The distribution of meteoric 36Cl/Cl in the United States ...quebec.hwr.arizona.edu/research/moysey03-36cl.pdf · donnes du sud-est des tats-Unis, dont on pense qu’elles prsentent

The distribution of meteoric 36Cl/Cl in the United States:a comparison of models

Stephen Moysey · Stanley N. Davis · Marek Zreda ·L. DeWayne Cecil

Abstract The natural distribution of 36Cl/Cl in ground-water across the continental United States has recentlybeen reported by Davis et al. (2003). In this paper, thelarge-scale processes and atmospheric sources of 36Cl andchloride responsible for controlling the observed 36Cl/Cldistribution are discussed.

The dominant process that affects 36Cl/Cl in meteoricgroundwater at the continental scale is the fallout of stablechloride from the atmosphere, which is mainly derivedfrom oceanic sources. Atmospheric circulation transportsmarine chloride to the continental interior, where distancefrom the coast, topography, and wind patterns define thechloride distribution. The only major deviation from thispattern is observed in northern Utah and southern Idahowhere it is inferred that a continental source of chlorideexists in the Bonneville Salt Flats, Utah.

In contrast to previous studies, the atmospheric flux of36Cl to the land surface was found to be approximatelyconstant over the United States, without a strong corre-lation between local 36Cl fallout and annual precipitation.However, the correlation between these variables wassignificantly improved (R 2=0.15 to R 2=0.55) when datafrom the southeastern USA, which presumably havelower than average atmospheric 36Cl concentrations, wereexcluded. The total mean flux of 36Cl over the continentalUnited States and total global mean flux of 36Cl are

calculated to be 30.5€7.0 and 19.6€4.5 atoms m�2 s�1,respectively.

The 36Cl/Cl distribution calculated by Bentley et al.(1986) underestimates the magnitude and variabilityobserved for the measured 36Cl/Cl distribution acrossthe continental United States. The model proposed byHainsworth (1994) provides the best overall fit to theobserved 36Cl/Cl distribution in this study. A process-oriented model by Phillips (2000) generally overestimates36Cl/Cl in most parts of the country and has severalsignificant local departures from the empirical data.

R�sum� La distribution naturelle du rapport 36Cl/Cl dansles eaux souterraines des �tats-Unis a �t� r�cemmentpr�sent�e par Davis et al. (2003). Dans ce travail, lesprocessus � grande �chelle et les sources atmosph�riquesde 36Cl et de chlorure responsables du contr�le de ladistribution observ�e du rapport 36Cl/Cl sont discut�s. Leprocessus dominant qui affecte le rapport 36Cl/Cl dans leseaux souterraines d’origine m�t�orique � l’�chelle conti-nentale est l’apport atmosph�rique de chlorure stable, quiprovient pour l’essentiel de sources oc�aniques. Lacirculation atmosph�rique transporte des chlorures marinsvers l’int�rieur des continents, o� la distribution dechlorure est d�finie par la distance � la c�te, la topogra-phie et les r�gimes des vents. La seule exception majeure� ce sch�ma est observ�e dans le nord de l’Utah et le sudde l’Idaho o� l’on suppose qu’il existe une sourcecontinentale de chlorure dans les bas-fonds sal�s deBonneville. Au contraire de pr�c�dentes �tudes (Knies etal. 1994; Phillips 2000), on trouve que le flux atmosph�-rique de 36Cl vers le sol est approximativement constantsur l’ensemble des �tats-Unis, sans forte corr�lation entrela retomb�e locale de 36Cl et les pr�cipitations annuelles.Cependant, la corr�lation entre ces variables devientsignificative (R 2=0.15 � 0.55) lorsqu’on supprime lesdonn�es du sud-est des �tats-Unis, dont on pense qu’ellespr�sentent des concentrations en 36Cl atmosph�riqueinf�rieures � la moyenne. Le flux total moyen de 36Clsur les �tats-Unis continentaux et le flux moyen global de36Cl sont respectivement �valu�s � 30.5 € 7.0 et 19.6 € 4.5atomes.m–2.s–1. La distribution du rapport 36Cl/Cl calcu-l�e par Bentley et al. (1986) sous-estime l’ordre degrandeur et la variabilit� observ�s pour la distributionmesur�e du rapport 36Cl/Cl sur les �tats-Unis continen-taux. Le mod�le propos� par Hainsworth (1994) fournit le

Received: 21 October 2002 / Accepted: 9 July 2003Published online: 17 October 2003

� Springer-Verlag 2003

S. Moysey ()) · S. N. Davis · M. ZredaDepartment of Hydrology and Water Resources,University of Arizona,Tucson, AZ 85721, USAe-mail: [email protected].: +1-650-7236117Fax: +1-650-7257344

L. D. CecilUS Geological Survey,900 N. Skyline Drive, Suite C, Idaho Falls, ID 83402, USA

Present address:S. Moysey, Department of Geophysics,Stanford University,Stanford, CA 94305, USA

Hydrogeology Journal (2003) 11:615–627 DOI 10.1007/s10040-003-0287-z

meilleur ajustement d’ensemble � la distribution durapport 36Cl/Cl observ�e dans cette �tude. Un mod�leorient� vers les processus propos� par Phillips (2000)surestime dans l’ensemble le rapport 36Cl/Cl dans laplupart des r�gions du pays et pr�sente plusieurs d�sac-cords locaux avec les donn�es empiriques.

Resumen Davis et al. (2003) han informado de ladistribucin natural de la proporcin 36Cl/Cl en las aguassubterrneas de la parte continental de los Estados Unidosde Am�rica [EUA]. En este art�culo, se discute cules sonlos procesos a gran escala y las fuentes atmosf�ricas del36Cl y del cloruro que dan lugar a la distribucinobservada de 36Cl/Cl.

El proceso dominante que afecta a la relacin 36Cl/Clen las aguas subterrneas de origen meterico a escalacontinental es el aporte de cloruro estable desde laatmsfera, que procede principalmente de los oc�anos. Lacirculacin atmosf�rica transporta el cloruro marino haciael interior, donde la distancia a la costa, topograf�a ycorrientes del viento definen la distribucin del cloruro.La fflnica desviacin principal de este esquema tiene lugaral norte de Utah y en el sur de Idaho, donde se deduce queexiste una fuente continental de cloruro en los RellanosSalados de Bonneville (Salt Flats).

En contraste con estudios previos (Knies et al. 1994;Phillips 2000), se ha descubierto que el flujo atmosf�ricode 36Cl hacia la superficie terrestre es aproximadamenteconstante en todos los estados, sin deducirse una corre-lacin fuerte entre el aporte de 36Cl y la precipitacinanual. Sin embargo, la correlacin entre estas variables seve mejorada de forma significativa, con coeficientes deregresin comprendidos entre 0,15 y 0,55, cuando seexcluyen los datos recogidos en el sudeste de los EUA,que tienen concentraciones de 36Cl atmosf�rico presunta-mente inferiores a la media. El flujo medio total de 36Clcalculado en la zona continental de los Estados Unidosvale 30,5€7,0 tomos por metro cuadrado y segundo,mientras que el flujo total global de 36Cl es de 19,6€4,5tomos por metro cuadrado y segundo.

La distribucin de 36Cl/Cl calculada por Bentley et al.(1986) infravalora la magnitud y variabilidad observadaen los valores medidos a lo largo de los Estados Unidos.El modelo propuesto por Hainsworth (1994) proporcionael mejor ajuste conjunto a la distribucin observada de36Cl/Cl en este estudio. El modelo orientado a procesos dePhillips (2000) sobreestima por lo general la distribucinde 36Cl/Cl en la mayor�a del pa�s y difiere significativa-mente de algunos valores locales emp�ricos.

Keywords Atmospheric deposition · Chlorine-36 ·Groundwater · United States

Introduction

Quantitative studies of groundwater systems using envi-ronmental nuclides have become popular because of therealization that the distribution of such nuclides consti-

tutes a historical record of flow and transport processeswithin aquifers. To interpret this record one must firstunderstand the inputs of the isotopes to the hydrologicsystem under study (Cecil and Vogt 1997; Davis et al.1998). Such a need explains the extensive research thathas been directed at quantifying the sources and behaviorof both stable nuclides, such as 2H and 18O, andradionuclides, such as 3H, 14C, and 36Cl, in groundwater(Clark and Fritz 1997).

The use of 36Cl (half-life, t1/2=301 ka) in groundwaterstudies, however, has been hampered by a deficiency inour understanding of its spatial and temporal inputfunction to the subsurface (Fabryka-Martin et al. 1987;Beasley et al. 1993; Davis et al. 1998). In particular, thereare few studies of the spatial distribution of naturallyoccurring 36Cl in groundwater beyond the aquifer scale(e.g., Bentley et al. 1986; Phillips 2000). Moysey (1999)and Davis et al. (2003) inferred the natural distribution of36Cl/Cl over the continental United States from analysesof groundwater samples. In this paper, the naturalvariability seen in the empirical distribution is exploredand discussed with an emphasis on identifying the large-scale processes that are most important in shaping theobserved distribution. Additionally, the measured distri-bution is compared to three models of the 36Cl/Cldistribution (Bentley et al. 1986; Hainsworth 1994;Phillips 2000).

The empirical 36Cl/Cl distribution presented in thispaper is based on that of Moysey (1999) and is essentiallythe same as that given by Davis et al. (2003); both havebeen generated from the same database. These data,however, are treated differently in constructing the two36Cl/Cl models. Moysey (1999) interpreted clusters ofdata jointly to determine regional estimates of 36Cl/Cl andassigned a weight to each of these values according to aqualitative measure of reliability. Davis et al. (2003) usedonly data from the samples that have the lowest chlorideconcentrations and that are believed to be free fromanthropogenic influence. The consistency between the36Cl/Cl maps produced by each of these studies, usingdifferent data interpretation techniques, shows that theunderlying data set is robust. The minor differences thatexist between the two models do not change theconclusions in this paper.

Origin of 36Cl and Stable Chloride in Nature

Most 36Cl found in actively circulating meteoric ground-water, not affected by anthropogenic sources, originatedin the atmosphere. The spallation of 40Ar by cosmic rays(Jiang et al. 1990; Lehmann et al. 1993) is the primarynatural atmospheric source of 36Cl. Subsequent toproduction, 36Cl atoms are deposited at the land surfaceand eventually carried to the subsurface by recharginggroundwater. In some cases, additional 36Cl may also beintroduced through secondary production at or below theground surface (Bentley et al. 1986; Phillips 2000).

616


About 60–75% of atmospheric 36Cl is produced in thestratosphere with the remainder produced in the tropo-sphere (Lal and Peters 1967; Bentley et al. 1986).Meridional circulation of the atmosphere causes themajority of exchange between the stratosphere andtroposphere to take place at mid-latitudes (Reiter 1975;Seinfeld and Pandis 1998). Once in the troposphere, 36Clreaches the ground surface relatively quickly as either wet(i.e., associated with precipitation) or dry fallout. Lal andPeters (1967) developed a model for the latitudinaldependence of deposition based on observations of falloutproduced during thermonuclear weapons testing in the1950s.

Precipitation and evaporation rates in the recharge areaof an aquifer can vary greatly through time, so ground-water studies often consider 36Cl/Cl rather than absolute36Cl concentrations. This ratio is insensitive to changes inthe supply of recharge water while the absolute concen-tration is not. Understanding variability in 36Cl/Cl,therefore, requires knowledge of the historical sourcesfor stable chloride as well as those for 36Cl.

The most important sources of chloride to the conti-nental landmasses are sea-spray and salt-based aerosolsderived from the oceans (Eriksson 1960; Semonin andBowersox 1982; Li 1992; Simpson and Herczeg 1994).The deposition of chloride from air masses originatingover the oceans decreases approximately exponentiallywith distance from the coast (Slinn et al. 1982; Simpsonand Herczeg 1994). With the probable exception ofextensive playas in northwestern Utah, continentalsources of wind-blown chloride appear to be only oflocal importance (Wood and Sanford 1995).

Methods

This study uses 36Cl/Cl measured in 183 groundwatersamples collected from 38 locations across the UnitedStates. Sampling locations were selected to obtain waterrecharged between 50–10,000 years ago. The younger agelimit was chosen to minimize the influence of anthro-pogenic 36Cl recharged after 1952 (Bentley et al. 1986;Davis et al. 1998; Phillips 2000); the older age limit waschosen because prior to 10,000 years ago, the globalproduction rate of 36Cl in the atmosphere was signifi-cantly different than it is today (Plummer et al. 1997;Phillips 2000). Waters with low chloride concentrationswere also targeted to minimize dilution of 36Cl/Cl due tothe addition of connate chloride or dissolution of haliteand other minerals in the subsurface. In general, clustersof samples were collected within a given region so thatboth local and continental scale variability in 36Cl/Clcould be investigated. A description of the project andsite-selection objectives is given by Davis et al. (2003)and specific site descriptions are provided by Moysey(1999).

Samples for 36Cl/Cl measurement were prepared at theUniversity of Arizona and measured by accelerator massspectrometry (AMS) at PRIME Lab, Purdue University.

All analytical results, as well as detailed descriptions ofanalytical procedures, are given by Moysey (1999).Unless otherwise stated, all 36Cl/Cl reported in this workhave been multiplied by 1015 and, therefore, have theimplied units of x10-15atoms 36Cl/atoms Cl.

Results

The regional estimate of 36Cl/Cl for each of the 38 studyareas is given in Table 1; each regional value representsthe synthesis of between 1 and 12 samples. The 36Cl/Clreported for each region was assigned based on datacollected during this study, existing geochemical andhydrologic models of the local flow system, and infor-mation from the literature. The use of multiple samplesfrom an area to provide a context for the determination ofthe regional meteoric 36Cl/Cl signal is critical becausevariability at the local aquifer scale, caused by processessuch as the dissolution of minerals or secondary 36Clproduction in the subsurface, can be of the same order ofmagnitude as continental scale variability (Moysey 1999).The “confidence index” (CI), also reported in Table 1, is aqualitative indicator of the reliability of the regional 36Cl/Cl estimate and was assigned based on the amount of datacollected from the area, the current understanding of theaquifer system as inferred from the literature, how wellthe data from this study can be explained by models of theaquifer system, and whether the data show signs of beinginfluenced by anthropogenic or subsurface sources of 36Clor chloride. A CI rank of 1 implies high confidence in the36Cl/Cl estimate, while a rank of 3 implies low confi-dence. Moysey (1999) gives details of how each regional36Cl/Cl estimate in Table 1 was determined.

Figure 1 shows the locations of the 38 regional-controlpoints used in this study. Nine control points wereconsidered unreliable (CI =3), and were not used for theanalyses discussed in this paper. Thirteen previouslypublished studies that obtained estimates of the meteoric36Cl/Cl signal (Table 2) are also used in this study. Thedistribution of 36Cl/Cl derived from these measurementsis given in Fig. 1. The isopleths of 36Cl/Cl in this figurewere drawn by hand because automated interpolationtechniques are not justified given the degree of uncer-tainty related to the 36Cl/Cl estimates and the relativelysparse distribution of data points compared to the small-scale spatial variability of the underlying process. The36Cl/Cl map presented here, therefore, is not meant to be aquantitative estimate of local 36Cl/Cl values, but rather actas a tool for understanding the large-scale processes thatcontrol the 36Cl/Cl distribution.

Note that the map shown in Fig. 1 is similar to thatinterpreted by Davis et al. (2003). The most significantdiscrepancy between the two 36Cl/Cl maps is a shift in thelocation of the maximum observed 36Cl/Cl approximately400 km (~250 miles) to the north in Fig. 1 compared withthe map of Davis et al. (2003). In general, the differencesbetween the two maps do not impact the conclusions ofthis paper.

617


Tab

le1

36C

l/C

lfr

omth

epr

esen

tst

udy

(reg

iona

les

tim

ates

)

Gro

upID

Gro

upna

me

Lat

. N

Lon

g. W

36C

l/C

l36

Cl/

Cl

Err

ora

Con

fide

nce

inde

x(C

I)b

Pre

cipi

tati

onc

(cm

/yea

r)W

etch

lori

dede

posi

tion

dW

et36

Cl

depo

siti

one

Tot

al36

Cl

depo

siti

one

(kg

ha-1

year

-1)

(ato

ms

m-2

s-1)

(ato

ms

m-2

s-1)

1W

Geo

rgia

32.5

83.7

156

81

159

3.1

26.0

35.0

2E

Geo

rgia

33.2

82.5

975

117

04.

020

.928

.13

NF

lori

da30

.484

.338

917

3–

6.5

136.

218

2.8

4F

lori

da27

.581

.545

22

160

10.0

24.2

32.5

5F

lori

da29

.181

.818

325

23

–13

.012

8.1

172.

06

Ark

ansa

s35

.390

.613

998

214

414

.010

4.8

140.

77

Mis

siss

ippi

31.2

89.4

503

213

86.

818

.324

.68

Okl

ahom

a35

.697

.427

778

284

1.7

25.4

34.0

9N

orth

Dak

ota

46.0

97.8

900

100

145

0.4

20.4

27.3

10S

outh

Dak

ota

44.4

103.

512

1519

11

350.

423

.631

.611

Neb

rask

a41

.110

0.8

493

192

370.

25.

37.

112

Wes

tV

irgi

nia

39.6

78.2

386

202

130

2.0

41.6

55.8

13N

ewY

ork

43.1

73.8

94

3–

1.5

0.7

1.0

14S

t..D

avid

(AZ

)31

.911

0.2

364

162

300.

917

.623

.715

SU

tah

37.2

113.

640

332

225

0.5

10.9

14.6

16N

Uta

h41

.911

1.6

699

961

3–

15.0

564.

875

8.1

17Id

aho

44.1

111.

335

230

92

300.

713

.317

.818

EO

rego

n44

.412

1.2

499

202

210

03.

080

.610

8.2

19W

Ore

gon

45.0

123.

021

916

42

200

6.0

70.8

95.0

20W

ashi

ngto

n45

.712

1.5

4752

3–

16.0

40.5

54.4

21W

Col

orad

o39

.210

4.8

952

142

240

0.5

23.1

31.0

22K

ansa

s38

.097

.550

020

02

601.

437

.750

.623

EC

olor

ado

39.2

102.

921

2362

43

–0.

445

.761

.424

Tah

oe(C

A)

39.1

120.

095

251

72

150.

27.

710

.325

Fre

sno

(CA

)36

.711

9.6

219

71

841.

618

.925

.326

Whi

teM

ount

ains

(AZ

)33

.810

9.1

1842

933

–0.

329

.840

.027

NW

yom

ing

43.8

104.

293

023

31

350.

418

.024

.228

SW

yom

ing

41.2

104.

812

1060

224

0.4

24.1

32.4

29In

dian

a41

.085

.227

575

184

1.5

22.2

29.8

30O

hio

40.1

83.8

252

147

3–

1.5

19.7

26.4

31K

entu

cky

37.5

83.1

218

752

130

1.5

17.4

23.3

32E

Ala

bam

a32

.085

.611

035

114

03.

520

.727

.833

WA

laba

ma

32.7

87.6

129

161

120

3.7

25.7

34.5

34M

onta

na48

.311

4.3

1,08

716

31

670.

317

.623

.635

Iow

a42

.492

.849

210

81

700.

821

.228

.536

Wis

cons

in43

.189

.536

821

53

–0.

917

.823

.937

Min

neso

ta45

.193

.265

715

91

750.

619

.526

.138

Ari

zona

32.2

110.

937

317

140

1.0

20.1

27.0

aT

here

port

eder

ror

isth

ela

rges

tof

the

anal

ytic

aler

ror,

grou

pst

anda

rdde

viat

ion,

orm

odel

erro

rb

CI

isa

subj

ecti

vem

easu

reof

conf

iden

cein

the

regi

onal

esti

mat

eof

36C

l/C

l;C

I=1

isth

ehi

ghes

tco

nfid

ence

,C

I=3

isth

elo

wes

tc

Pre

cipi

tati

onva

lues

are

esti

mat

edfr

omth

eN

AD

P(2

000)

tota

lpr

ecip

itat

ion

map

for

1994

dW

etch

lori

dede

posi

tion

ises

tim

ated

from

the

mea

nof

NA

DP

(200

0)m

easu

rem

ents

betw

een

1979

–199

7e

Cal

cula

ted

valu

es.

Tot

alde

posi

tion

assu

mes

that

wet

depo

siti

onac

coun

tsfo

r74

%of

tota

lfa

llou

t

618


The 36Cl/Cl ranges from a minimum of less than 100near the oceans to a maximum of about 1,200 in theinterior of the continent over the central High Plainsregion of Wyoming (Fig. 1). The 36Cl/Cl isopleths areapproximately parallel to the west coast, where they forma steep inland gradient. In contrast, 36Cl ratios in theeastern United States slowly increase with distance fromthe Gulf of Mexico and northward along the easternseaboard, showing only secondary influence from theAtlantic coastline. The specific arrangement of 36Cl/Clisopleths along the northeastern coastline is uncertain dueto a lack of data in this area. A region of low 36Cl/Cl isalso apparent in northwestern USA (northern Utah andsouthern Idaho).

The pattern observed for the spatial distribution of36Cl/Cl is similar to that measured by the NationalAtmospheric Deposition Program (NADP 2000) forstable-chloride deposition (Fig. 2), though the magnitudesare inversely related. A notable exception is along theeastern seaboard where the chloride contours tend to moreclosely parallel the coastline. The strong correlationbetween 36Cl/Cl and chloride deposition is shown quan-titatively in Fig. 3 for data obtained in this project as well

as 13 other studies conducted around the United States(Table 1). Based on this high level of correlation, thedeposition of stable chloride is inferred to be the primarycontrol on spatial variability of 36Cl/Cl.

Most of the chloride that is deposited on the continentoriginates as particulates released to the atmosphere fromthe oceans (Eriksson 1960; Semonin and Bowersox 1982;Li 1992; Simpson and Herczeg 1994). Once chloride issuspended in the atmosphere, it can be carried inland bywinds. Atmospheric chloride availability decreases expo-nentially with distance from the ocean, resulting in higherchloride deposition rates and lower 36Cl/Cl near the coastcompared with the interior of the continent. Synopticscale weather patterns and topographic barriers also playan important role in controlling the migration of chlorideparticulates over the continent (Semonin and Bowersox1982). The western mountain ranges act as an effectivebarrier to the inland movement of air masses carryingchloride. This results in the sharp gradient in 36Cl/Clalong the west coast, as shown in Fig. 1. In the easternUSA, northward-trending storms are generally limited tocoastal areas while air masses moving north from the Gulfof Mexico are the primary source of moisture for the

Fig. 1 Empirical 36Cl/Cl distribution in the United States inferred from groundwater measurements. 36Cl/Cl are shown with control pointID from Table 1 in parenthesis

619


interior of the continent (Barry and Chorley 1992). Theseair masses, which also carry chloride from the Gulf ofMexico are, therefore, the predominant influence onchloride deposition in the southeastern interior of theUSA, resulting in the broad Gulf-dominated isoplethsshown in Fig. 1.

The low 36Cl/Cl observed near the border of Utah andIdaho is the only clear exception to the oceanic forcingdescribed above. This deviation from the continentalpattern is likely caused by a continental source ofchloride, predominantly the Bonneville Salt Flats innorthern Utah. Local winds can pick up and transportparticles from the salt flats resulting in abnormally highchloride deposition in this area. This effect is apparent inchloride deposition measurements that are locally two tosix times higher than those measured in nearby regions(NADP 2000).

The historical depositional flux of 36Cl may beestimated by taking the product of the 36Cl/Cl ratios,measured in groundwater, and modern chloride-deposi-tion rates. This approach relies on two assumptions: (1)after their initial deposition, 36Cl and chloride behaveconservatively with no subsurface sources, and (2)modern chloride deposition is representative of theaverage chloride deposition over the last 10,000 years.From the data in Tables 1 and 2, excluding Arkansas andeastern Oregon (items 6 and 18 in Table 1), the mean wet36Cl flux over the United States is 22.7€1.8 atoms m-2 s-1.The deposition rates from eastern Oregon and Arkansaswere not included in the calculation of the mean flux asthese values were found to be statistical outliers usingChauvenet’s Criterion (Taylor 1982).

The chloride-deposition rates measured by the NADP(Fig. 2) reflect only the wet component of modernchloride fallout and, therefore, the calculated 36Cl fluxrepresents a “wet-only” estimate. Few reliable estimatesof chloride deposition currently exist for the dry compo-nent of fallout over the continent. Using the wet 36Cl fluxcalculated above and the slope of the regression line fromFig. 3 (16.9€1.2 atoms m-2 s-1), it is possible to estimatethat 26€11% of the total mean 36Cl fallout for thecontinental United States occurs as dry deposition. This isin good agreement with Eriksson (1960) and Bentley et al.(1986) who suggested that dry fallout should be approx-imately 23% of total deposition. Taking the relativefraction of dry deposition to be 26€11%, the total meanflux of 36Cl over the United States is 30.5€7.0 atoms m-

2 s-1. The resulting 36Cl fluxes calculated in this way(Tables 1 and 2) agree with the model used by Bentley etal. (1986) as they show little dependence on latitude(Fig. 4) and closely match the deposition rate used bythese authors. If the latitudinal dependence proposed byLal and Peters (1967; Fig. 4) is assumed to hold, thenthe estimated global “wet-only” fallout of 36Cl is14.6€1.1 atoms m-2 s-1 and the total global fallout, i.e.,combined wet and dry deposition, is 19.6€4.5 atoms m-2 s-1.

Tab

le2

36C

l/C

lfr

ompr

evio

usly

publ

ishe

dm

easu

rem

ents

IDS

tate

36C

l/C

lP

reci

pita

tion

aW

etch

lori

dede

posi

tion

bW

et36

Cl

depo

siti

onc

Tot

al36

Cl

depo

siti

ond

Ref

eren

ce(c

m/y

ear)

(kg

ha-1

year

-1)

(ato

ms

m-2

s-1)

(ato

ms

m-2

s-1)

AZ

1A

rizo

na88

245

0.3

14.3

19.1

Lop

esan

dH

offm

ann

(199

7)A

Z2

Ari

zona

379

401.

020

.427

.4B

entl

eyet

al.

(198

6)C

AC

alif

orni

a75

0d11

01.

040

.454

.2P

hill

ips

etal

.(1

995)

FL

Flo

rida

6014

06.

019

.426

.0P

help

s(1

997,

pers

onal

com

mun

icat

ion)

IDId

aho

580

250.

825

.033

.5C

ecil

etal

.(1

992)

INIn

dian

a40

094

1.1

23.7

31.8

Vog

tet

al.

(199

4)M

DM

aryl

and

252

126

4.0

54.3

72.9

Pur

dyet

al.

(198

7)N

M1

New

Mex

ico

465

410.

512

.516

.8P

hill

ips

etal

.(1

988)

NM

2N

ewM

exic

o72

030

0.4

15.5

20.8

Phi

llip

set

al.

(198

8)N

VN

evad

a51

940

0.8

22.4

30.0

Nor

ris

etal

.(1

987)

SC

Sou

thC

arol

ina

9015

54.

019

.426

.0B

easl

eyet

al.

(199

2)T

NT

enne

ssee

177

140

2.2

21.0

28.2

Vou

rvop

oulo

uset

al.

(199

0)T

X1

Tex

as46

041

0.7

17.3

23.3

Sca

nlon

(199

0)T

X2

Tex

as64

475.

017

.223

.1B

entl

eyet

al.

(198

6)W

AW

ashi

ngto

n73

550

0.5

19.8

26.6

Pry

ch(1

992,

pers

onal

com

mun

icat

ion)

aP

reci

pita

tion

valu

esar

ees

tim

ated

from

the

NA

DP

(200

0)to

tal

prec

ipit

atio

nm

apfo

r19

94b

Wet

chlo

ride

depo

siti

onis

esti

mat

edfr

omth

em

ean

ofN

AD

P(2

000)

mea

sure

men

tsbe

twee

n19

79–1

997

cC

alcu

late

dva

lues

.T

otal

depo

siti

onas

sum

esth

atw

etde

posi

tion

acco

unts

for

74%

ofto

tal

fall

out.

dO

rigi

nall

yes

tim

ated

from

ach

lori

debu

dget

.

620


Discussion

Previous attempts at modeling 36Cl/Cl variability over theUnited States have been made by Bentley et al. (1986),Hainsworth (1994), and Phillips (2000). The approachtaken in constructing these models has been to estimatethe fallout of 36Cl and stable chloride as a function ofgeographic location and then combine these to produce amap of the resulting 36Cl/Cl. In all three cases, theinvestigators have included measured stable chloridedeposition over the United States as part of their

calculations; Bentley et al. (1986) used chloride falloutfrom Eriksson (1960), and Hainsworth (1994) and Phillips(2000) used data from the NADP (2000).

Bentley et al. (1986) and Phillips (2000) used thelatitude dependence of the 36Cl depositional flux (Fig. 4)proposed by Lal and Peters (1967) in their models (Fig. 5a, c). Each author, however, used a different estimate ofthe global 36Cl production rate (16 atoms m-2 s-1 forBentley et al. 1986; 30 atoms m-2 s-1 for Phillips 2000).Phillips (2000) also attempted to account for variability of36Cl deposition by allowing local deviations from the

Fig. 2 Isopleth map of averageannual wet fallout of stablechloride (kg/ha) in the UnitedStates for the period 1979–1997(data from NADP 2000)

Fig. 3 Correlation between36Cl/Cl and chloride deposition.The single most important fac-tor controlling 36Cl/Cl is stable-chloride deposition. Closed cir-cles are from this study (Ta-ble 1); open circles representdata from 13 previously pub-lished studies (Table 2). Thesamples from Nebraska andLake Tahoe (items 11 and 24from Table 1) appear as outliersin the figure and were notincluded in the regression.Chloride deposition was in-ferred from measurements bythe NADP (2000)

621


mean flux proportional to the difference between localprecipitation and the average precipitation rate for a givenlatitude. Hainsworth (1994; Fig. 5b) took a considerablydifferent approach; she attributed 70% of the geographicvariability of fallout to the transfer of 36Cl between thestratosphere and troposphere. The stratospheric compo-nent of the 36Cl flux was calibrated using the measuredfallout of 90Sr produced by thermonuclear weapons tests.The tropospheric component of 36Cl fallout at a givenlocation was assumed to be proportional to the ratio of themean local rainfall to the mean zonal-precipitation rate.Hainsworth (1994) used the global 36Cl flux calculated byBlinov (1988) of 19 atoms m-2 s-1.

All three models qualitatively reproduce the mainfeatures of the 36Cl/Cl distribution observed in this study.Low ratios occur near the coast and high ratios occur inthe interior of the continent (Fig. 5). The reproduction ofthis large-scale variability is tied to the dominating effectof stable-chloride fallout. All of the models, nevertheless,deviate somewhat from the empirical values listed inTable 1 and shown in Fig. 1. The significance of some ofthese differences, such as the behavior of the 36Cl/Cl

isopleths along the northeast coastline, cannot be assesseddue to insufficient data. However, each of the Bentley etal. (1986), Hainsworth (1994), and Phillips (2000) modelsdo exhibit several features that are significantly differentfrom the empirical distribution in Fig. 1.

The Bentley et al. (1986) model generally underesti-mates the empirical 36Cl/Cl. This model (Fig. 5a) reachesa maximum 36Cl/Cl of only 640 and predicts a regionwhere the distribution is approximately constant in theinterior of the United States. These low 36Cl/Cl ratios areprimarily related to overestimation of stable-chloridefallout. The chloride-deposition rates used by Bentley etal. (1986) were taken from Eriksson (1960). Comparisonwith modern data collected by the NADP (2000) indicatesthat the Eriksson chloride deposition values overestimatethe flux in the interior of the United States, leading to thelow 36Cl/Cl predicted by Bentley et al. (1986).

The model by Hainsworth (1994) in Fig. 5b capturessimilar patterns of variability as seen in the empiricalresults (Fig. 1). The maximum 36Cl/Cl predicted by hermodel (1,600), however, is somewhat higher than thatobserved in our study (~1,200). The significance of thisdiscrepancy is difficult to evaluate given the lack of 36Cl/Cl measurements from eastern Montana and NorthDakota. The model of Hainsworth (1994) also tends tounderestimate 36Cl/Cl in the western and southwesternparts of the country. The cause of this underestimation isuncertain, though it could be related to annual fluctuationsin the stable-chloride measurements used in the model,differences between the spatial distribution of 36Cl falloutand that of 90Sr, or over correction of the 36Cl flux toaccount for precipitation.

The 36Cl/Cl map constructed by Phillips (2000), shownin Fig. 5c, is considerably different from the patternshown in the empirical data (Fig. 1) in four ways. First,the irregular tight curves in the isopleths in California andNevada reflect large regional variations in precipitationthat control the 36Cl/Cl values calculated for the Phillips(2000) model. This type of small-scale variability cannotbe captured by the empirical data due to the sparsity ofsample locations. Second, the Phillips (2000) modelshows only small variations in 36Cl/Cl over a large regionencompassing northern Colorado, western Nebraska andSouth Dakota, and all of Wyoming. The empirical data(Fig. 1) show large variations in the same region. Anexplanation of the contrast is not evident. Third, themeasured distribution (Fig. 1) shows that a zone of low36Cl/Cl exists in northern Utah and southeastern Idaho.This probably reflects contributions of “dead” chloride,i.e., 36Cl/Clffi0, from the Bonneville Salt Flats, Utah, andother playas in the region. This zone is not evident in thePhillips (2000) model. Fourth, values of 36Cl/Cl in thePhillips (2000) model (Fig. 5c) generally tend to overes-timate those of the empirical result (Fig. 1), especially inthe southeastern USA (e.g., Mississippi, Alabama, Geor-gia, and Florida) and the Midwest (e.g., Minnesota, Iowa,and Indiana) where the predicted 36Cl/Cl are roughlytwice the observed values. This is a large-scale trend thatillustrates a significant difference between the Phillips

Fig. 4 Plot of 36Cl vs. latitude. No correlation can be observedbetween the 36Cl flux and latitude given the range of this study. The36Cl flux was calculated as the product of 36Cl/Cl and stable-chloride fallout (NADP 2000) assuming that dry depositionaccounts for 26% of total fallout. The solid line is the model fromBentley et al. (1986) after correction of dry deposition from 23% to26%; closed circles are for qualitatively more reliable data (i.e.,confidence index (CI) =1), open circles are for qualitatively lessreliable data (i.e., CI =2)

622


Fig. 5 Isopleth maps of 36Cl/Clbased on previously publishedmodels of 36Cl/Cl (x10-15) in theUnited States: a Bentley et al.(1986); b Hainsworth (1994);c Phillips (2000)

623


(2000) model and the empirical results of the currentstudy. Parts of the West and Southwest (e.g., regions ofArizona, New Mexico, Nevada, Utah, Colorado, Wyom-ing, and South Dakota) are an exception to this trend asthe predicted and observed 36Cl/Cl are of similar magni-tude in these areas.

The higher 36Cl ratios predicted by Phillips (2000)could be related to the relatively high estimate of themean global 36Cl flux used in his model (30 atoms m-2

s-1) compared with that found in this study (~20 atoms m-2

s-1). The precipitation correction to the 36Cl flux used inthe Phillips model would have reduced the effective 36Cldeposition rate in the Southwest, which could explainwhy the model and empirical results are in betteragreement in this region. Several estimates of the globalmean 36Cl flux are given in Table 3. The theoreticallycalculated 36Cl production rates are in good agreementwith the flux calculated in this study, particularly thecalculation of 19 atoms m-2 s-1 by Masarik and Beer(1999). Estimates of the 36Cl flux derived from measure-ments are highly variable (Table 3).

The correction of the 36Cl flux for local variations,suggested by Phillips (2000), relates local deviations frommean latitudinal precipitation to local deviations in meanlatitudinal 36Cl deposition:

Dm36 ¼ �D36ðlÞ þ SDðlÞ Plocal � �PðlÞð Þ ð1Þ

where Dm36 is the local or measured 36Cl flux, �D36ðlÞ is the

mean 36Cl flux for a given latitude band l, Plocal is thelocal mean annual precipitation rate, and �PðlÞ is the meanprecipitation rate for a given latitude band. The parameterS D(l) describes the relationship between 36Cl depositionand precipitation rate for a given latitude band, i.e., is theslope of a regression line on a plot of the 36Cl flux vs.precipitation. This correction assumes a linear depen-dence between 36Cl deposition and precipitation. In thecurrent study, a strong linear correlation is not observedbetween the 36Cl deposition rate and precipitation(Fig. 6). The reason for this lack of correlation is unclear;however, it may be related to spatial variability ofdepositional processes.

The net flux of a constituent from the atmosphere tothe land surface at a given location can be expressed asthe sum of wet and dry deposition rates, which are afunction of concentration (Davidson 1989):

FT ¼ Fwet þ Fdry ¼ CPPþ vCa � KPþ v�ð Þ�Ca ð2Þwhere FT, Fwet, and Fdry are the net, wet and drydepositional fluxes of the constituent to the surface. Theconcentration of the constituent in precipitation is givenby C P, Ca is the concentration in air at the groundsurface, �Ca is the mean air-phase concentration over theentire atmospheric column, K is a vertically averagedpartitioning coefficient which describes scavenging be-tween the air and precipitation phase, P is the localprecipitation rate, and v is a dry deposition velocity,which describes processes other than precipitation thatcarry the constituent to the ground surface. The lastvariable in Eq. (2) is v*, which is a dry deposition velocitythat is scaled by the ratio between surface-air concentra-tion and mean-air concentration (i.e., v� ¼ v � Ca=�Cað Þ.

From Eq. (2), it is inferred that for the linearrelationship expressed in Eq. (1) to hold, the parametersK, v*, and �Ca must be approximately constant within agiven latitude band or vary in a complementary manner. Itis likely that there are many situations in nature wherethese conditions will not hold. For example, the atmo-spheric parameters which control deposition, i.e., K andv*, are spatially variable because site-specific conditions,such as topography, have an important effect on localatmospheric conditions, such as turbulence. Furthermore,and perhaps more importantly, if the atmospheric con-centration of 36Cl (i.e.,�Ca) varies spatially within alatitude band, then the net 36Cl flux cannot dependlinearly on precipitation. Spatial variability of the atmo-spheric 36Cl concentration can be expected if there existsspatial heterogeneity in the mean location of strato-sphere–troposphere exchange within a given latitude belt.

At a given location, the deposition parameters (K and v*)and atmospheric 36Cl concentration may be approximatelyconstant through time. According to Eq. (2), this wouldresult in a linear relationship between precipitation and the36Cl flux for a particular site. The high degree of linear

Table 3 Previously publishedestimates of the global 36Cl flux

Global production/fallout rate

Estimation method Hemisphere ofmeasurement

Reference

(atoms m-2 s-1)

11 Calculated – Lal and Peters (1967)19 Calculated – Masarik and Beer (1999)19 Calculated – Blinov (1988)17–26a Calculated – Oeschger et al. (1969)14–24a Calculated – Huggle et al. (1996)24 Measured Northern Andrews et al. (1994)28 Measured Northern Hainsworth et al. (1994)30 Measured Both Phillips (2000)15 Measured Southern Keywood et al. (1998)10–14 Measured Southern Nishiizumi et al. (1979)

Finkel et al. (1980)Nishiizumi et al. (1983)

a Minimum and maximum production rates reflect changes in activity due to sunspot cycle

624


correlation observed in the data of Knies et al. (1994) maybe the result of such an effect. The opposite was shown byHainsworth et al. (1994) who found little correlationbetween precipitation and 36Cl fallout at a site in Maryland,USA.; they concluded that seasonal differences in thetropospheric concentration of 36Cl were responsible for thepoor correlation. Recently, Sterling (2000) has suggestedthat variability in the atmospheric concentration of chlorideleads to the need for spatially and temporally dependentdeposition parameters [i.e., SD(l) in Eq. (1)] to accuratelymodel the deposition of chloride. It is likely that the same istrue for modeling the deposition of 36Cl.

Phillips (personal communication, 2002) has speculat-ed that the southeastern United States may have anoma-lously low atmospheric 36Cl concentrations. This is due tothe persistent inland flow of air masses from lowerlatitudes of the Atlantic, where 36Cl availability isdiminished (Fig. 4). Equation (2) indicates that, for agiven precipitation rate, lower atmospheric 36Cl concen-trations will lead to lower 36Cl deposition rates. Since theprecipitation rate in the southeastern USA is higher thanthe continental average, a low atmospheric 36Cl concen-tration could result in a 36Cl flux for this region that isequal to the mean flux over the United States, as observedin Fig. 6. By excluding the data from the southeastern partof the country (i.e., Florida, South Carolina, Tennessee,Georgia, Alabama, Kentucky, and Mississippi), the coef-ficient of determination for a linear regression is signif-icantly higher (R2=0.55) than when this region is retainedin the data set (R2=0.15). This improvement in thecorrelation between 36Cl deposition and precipitation is a

good indicator that factors, such as atmospheric 36Clconcentrations, must be considered in addition to precip-itation when modeling 36Cl deposition.

Conclusions

The use of 36Cl as a hydrologic tracer requires that theinput of this nuclide to groundwater systems be known.As a result, an intensive groundwater sampling programwas undertaken by Davis et al. (2003) to estimate 36Cl/Clin the United States. Significant variability in 36Cl/Clexists at the aquifer scale due to processes such asdissolution of chloride in the subsurface and mixing ofyoung water, which contains an anthropogenic componentof 36Cl, with old water having natural levels of 36Cl. Byattempting to remove the effects of these processes, thenatural distribution of 36Cl/Cl in meteoric groundwateracross the continental United States was inferred byMoysey (1999) and Davis et al. (2003).

Processes that are responsible for the continental scaledistribution of 36Cl and chloride fallout from the atmo-sphere ultimately control the meteoric 36Cl/Cl signal ingroundwater. The single most important factor controllingthe 36Cl/Cl distribution is the spatial dependence of stable-chloride fallout, which can be shown to statisticallyaccount for almost all of the variability observed in 36Cl/Cl across the United States. The lowest 36Cl/Cl ratios(<100) are observed in coastal areas, which coincide withthe location of high chloride deposition. The influence ofthe oceans steadily decreases inland and a maximum 36Cl/Cl of around 1,200 is reached over the central High Plains

Fig. 6 Plot of 36Cl depositionrate vs. precipitation. A strongcorrelation between 36Cl falloutand annual precipitation, asproposed by Phillips (2000), isnot observed for data distribut-ed throughout the United States(R2=0.15). The correlation issignificantly improved whendata from the southeasternUSA, presumably an area withlow atmospheric 36Cl concen-trations, are not included in theregression (R2=0.55). Previous-ly published values refer tothose studies listed in Table 2.The Phillips (2000) curve hasbeen adjusted assuming that wetdeposition accounts for 74% oftotal deposition. Western Ore-gon was not included in theregressions due to the largeerror associated with this datumpoint

625


in eastern Wyoming. The overall pattern is modifiedsomewhat by topography and dominant wind patterns,resulting in high 36Cl/Cl gradients along the west coast andlow gradients extending from the Gulf of Mexico in theeast. An important continental source of chloride, associ-ated with the Bonneville Salt Flats, is apparent in northernUtah where a large depression in 36Cl/Cl is observed.

The general pattern of variability observed in theempirical distribution of 36Cl/Cl was captured in themodels presented by Bentley et al. (1986), Hainsworth(1994), and Phillips (2000). This is due to the dependenceof 36Cl/Cl on stable-chloride deposition, which wasincorporated explicitly into these models through thedirect use of empirical data. The best general agreementwith the empirical distribution found in this study wasprovided by the Hainsworth (1994) model, which cap-tured most of the observed spatial variability. The modelof Bentley et al. (1986) underestimates the observed 36Cl/Cl throughout the country. The model proposed byPhillips (2000) generally overestimates the observed36Cl/Cl in the eastern part of the United States, butmatches the measured values in the Southwest.

The mean flux of 36Cl over the United States wascalculated to be 30.5€7.0 atoms m-2 s-1 (assuming that drydeposition is on average 26% of the total) and was foundto have no dependence on latitude, given the range of theinvestigation. This average continental fallout correspondsto a mean global 36Cl flux of 19.6€4.5 atoms m-2 s-1,which was calculated using the latitude dependencefor 36Cl fallout proposed by Lal and Peters (1967).This estimate is in good agreement with the flux of19 atoms m-2 s-1 calculated by Masarik and Beer (1999)from theoretical considerations, but is low compared to theflux of 30 atoms m-2 s-1 estimated by Phillips (2000) frommeasurements of 36Cl deposition obtained by variousstudies from around the globe.

The 36Cl deposition rates in dry areas, such as theSouthwest, were generally found to be lower than the meanflux for the continental USA. However, the strong linearcorrelation between 36Cl deposition and precipitation foundby Knies et al. (1994) and Phillips (2000) was not observedfor the data in this study, which are representative of alarge range of the climactic conditions throughout thecontinental United States. The lack of this linear relation-ship is likely due to spatial variability of depositionalprocesses and, in particular, atmospheric 36Cl concentra-tions over the continent. This hypothesis is supported by asignificant improvement in the coefficient of determina-tion, R2, from 0.15 with all data considered, to 0.55 withdata from the southeastern USA excluded. Since air massesin the South typically originate at low latitudes and travelnorth through the Gulf of Mexico, the atmospheric 36Clconcentrations in this region are probably low compared tothose found at similar latitudes in the rest of the UnitedStates. The diminished chloride availability in the atmo-sphere would lead to lower than expected 36Cl depositionrates. This example emphasizes the fact that a combinationof processes, at global and local scales, contributes to theremoval of 36Cl from the stratosphere and subsequent

distribution of this nuclide at the land surface. Accuratemodeling of the 36Cl/Cl distribution will require a quan-titative understanding of these processes.

The map of 36Cl/Cl presented in this study is notintended to provide a rigorous estimate of the initial 36Cl/Cl for the United States. In previous publications (Moysey1999; Davis et al. 2003), the high degree of localvariability that exists in 36Cl/Cl was discussed in detail.The results presented here should be taken as a tool toimprove our understanding of 36Cl and chloride depositionand variability. For individual aquifer studies, the initialvalue of 36Cl/Cl should be estimated locally from theavailable data whenever possible. The map provided herecan be used, however, as a general reference as long ascare is taken to note the uncertainty underlying Fig. 1. Theempirical 36Cl/Cl distribution presented in this study willbe most useful as a reference for comparison against futuremodels of 36Cl/Cl variability, for identifying significantlocal variations from the continental pattern of 36Cl/Cl,and for better understanding the processes of stratosphere–troposphere exchange and atmospheric deposition.

Acknowledgments This work was funded by the National ScienceFoundation (Grant EAR9526881) and is the product of thecooperation of more than 20 individuals from private, academic,municipal, state, and federal organizations. Many of these individ-uals also kindly provided chemical and geologic information on theregional and local systems that were sampled. Others spent manyhours in the field with us. All 36Cl analyses were conducted under thedirection of Pankaj Sharma of PRIME Lab, Purdue University.Augusta Davis assisted with the collection of most of the watersamples. Nadia Moysey and Timothy Shanahan made significantcontributions in the preparation of samples for 36Cl/Cl measurement.The discussion and analysis in this paper were significantly improvedby the thorough and insightful reviews given by Dr. F.M. Phillipsand Dr. B.E. Lehmann. We are deeply grateful to all of them.

References

Andrews JN, Edmunds WM, Smedley PL, Fontes J-C, Fifield LK,Allan GL (1994) Chlorine-36 in groundwater as a palaeocli-matic indicator: the East Midlands Triassic sandstone aquifer(UK). Earth Planet Sci Lett 122:159–171

Barry RG, Chorley RJ (1992) Atmosphere, weather and climate.Routledge, New York, 392 pp

Beasley TM, Elmore D, Kubik PW, Sharma P (1992) Chlorine-36releases from the Savannah River site nuclear fuel reprocessingfacilities. Ground Water 30(4):539–548

Beasley TM, Cecil LD, Sharma P, Kubik PW, Fehn U, Mann LJ,Gove HE (1993) Chlorine-36 in the Snake River Plain aquiferat the Idaho National Engineering Laboratory: origin andImplications. Ground Water 31(2):302–310

Bentley HW, Phillips FM, Davis SN (1986) Chlorine-36 in theterrestrial environment. In: Fritz P, Fontes JC (eds) Handbookof environmental isotope geochemistry, vol 2, ch 10. Elsevier,Amsterdam, pp 427–480

Blinov A (1988) The dependence of cosmogenic isotope productionrate on solar activity and geomagnetic field variations. In:Stephenson FR, Wolfendale AW (eds) Secular solar andgeomagnetic variations in the last 10,000 years, Proceedingsof the NATO Advanced Research Workshop (1987). KluwerAcademic, New York

Cecil LD, Vogt S (1997) Identification of bomb-produced 36Cl inmid-latitude glacial ice of North America. Nucl InstrumMethods Phys Res B123:287–289

626


Cecil LD, Pittman JR, Beasley TM, Michel RL, Kubik PW, SharmaP, Fehn U, Gove HE (1992) Water infiltration rates in theunsaturated zone at the Idaho National Engineering Laboratoryestimated from chlorine-36 and tritium profiles, and neutronlogging. In: Kharaka YK, Maest AS (eds) Water–rock inter-action. Balkema, Rotterdam, pp 709–714

Clark I, Fritz P (1997) Environmental isotopes in hydrogeology.Lewis Publishers, Boca Raton, 328 pp

Davis SN, Cecil LD, Zreda M, Sharma P (1998) Chlorine-36 andthe initial value problem. Hydrogeol J 6(1):104–114

Davis SN, Fabryka-Martin J, Wolfsberg L, Moysey S, Shaver R,Alexander C Jr, Krothe N (2000) Chlorine-36 in ground watercontaining low chloride concentrations. Ground Water38(6):912–921

Davis SN, Moysey S, Cecil LD, Zreda M (2003) Chlorine-36 ingroundwater of the United States: empirical data. Hydrogeol J11(2):217–227

Davidson CI (1989) Mechanisms of wet and dry deposition ofatmospheric contaminants to snow surfaces. In: Oeschger H,Langway CC Jr (eds) The environmental record in glaciers andice sheets. Wiley, Toronto, pp 29–51

Eriksson E (1960) The yearly circulation of chloride and sulfur innature; meteorological, geochemical and pedological implica-tions, part II. Tellus 12:63–109

Fabryka-Martin J, Davis SN, Elmore D (1987) Applications of 129Iand 36Cl in hydrology. Nucl Instrum Methods Phys ResB29:361–371

Finkel RC, Nishiizumi K, Elmore D, Ferraro RD, Gove HE (1980)36Cl in polar ice, rainwater and seawater. Geophys Res Lett7:983–986

Hainsworth L (1994) Spatial and temporal variations in chlorine-36deposition in the northern United States. PhD Thesis, Univer-sity of Maryland, USA, 215 pp

Hainsworth LJ, Mignerey AC, Helz GR, Sharma P, Kubik PW(1994) Modern chlorine-36 deposition in southern Maryland,USA. Nucl Instrum Methods Phys Res B92:345–349

Huggle D, Blinov A, Stan-Sion C, Korschinek G, Scheffel C,Massonet S, Zerle L, Beer J, Parrat Y, Gaeggeler H, Hajdas W,Nolte E (1996) Production of cosmogenic 36Cl on atmosphericargon. Planet Space Sci 44:147–151

Jiang SS, Hemmick TK, Kubik PW, Elmore D, Gove HE, Tullai-Fitzpatrick S, Hossain TZ (1990) Measurement of the 36Ar(n,p)36Cl cross section at thermal energies using the AMStechnique. Nucl Instrum Methods Phys Res B52:608–611

Keywood MD, Fifield LK, Chivas AR, Cresswell RG (1998)Fallout of chlorine-36 to the Earth’s surface in the southernhemisphere. J Geophys Res 103(D7):8281–8286

Knies DL, Elmore D, Sharma P, Vogt S, Li R, Lipschutz ME, PettyG, Ferrell J, Monaghan MC, Fritz S, Agee E (1994) Be-7, Be-10, and Cl-36 in precipitation. Nucl Instrum Methods Phys ResB92:340–344

Lal D, Peters B (1967) Cosmic ray produced radioactivity on theEarth. Handbuch der Physik, vol 46. Springer, Berlin Heidel-berg New York, pp 551–612

Lehmann BE, Davis SN, Fabryka-Martin JT (1993), Atmosphericand subsurface sources of stable and radioactive nuclides usedfor groundwater dating. Water Resour Res 29:2027–2040

Li Y-H (1992) Seasalt and pollution inputs over the continentalUnited States. Water Air Soil Pollut 64:561–573

Linsley RK Jr, Kohler MA, Paulhus JLH (1975) Hydrogeology forengineers. McGraw-Hill, New York, 88 pp

Lopes TJ, Hoffmann JP (1997) Geochemical analyses of ground-water ages, recharge rates, and hydraulic conductivity of the NAquifer, Black Mesa Area, Arizona. US Geol Survey Water-Resources Investigations Report 96-4190, 42 pp

Masarik J, Beer J (1999) Simulation of particle fluxes andcosmogenic nuclide production in the Earth’s atmosphere. JGeophys Res D104(D10):12099–12111

Moysey S (1999) Meteoric 36Cl in the contiguous United States.MSc Thesis, University of Arizona, Tucson, Arizona, USA,163 pp

National Atmospheric Deposition Program (NADP)/NationalTrends Network (2000) NADP Program Office, Illinois StateWater Survey, 2204 Griffith Dr., Champaign, IL 61820

Nishiizumi K, Arnold JR, Elmore D, Ferraro RD, Gove HE, FinkelRC, Beukens RP, Chang KH, Kilus LR (1979) Measurementsof 36Cl in Antarctic meteorites and Antarctic ice using a Van DeGraaff accelerator. Earth Planet Sci Lett 45:285–292

Nishiizumi K, Arnold JR, Elmore D, Ma X, Newman D, Gove HE(1983) 36Cl and 53Mn in Antarctic meteorites and 10Be-36Cldating of Antarctic ice. Earth Planet Sci Lett 62:407–417

Norris AE, Wolfsberg W, Gifford SK, Bentley HW, Elmore D(1987) Infiltration at Yucca Mountain, Nevada, traced by Cl-36. Nucl Instrum Methods Phys Res B29:376–379

Oeschger H, Houtermans J, Loosli H, Wahlen M (1969) Theconstancy of radiation from isotope studies in meteorites and onthe Earth. In: Olssun IU (ed) Radiocarbon variations andabsolute chronology, 12th Nobel Symposium, Wiley-Intersci-ence, New York, pp 471–496

Phillips FM (2000) Chlorine-36. In: Cook PG, Herczeg AL (eds)Environmental tracers in subsurface hydrology. Kluwer Aca-demic, Boston, pp 299–348

Phillips F, Mattick J, Duval T, Elmore D, Kubik P (1988) Chlorine36 and tritium from nuclear weapons fallout as tracers for long-term liquid and vapour movement in desert soils. Water ResourRes 24:1877–1891

Phillips FM, Rogers DB, Dreiss SJ, Jannik NO, Elmore D (1995)Chlorine 36 in Great Basin waters: revisited. Water Resour Res31:3195–3204

Plummer MA, Phillips FM, Fabryka-Martin J, Turin HJ, WigandPE, Sharma P (1997) Chlorine-36 in fossil rat urine: an archiveof cosmogenic nuclide deposition during the past 40,000 years.Science 277:538–540

Purdy CB, Mignerey AC, Helz GR, Drummond DD, Kubik PW,Elmore D, Hemmick T (1987) Cl-36: A tracer in groundwaterin the Aquia Formation of southern Maryland. Nucl InstrumMethods Phys Res B29:372–375

Reiter E (1975) Stratospheric–tropospheric exchange processes.Rev Geophys Space Phys 13:459–474

Scanlon BR, Kubik PW, Sharma P, Richter BC, Gove HE (1990)Bomb chlorine-36 analysis in the characterization of unsatur-ated flow at a proposed radioactive waste disposal facility,Chihuahuan Desert, Texas. Nucl Instrum Methods Phys ResB52:489–492

Seinfeld JH, Pandis SN (1998) Atmospheric chemistry and physics.Wiley, Toronto, 1326 pp

Semonin RG, Bowersox VC (1982) Characterization of the inorganicchemistry of the precipitation of North America. In: PruppacherHR, Semonin RG, Slinn WGN (eds) Precipitation scavenging,dry deposition, and resuspension, vol 2: dry deposition andresuspension. Elsevier, Santa Monica, Calif., pp 191–200

Simpson HJ, Herczeg AL (1994) Delivery of marine chloride inprecipitation and removal by rivers in the Murray-DarlingBasin, Australia. J Hydrol 154:323–350

Slinn WGN, Radke LF, Katen PC (1982) Inland transport, mixingand dry deposition of sea-salt particles. In: Pruppacher HR,Semonin RG, Slinn WGN (eds) Precipitation scavenging, drydeposition and resuspension, vol 1: precipitation scavenging,Elsevier, Santa Monica, Calif., pp 1037–1046

Sterling JM (2000) Spatial distribution of chloride and 36Cldeposition in the coterminous United States. MSc Thesis,New Mexico Institute of Mining and Technology, USA, 155 pp

Taylor JR (1982) An introduction to error analysis. UniversityScience Books, Mill Valley, CA, USA, 270 pp

Vogt S, Elmore D, Fritz SJ (1994) Cl-36 in shallow, perchedaquifers from central Indiana. Nucl Instrum Methods Phys ResB92:398–403

Vourvopoulos G, Brahana JV, Nolte E, Korschinek G, Priller A,Dockhorn B (1990) 36Cl measurements and the hydrology of anacid injection site. Nucl Instrum Methods Phys Res B52:451–454

Wood WW, Sanford WE (1995) Eolian transport, saline lakebasins, and groundwater solutes. Water Resour Res31(12):3121–3129

627


Documents

The distribution of meteoric 36Cl/Cl in the United States ...quebec.hwr.arizona.edu/research/moysey03-36cl.pdf · donnes du sud-est des tats-Unis, dont on pense qu’elles prsentent