edited by H. P.Taylor, J. R. O’Neil, and I. R. Kaplan, Spec. Publ. Geochem. Soc., 3,205–209.

Stichler, W. (1987), Snowcover and snowmelt processes studied by meansof environmental isotopes, in Seasonal Snowcovers: Physics, Chemistry,Hydrology, edited by H. G. Jones and W. J. Orville-Thomas, pp. 297–308, Springer, New York.

Stichler, W., W. Rauert, and J. Martinec (1981), Enviromental isotopestudies of an Alpine snowpack, Nord. Hydrol., 12, 297–308.

Stichler, W., U. Schotterer, K. Frolich, P. Ginot, C. Kull, H. Gaggeler, andB. Pouyard (2001), Influence of sublimation on stable isotope recordsrecovered from high-altitude glaciers in the tropical Andes, J. Geophys.Res., 106, 22,613–22,620.

Szecsody, J. E., R. L. Jacobson, and M. E. Campana (1983), Environmentalisotopic and hydrogeochemical investigation of recharge and subsurfaceflow in Eagle Valley, Nevada, Publ. 42037, Desert Res. Inst. WaterResour. Cent., Reno, Nev.

Taylor, S., X. Feng, C. E. Renshaw, and J. W. Kirchner (2002), Isotopicevolution of snowmelt: 2. Verification and parameterization of a one-dimensional model using laboratory experiments, Water Resour. Res.,38(10), 1218, doi:10.1029/2001WR000815.

Unnikrishna, P. V., J. J. McDonnell, and C. Kendall (2002), Isotope varia-tions in a Sierra Nevada snowpack and their relation to meltwater,J. Hydrol., 260, 38–57.

Vaccaro, J. J. (1992), Sensitivity of groundwater recharge estimates toclimate variability and change, Columbia Plateau, Washington, J. Geo-phys. Res., 97, 2821–2833.

Wilson, L. G., K. J. DeCook, and S. P. Neuman (1980), Regional rechargeresearch for Southwest alluvial basins, unnumbered report, Water Resour.Res. Cent., Univ. of Ariz., Tucson.

Winograd, I. J., A. C. Riggs, and T. B. Coplen (1998), The relativecontributions of summer and cool-season precipitation to groundwaterrecharge, Spring Mountains, Nevada, USA, Hydrogeol. J., 6, 77–93.


17 of 18

D09302

Wright, W. E. (2001), dD and d18O in mixed conifer systems in the U. S.Southwest: The potential of d18O in Pinus ponderosa tree rings as anatural environmental recorder, Ph.D. dissertation, Univ. of Ariz.,Tucson.

Younger, P. L., G. Teutsch, E. Custudio, T. Elliot, M. Manzano, M. Sauter(2002), Assessments of the sensitivity to climate change of flow andnatural water quality in four major carbonate aquifers of Europe, inSustainable Groundwater Development, edited by K. M. Hiscock, M.O. Rivett, and R. M. Davison, Geol. Soc. Spec. Publ., 193, 303–323.

Yusoff, I., K. M. Hiscock, and D. Conway (2002), Simulation of the im-pacts of climate change on groundwater resources in eastern England, in

Sustainable Groundwater Development, edited by K. M. Hiscock, M. O.Rivett, and R. M. Davison, Geol. Soc. Spec. Publ., 193, 325–344.

��A. R. Campbell and F. M. Phillips, Earth and Environmental Science

Department, New Mexico Institute of Mining and Technology, MSEC 208,801 Leroy Place, Socorro, NM 87801, USA.S. Earman, Division of Hydrologic Sciences, Desert Research Institute,

2215 Raggio Parkway, Reno, NV 89512-1095, USA. ([email protected])B. D. Newman, Earth and Environmental Sciences Division, Los Alamos

National Laboratory, P. O. Box 1663, Los Alamos, NM 87545, USA.


18 of 18

D09302

Documents

Isotopic exchange between snow and atmospheric water vapor: Estimation of the snowmelt component of groundwater recharge in the southwestern United States