Mercury and methylmercury concentrations and loads in the ... · is Harley Gulch (Fig. 1), which drains an aban-doned mercury mine complex (the Turkey Run and Abbott mines ). Another

Science of the Total Environment 327(2004) 215–237

0048-9697/04/$ - see front matter� 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2004.01.013

Mercury and methylmercury concentrations and loads in the CacheCreek watershed, California

Joseph L. Domagalski *, Charles N. Alpers , Darell G. Slotton , Thomas H. Suchanek ,a, a b c

Shaun M. Ayersb

US Geological Survey, Placer Hall, 6000 J Street, Sacramento, CA 95819-6129, USAa

Department of Environmental Science and Policy, University of California, One Shields Avenue, Davis, CA 95616, USAb

US Fish and Wildlife Service, 2800 Cottage Way, Sacramento, CA 95825, USAc

Received 16 September 2003; accepted 13 January 2004

Abstract

Concentrations and loads of total mercury and methylmercury were measured in streams draining abandonedmercury mines and in the proximity of geothermal discharge in the Cache Creek watershed of California during a17-month period from January 2000 through May 2001. Rainfall and runoff were lower than long-term averagesduring the study period. The greatest loading of mercury and methylmercury from upstream sources to downstreamreceiving waters, such as San Francisco Bay, generally occurred during or after winter rainfall events. During thestudy period, loads of mercury and methylmercury from geothermal sources tended to be greater than those fromabandoned mining areas, a pattern attributable to the lack of large precipitation events capable of mobilizingsignificant amounts of either mercury-laden sediment or dissolved mercury and methylmercury from mine waste.Streambed sediments of Cache Creek are a significant source of mercury and methylmercury to downstream receivingbodies of water. Much of the mercury in these sediments is the result of deposition over the last 100–150 years byeither storm-water runoff, from abandoned mines, or continuous discharges from geothermal areas. Several geochemicalconstituents were useful as natural tracers for mining and geothermal areas, including the aqueous concentrations ofboron, chloride, lithium and sulfate, and the stable isotopes of hydrogen and oxygen in water. Stable isotopes ofwater in areas draining geothermal discharges showed a distinct trend toward enrichment of O compared with18

meteoric waters, whereas much of the runoff from abandoned mines indicated a stable isotopic pattern more consistentwith local meteoric water.� 2004 Elsevier B.V. All rights reserved.

Keywords: Mercury; Methylmercury; Abandoned mines; Water quality

1. Introduction

The Cache Creek watershed(Fig. 1) is animportant source of total inorganic mercury to

*Corresponding author. Tel.:q1-916-278-3077.E-mail address: [email protected](J.L. Domagalski).

downstream areas including the San Francisco Bay,and the region known as the Delta of the Sacra-mento and San Joaquin rivers(Domagalski, 1998,2001; Domagalski and Dileanis, 2000; Foe andCroyle, 1999). Although the Cache Creek drainagebasin covers only approximately 4% of the areadrained by the Sacramento River(Fig. 1), the

SDMS DOCID# 1106040

216 J.L. Domagalski et al. / Science of the Total Environment 327 (2004) 215–237

Fig. 1. Map of study area and location of sampling sites.

217J.L. Domagalski et al. / Science of the Total Environment 327 (2004) 215–237

amount of total mercury transported downstreamcan be as high as 50% of the total annual load ofthe Sacramento River(Foe and Croyle, 1999).Sources of mercury within the Cache Creek water-shed include natural geothermal springs, aban-doned and inactive mercury mines, and mercuryprospects(Fig. 1). Sulphur Creek(Fig. 1) hasseveral active geothermal springs within itsdrainage.Mercury was mined at several locations in the

Cache Creek basin. The mercury–sulfide mineral,cinnabar(HgS), occurs in deposits near the ClearLake area(Fig. 1) and other locations within theCache and Putah creek drainages. The mercurydeposits in the region near Clear Lake are hydro-thermal in origin, of Cenozoic Age, and are themost northern part of a group of similar depositsassociated with volcanism and the migration of atransform fault system within the central portionof the California Coast Ranges(Rytuba, 1996).Production of mercury in California was especiallyhigh during the period after 1850 following thediscovery of placer gold deposits in the CaliforniaSierra Nevada. Peak production of mercuryoccurred in 1877 when the various mines of theCalifornia Coast Ranges produced approximately2776 metric tons of elemental mercury that wereprimarily used in gold recovery operations(Brad-ley, 1918). Residues from the abandoned mercurymines remain a source of total mercury to CacheCreek and downstream receiving bodies of water(Foe and Croyle, 1999; Domagalski, 1998). Min-ing wastes can enter streams primarily throughrunoff associated with rain, and the highestobserved concentrations and loads of total mercuryin Cache Creek have followed rainfall events(Domagalski, 1998, 2001). Mercury from geother-mal sources enters the creeks year-round.The mercury transported from the Cache Creek

basin to receiving waters may pose a human healthproblem if that mercury enters the aquatic foodweb and eventually bioaccumulates as methylmer-cury in fish to levels above health guidelines.Currently, the potential for the mercury of theCache Creek basin to change to the methylatedform, either within the Cache Creek basin or whentransported to a downstream receiving body ofwater such as the Delta, is largely unknown. Most

of the mercury transported through Cache Creekis presumably in the form of cinnabar as a sus-pended solid. The cinnabar must dissolve to lib-erate aqueous Hg(II) before the mercury can betransformed to methylmercury. No studies haveyet documented the fraction of mercury present asmethylmercury at various locations within theCache Creek watershed.The present study was designed to determine

local sources of total mercury and methylmercurywithin the Cache Creek watershed and to deter-mine the mass loadings from those sources. It washypothesized that the greatest loadings wouldoccur from previously mined regions where littleor no removal of mine wastes have occurred, andthat loads from geothermal sources would bestable, but minor compared to those of minewastes. Furthermore, it was hypothesized that sta-ble isotopes of hydrogen and oxygen in water, orother geochemical tracers such as born and lithiumcould be used to help relate mercury concentrationsor loads in water to source areas. The studydetermined the concentrations and loadings duringa time frame from January 2000 through May2001 and assessed the seasonal effects on concen-trations and loadings. The study is part of a largerinvestigation on mercury and its effects both withinthe Cache Creek watershed and downstream in theDelta of the Sacramento and San Joaquin rivers.The larger investigation is examining mercurybioaccumulation, the potential of mine remediationwithin the Cache Creek watershed to reduce mer-cury loads, and mercury bioaccumulation-relatedissues associated with ecosystem restoration pro-jects within the Delta of the Sacramento and SanJoaquin rivers.

2. Study area and methods

The Cache Creek Basin occupies approximately3000 km in Northern California(Fig. 1). The2

area upstream of Rumsey is characterized by thelow hills of the California Coast Ranges whereasthe area downstream of Rumsey flattens out tobecome part of the Sacramento Valley. Land coverin the upstream portion of the basin is mainlyforest and grazing land with minor amounts oforchards and cropland. The amount of land used


Table 1Site names and numbers from Fig. 1

Site number Site name

1 Cache Creek near Lower Lake2 Bear Creek above Sulphur Creek3 Bear Creek above Holsten Chimney Canyon4 Sulphur Creek at Wilbur Springs5 Harley Gulch near Wilbur Springs6 North Fork Cache Creek near Clearlake Oaks7 North Fork Cache Creek at Highway 208 Davis Creek Reservoir at dam, near Knoxville9 Cache Creek at Rumsey10 Cache Creek near Highway 50511 Cache Creek into Settling Basin12 Cache Creek out of Settling Basin13 Yolo Bypass at Interstate 80 near West Sacramento14 Lower Yolo Bypass

for crops increases downstream of Rumsey. Formermine sites represent a relatively small amount ofthe total land cover. Cache Creek has its origin asoutflow from Clear Lake. The largest tributary toCache Creek, the North Fork of Cache Creek, hasits origin in the northern part of the basin andincludes the Indian Valley Reservoir. Anothermajor tributary to Cache Creek is Bear Creek. TheBear Creek watershed does not have a reservoir.Both Clear Lake and Indian Valley Reservoir aremanaged to supply irrigation water to farmers inthe lower parts of the Cache Creek basin. In fact,summertime flows in Cache Creek and the NorthFork of Cache Creek are entirely managed forirrigation usage, and essentially no water reachesthe Sacramento River during the summer and fallmonths(Domagalski et al., 2000a). Fall and earlywintertime flows tend to be low because releasesfrom Clear Lake and Indian Valley Reservoir arevery low. Higher flows occur in winter in responseto seasonal rainfall. In addition, the level of ClearLake or Indian Valley Reservoir is occasionallylowered during the winter months for flood protec-tion. The rainy season occurs from Novemberthrough March, although each year is variable withrespect to timing and amount of rainfall. The waterfrom Cache Creek that leaves the basin enters aseasonal flood-control channel of the SacramentoRiver system known as the Yolo Bypass(Fig. 1),which is designed to reduce the potential of flood-

ing in adjacent areas and cities. The Yolo Bypassdischarges very little water during the dry season.There are numerous smaller tributaries to Cache

Creek; some draining abandoned mining sites andgeothermal areas. One of these smaller tributariesis Harley Gulch(Fig. 1), which drains an aban-doned mercury mine complex(the Turkey Runand Abbott mines). Another is Davis Creek(Fig.1), which drains the Reed Mine. The SulphurCreek (Fig. 1) drainage includes both naturalsources of mercury from geothermal springs andsome mine wastes, including both mercury andgold mines. Sulphur Creek drains into Bear Creek,a tributary to Cache Creek above Rumsey.Sampling sites(Fig. 1 and Table 1) were select-

ed to assess representative locations of potentialsources of mercury within the Cache Creek water-shed. Stream sites immediately downstream of thedams on both Clear Lake and Indian Valley Res-ervoir were sampled to determine mercury andmethylmercury concentrations from either the lakeor reservoir. Sampling sites were situated on smalltributaries or other water bodies near mercurymines or natural mercury sources. There were twosites on Bear Creek which included an upper site,Bear Creek above Sulphur Creek(Fig. 1, site 2);and a lower site, Bear Creek above Holsten Chim-ney Canyon(Fig. 1, site 3). There was one siteon Sulphur Creek(Fig. 1, site 4), one on HarleyGulch (Fig. 1, site 5), and one on Davis Creek


Reservoir at its spillway(Fig. 1, site 8). The upperBear Creek site is situated above the mercurymines and geothermal springs. Additional sitessituated on Cache Creek included a site at Rumsey(Fig. 1, site 9), which is centrally located in theCache Creek Basin, and sites just upstream of thepoint where Cache Creek discharges to the YoloBypass(Fig. 1, sites 10, 11 and 12); these lattertwo sites bracket the Cache Creek Settling Basin,which is designed to trap sediment transported outof the Cache Creek watershed. Finally, two siteswere situated in the Yolo Bypass(Fig. 1, sites 13and 14): one in the central portion of the YoloBypass(Yolo Bypass at Interstate 80 near WestSacramento) and the second site(Lower YoloBypass) just upstream of where the Yolo Bypassdischarges into the Delta region.In addition to the sites listed in Table 1, two

mine-site areas in the Cache Creek watershed weresampled in more detail by a separate team fromthe University of California, Davis(UCD). Thelocation of several sampling sites in both theAbbott–Turkey Run mine-site area and the SulphurCreek mine-site area are described in detail bySuchanek et al.(2002).The US Geological Survey(USGS) sampled

some sites, including the Yolo Bypass sites andthe site immediately downstream of the dam onIndian Valley Reservoir, and the UCD team sam-pled some sites, such as the Bear Creek aboveSulphur Creek and Cache Creek near Highway505. Other sites were sampled by both the USGSand UCD.Water samples collected by the USGS were

width- and depth-integrated and were collectedwith a sampler designed for the collection ofisokinetic samples(Shelton, 1994; Edwards andGlysson, 1988; Ward and Hair, 1990; US Geolog-ical Survey, 1999). The water samples were col-lected in 3-l Teflon bottles that had been cleanedfor the purpose of collecting water samples formercury and trace metals. The Teflon bottles wereoriginally cleaned by immersion in 10% hydroch-loric acid at 658 C for a period of 3 days. Afterthorough rinsing with ultra-clean water, the bottleswere tightly capped and double-wrapped in plasticfor transport to a field site. After collection of awater sample, the bottles were rinsed with ultra-

clean water and then field-cleaned with a dilutedetergent, followed by thorough rinsing with ultra-clean water, a rinse with 5% hydrochloric acid,and a final series of rinses with ultra-clean water.One set of sampling bottles was only used forgeothermal or mercury mine sites expected to havehigher mercury concentrations; while another setwas only used for downstream sites on the largercreeks and rivers expected to have lower mercuryconcentrations. After collection, the water sampleswere composited in an 8-l Teflon-lined stainless-steel churn. The cleaning procedure for a churninvolved washing with dilute detergent followedby thorough rinsing with ultra-clean water, a thor-ough internal rinse with 5% hydrochloric acid, anda final series of rinses with ultra-clean water.Similar to the procedure with the 3-l Teflon sam-pling bottles, one churn was used only for miningor geothermal sites, and a second churn was usedonly for the sites on the larger creeks and rivers.Water samples were taken from the churn for

various analyses of whole-water(unfiltered) sam-ples such as suspended sediment concentration,mercury and methylmercury in unfiltered water,trace elements in unfiltered water, and measure-ment of pH and specific conductance. Water sam-ples were then collected for analyses of filteredsamples with a 0.45mm, high-capacity capsulefilter. The filtration order for samples was totalmercury, methylmercury, other trace metals, andfinally alkalinity.Water samples collected by UCD differed from

those taken by the USGS in that the UCD sampleswere grab samples collected in that part of theriver or stream judged to have the greatest dis-charge. The UCD group did not filter samples inthe field, but rather transported the samples byovernight courier or ground transport to the cor-responding laboratory, where samples were imme-diately filtered and preserved. Samples taken formercury and methylmercury were collected in 1 l,pre-washed glass bottles. Samples for trace metals,alkalinity and stable isotopes were taken in 4-lpolyethylene bottles that were cleaned with thesame procedure described above for the Teflon-lined churns.Water samples collected by the USGS for the

measurement of total mercury in water were ana-


lyzed according to the method of Roth(1994),which utilizes cold vapor atomic fluorescencespectrometry. Water samples collected by UCDwere also analyzed with a cold vapor atomicfluorescence methodology, and complete detailsare given in Puckett and van Buuren(2000). Themethod was based on that of the US EnvironmentalProtection Agency(US EPA, 1996). Methylmer-cury in water was measured by a similar methodfollowing distillation and ethylation of aqueoussamples(Puckett and van Buuren, 2000). Selectedtrace elements in water were analyzed by induc-tively coupled plasma mass spectroscopy(Alperset al., 2000).Full details of laboratory and field quality assur-

ance requirements are given by Puckett and vanBuuren(2000). Field-level quality assurance con-sisted of the collection of blanks and replicates.Seven field blanks were collected for total mercuryin unfiltered and filtered water during the threesampling events completed by the USGS. Theconcentrations of total mercury in unfiltered waterblanks ranged from less than detection(0.5 ngyl)to 1.2 ngyl. The median level was 0.6 ngyl. Themedian was estimated by setting the concentrationsof non-detected values to one half the detectionlimit. The concentrations of total mercury in fil-tered blank samples ranged from less than detec-tion (0.5 ngyl) to 1.2 ngyl. The median value wasless than the detection limit. Therefore, bias causedby contamination does not affect the data set, asthe levels of mercury measured in environmentalsamples are much higher than those measured inblank samples. The range in measured concentra-tions of total mercury in environmental sampleswas less than the detection limit to 3070 ngyl, and98% of all measured concentrations exceeded aconcentration of 1 ngyl.Six field blanks were collected for methylmer-

cury in unfiltered and filtered water during thethree sampling events completed by the USGS.Most of the measurements were less than thedetection limit of 0.02 ngyl. Two samples ofmethylmercury in unfiltered water had concentra-tions just slightly above the detection limit, andthe highest concentration was 0.03 ngyl. Therefore,bias caused by contamination does not affect themeasurements of methylmercury.

All samples collected for total mercury in unfil-tered and filtered water by the USGS were takenin duplicate, and each of the duplicate sampleswas analyzed in triplicate. The median relativepercent difference(RPD) for the values of totalmercury in unfiltered water samples was 3.5%,whereas that for filtered water samples was 6.4%.The higher RPD for the filtered water samplesmay be attributed to the relatively lower concen-trations of total mercury in filtered water. A totalof six replicates were collected by the USGS formethylmercury analysis. The median RPD formethylmercury in whole water samples was 8.5%,whereas that in filtered water was 4.5%.The UCD sampling team also collected field

blanks and replicates. The median concentrationof total mercury in unfiltered water blanks was0.32 ngyl. The median concentration of methylmercury in unfiltered water blanks was less thanthe detection limit of 0.02 ngyl. The medianconcentration for the field blanks taken by theUCD team and filtered at the Batelle Laboratorywas 0.072 ngyl for filtered total mercury, andbelow the detection limit of 0.024 ngyl for filteredmethylmercury. The median RPD of total mercuryin unfiltered water for the UCD sampling teamwas 8.6% and that for methylmercury in unfilteredwater was 13.3%. The RPD for methylmercury infiltered water for the samples collected by UCDwas 7.5% and that for methylmercury was 20.1%.A separate laboratory quality assurance program

was used for the analysis of oxygen and hydrogenisotope ratios in unfiltered water samples. Isotopicanalyses of oxygen and hydrogen atoms in waterwere recorded as ratios relative to Standard MeanOcean Water(SMOW-V, O’Neil, 1986). Isotoperatios of oxygen( Oy O, expressed asd O)18 16 18

and hydrogen( Hy H, expressed asdD) in water2 1

were measured with a light stable isotope ratiomass spectrometer. Oxygen isotope measurementswere done on CO after equilibration with the2

water at 258C. Hydrogen isotope measurementswere done on H after reduction of the water with2

zinc with a platinum catalyst(Bigeleisen et al.,1952; Kendall and Coplen, 1985). The laboratoryuses a calibration procedure with three uniquestandards, with each standard analyzed in duplicatewith each analytical run. In all cases, the laboratory


was able to calibrate the instruments according tothe known values of isotope ratios in the standards.The laboratory also completed 18 duplicate meas-urements ofd O and 16 duplicate measurements18

of dD during the time when the environmentalsamples from this study were analyzed. The aver-age differences in the replicates ford O were18

0.03 per mil and that fordD were 0.4 per mil.Another quality assurance check was made byincluding 13 samples of de-ionized(DI) water asblind replicates. The S.D. for 13 measurements ofthe DI water were 0.07 per mil ford O and 0.818

per mil for dD.

3. Results and discussion

The general strategy for determination of massloads was two-fold:(1) to collect samples of stormwater run-off because of the higher river flowsand the greater potential for transport of mercuryand methylmercury; and(2) to collect samples atpre-planned intervals during the dry season. Therainy season and peak river flows generally occurbetween November and March, with little or norainfall and low river discharge during the remain-der of the year. Rainfall was below normal duringthe period of this study, and discharge from CacheCreek was relatively low compared to historicalrecords. The discharge for water year 2000(1October 1999 through 30 September 2000) for theCache Creek at Yolo was only 55% of the long-term average based on data from 1903 to 2000(Anderson et al., 2001). Discharge during wateryear 2001 was even less than that of water year2000. Therefore, the results of this study reflectlow-flow conditions in these streams and rivers.Concentrations of total mercury in whole and

filtered water samples are shown in Fig. 2 for thelarger stream sites, and in Fig. 3 for the smallerstream sites. As expected, the mining and geother-mal sites had the highest concentrations, and mostof the total mercury was associated with suspendedsediment, especially for the sites on Cache Creekdownstream of Bear Creek. The fraction of totalmercury associated with particles increased down-stream from the mining sites. The median ratio forthe filtered to unfiltered total mercury for theCache Creek at Rumsey site was 0.13, while that

for the inflow to the Cache Creek Settling Basinwas 0.1, and that for the Yolo Bypass was 0.12.In contrast, the median ratio of filtered to unfilteredtotal mercury for the Sulphur Creek at WilburSprings site was 0.26, while the median ratio forthe Harley Gulch site was 0.45, and that for theBear Creek above Holsten Chimney Canyon sitewas 0.43. Therefore, soluble mercury enters thestreams near the mine or geothermal sites, but theneither precipitates or becomes associated with par-ticles as the water from those tributaries and itsassociated load of mercury, discharge into CacheCreek. Concentrations of total mercury were lowerat most of the downstream locations than at themining and geothermal sites, because of theirdistance from these sources and because of dilutionfrom the two largest sources of water, Clear Lakeand Indian Valley Reservoir, or precipitation of thedissolved mercury.Concentrations of methylmercury in whole

water and filtered water samples for selected sitesare shown in Figs. 4 and 5. The highest concentra-tions were measured in water from Sulphur Creekat Wilbur Springs, Bear Creek above HolstenChimney Canyon, and Harley Gulch near WilburSprings. Concentrations of methylmercury weregenerally higher in whole water samples than infiltered samples. The ratio of methylmercury inthe filtered water samples to that for total meth-ylmercury in unfiltered water was higher than thecorresponding ratio for total mercury. The ratio ofmethylmercury in filtered water samples to wholewater samples ranged from approximately 0.1 togreater than 1, and the median ratio ranged from0.29 to 0.78 for all sites. These ratios for methyl-mercury were considerably more variable than thecorresponding ones for total mercury, at all sites.Tables of instantaneous daily loads of total

mercury for selected sites are given for data fromFebruary and March 2000(Table 2), and February2001 (Table 3). The periods of time presented inTables 2 and 3 correspond to when samples ofstorm water runoff were collected at the sites. Forthe year 2000 sampling, the total instantaneousmercury loads were relatively low near the miningand geothermal sites relative to the downstreamsites because of the lower discharges from themining and geothermal sites. Loads of total mer-


Fig. 2. Mercury concentrations at large river sites.


Fig. 3. Mercury concentrations at small stream sites.

cury increased downstream during the February–March 2000 storm, and the downstream loadsexceeded the sum of the loads from the miningand geothermal sites. For the samples collectedduring February 2000, part of the downstreamincrease could be attributed to hydrological orwater management factors. Midway during theperiod of time that samples were being collected,on 28 February 2000, water began to be releasedfrom the Clear Lake dam to lower water levels ofClear Lake to reduce the risk of flooding tolakeside property. The discharge of Cache Creekat Rumsey and at downstream sites during sam-pling, for example, contained both storm-waterrunoff and released water from Clear Lake, where-

as discharge at the time of sampling at the miningand geothermal sites, as well as the sites on theNorth Fork of Cache Creek, contained only storm-water runoff. Therefore, the higher loads at theCache Creek at Rumsey, and downstream at thesettling basin can be logically attributed primarilyto re-suspension of previously deposited mercuryin the bottom sediments as a result of higher flowsfrom the released water. The load of total mercuryat the sites furthest downstream, i.e. the YoloBypass at Interstate 80 near West Sacramento andthe Lower Yolo Bypass resulted from combinedflows from Cache Creek and the Sacramento River.No attempt was made, nor are data available, todiscriminate between these sources.


Fig. 4. Methylmercury concentrations at large river sites.

The loads of mercury and methylmercury fromthe storm of late February and March 2001 weremuch less than those during the storm of Febru-

ary–March 2000(Tables 2 and 3). The sum ofthe loads originating from the mining and geother-mal sites, as measured on the storm of February


Fig. 5. Methylmercury concentrations at small stream sites.

2001 was close to that measured at the CacheCreek at Rumsey and downstream. During thatstorm, there was very low discharge of water fromboth Clear Lake and Indian Valley Reservoir. Asa result, the mercury loads measured downstreamat Cache Creek were more representative of theloads from the mining and geothermal sites.Patterns of instantaneous loads of total methyl-

mercury for individual sites were generally similarto those for total mercury. The upstream loads oftotal methylmercury during the February–March2000 sampling were less than the loads at thedownstream sites. During the February 2001 sam-pling, however, the highest load of total methyl-

mercury occurred at the Cache Creek at Rumseysite, with two upper-watershed streams, Bear Creekand the North Fork of Cache Creek, contributingmost to the load at Rumsey.It was not possible to calculate accurate annual

discharge or annual loads of mercury from all ofthe sampling sites, or at all times at individualsites. Continuous records of discharge were onlyavailable for Sulphur Creek at Wilbur Springs,Bear Creek above Holsten Chimney Canyon, Har-ley Gulch near Wilbur Springs, outflow from ClearLake (Cache Creek near Lower Lake), North ForkCache Creek near Clearlake Oaks(outflow fromIndian Valley Reservoir), and the Yolo Bypass.


Table 2Mercury mass loadings during two storm events

Site name Hg load(gyday)

Storm 1a Storm 2b

Cache Creek near Lower Lake 86 0.2Bear Creek above Holsten Chimney Canyon 280 67Sulphur Creek at Wilbur Springs 51 35Harley Gulch near Wilbur Springs 5 0.009North Fork Cache Creek near Clearlake Oaks 29 0.1North Fork Cache Creek at Highway 20 129 12Davis Creek Reservoir at dam, near Knoxville 5 Not sampledCache Creek at Rumsey 213 149Cache Creek into Settling Basin 2386 134Cache Creek out of Settling Basin 1864 122Yolo Bypass at Interstate 80 Near West Sacramento 2267 119Lower Yolo Bypass 1496 128

Storm 1, Sampling occurred from 27 February 2000 through 30 March 2000.a

Storm 2, Sampling occurred from 20 February 2001 to 23 February 2001.b

Table 3Methylmercury mass loadings during two storm events

Site name Methylmercury load(gyday)

Storm 1a Storm 2b

Cache Creek near Lower Lake 0.63 0.002Bear Creek above Holsten Chimney Canyon 0.43 0.3Sulphur Creek at Wilbur Springs 0.03 0.03Harley Gulch near Wilbur Springs 0.001 0.00003North Fork Cache Creek near Clearlake Oaks 0.16 Not detectedNorth Fork Cache Creek at Highway 20 0.45 0.11Davis Creek Reservoir at Dam, near Knoxville 0.05 Not sampledCache Creek at Rumsey 0.67 1.45Cache Creek into Setting Basin 6.7 1.12Cache Creek out of Settling Basin 5.1 0.75Yolo Bypass at Interstate 80 near West Sacramento 21.8 1.22Lower Yolo Bypass 17.7 0.73

Storm 1, sampling occurred from 27 February 2000 through 30 March 2000.a

Storm 2, sampling occurred from 20 February 2001 to 23 February 2001.b

Discharge for the input to the Cache Creek SettlingBasin (Lower Cache Creek) was estimated fromthe discharge record of a nearby gauging station.Annual discharge and percentage of discharge

for the entire Cache Creek watershed and YoloBypass for water years 2000 and 2001 for the siteswith continuous records are given in Tables 4 and5.A significant relationship between concentration

and discharge is required to calculate an accurateannual load of any constituent in a river with few

observations of the concentration of that constitu-ent and a continuous record of discharge. Linearleast-squares regression of the data on streamdischarge and total mercury concentrations for themining or geothermal sites and the input to theCache Creek Settling Basin had generally poor fits(r between 0.008 and 0.14). The best relation2

between stream discharge and total mercury wasfor the Cache Creek into Settling Basin site(r s2

0.7), but the regression equation had poor predic-tive value with a positivey-intercept. This lack of


Table 4Annual discharge for selected sites for water year 2000(1 October through 30 September), and percent of Cache Creek and YoloBypass discharge

Site Annual discharge Percent of Cache Creek Percent of Yolo(m )3 Watershed discharge Bypass discharge

Sulphur Creek 2 784 930 1 0.1Bear Creek 33 270 770 12 1Harley Gulch 395 290 0.15 0.01Lower Cache Creek 268 342 150 100 8

Table 5Annual discharge for water year 2001(1 October through 30 September), and percent of Cache Creek and Yolo Bypass discharge

Site Annual discharge Percent of Cache Creek Percent of Yolo(m )3 Watershed discharge Bypass discharge

Sulphur Creek 1 834 700 2 -1Bear Creek 18 267 000 19 9Harley Gulch 5020 0.005 -0.01Lower Cache Creek 93 931 500 100 45

Fig. 6. Annual mercury loads at sites with continuous stream-flow records.

a clear relation between discharge and mercuryconcentration limited our ability to calculate annu-al loads of mercury for these streams.

A crude estimate of annual mercury loading atselect sites may be obtained with an estimatedaverage mercury concentration of the dry seasonand an average mercury concentration for the wetseason, or runoff events, combined with flow datafor sites that have reliable records of discharge.For the Clear Lake outflow, a dry-season estimatedaverage of 4 ngyl of total mercury and a wet-season estimated average of 12 ngyl of totalmercury was used. For the Indian Valley Reservoir,an estimated average of 2 ngyl of total mercuryfor the dry season and 5 ngyl during the wetseason were used. For Harley Gulch, an estimatedaverage of 169 ngyl for the dry season and 279ngyl for the wet season were used. For BearCreek, similar concentrations of 38 ngyl and 131ngyl were used, respectively. For Sulphur Creek,values of 758 and 1095 ngyl were used. For theCache Creek Settling Basin, concentrations of 1.3and 51.3 ngyl were used.Annual loads for each year of the study were

calculated for sites that have continuous stream-flow records. These annual loads are shown in Fig.6. For both water years, the loads from SulphurCreek are greater than those form either ClearLake or the Indian Valley Reservoir. Loads fromSulphur Creek are also much greater than those


Fig. 7. Total mercury in streambed sediment at selected sites.

from Harley Gulch, which is immediately down-stream of a large mercury mine. Because of thelower than normal rainfall amounts, mercury loadsfrom Harley Gulch were relatively low. Mercuryloads from Bear Creek increase slightly from thoseof Sulphur Creek, probably from re-suspension ofpreviously deposited mercury from Sulphur Creek.When the combined loads of the upstream tribu-taries are summed, the resulting combined load isless than that calculated for the most downstreamsite, the Cache Creek Settling Basin. This increasein loads between the mine and geothermal sitesand the most downstream site can be logicallyattributed to re-suspension of previously depositedmercury. Erosion of mercury from these minewaste piles has been ongoing for over 100 years.As a result, the bed sediments of Cache Creek area source of mercury to downstream water bodies.Concentrations(dry weight) of total mercury in

streambed sediments collected during the late fallof 2000 are shown on Fig. 7. As expected, con-centrations were considerably higher in the sedi-

ments from the mining or geothermal sites relativeto the sites downstream on Cache Creek or theNorth Fork of Cache Creek.Mercury mining began approximately 150 years

ago at locations in the Cache Creek basin, and thewastes from those operations remain a continuingsource to downstream locations. Because thesesites are impacted by mining activities, there is noway of determining what the concentration ofmercury in streambed sediment would be in theabsence of mining. Mercury concentrations were100 ngyg of dry sediment in streambed sedimenttaken just below Clear Lake and 87 ngyg of drysediment in the North Fork of Cache Creek nearHighway 20. It is to be expected that mercuryconcentrations in sediment from the Cache Creekat Rumsey site would be higher than those at theClear Lake site because of runoff from geothermalsources and of naturally occurring mercury inupstream soils. Although no ‘pre-mining back-ground’ concentration of mercury in sedimentcould be derived for the Cache Creek at Rumseysite, the present concentrations have probably beeninfluenced by human activities for more than 150years and almost certainly exceed the pre-mininglevels. Because of the anthropogenic influences onthe Cache Creek at Rumsey site, the streambedsediments along Cache Creek can be consideredas an additional source of mercury to downstreamareas.The aqueous chemistry at various locations

within the Cache Creek Basin can vary seasonallybecause of chemical differences among the inflow-ing waters from different sources. These variationsin chemistry, related to water source, may be usefulin relating the sources of mercury or methylmer-cury at downstream locations if the water associ-ated with these mercury sources have distinguish-able geochemical signatures. Some potentially use-ful chemical tracers include the aqueous concen-trations of chloride(Cl), sulfate(SO ), boron(B),4

and lithium(Li), the relative amounts of dissolvedor suspended mercury, and the stable isotopes ofhydrogen and oxygen in water.An example of a chemical signature for the

Cache Creek at Rumsey site is aqueous chlorideand sulfate. A 2-year profile of ClySO ratios with4

data from a previous study is shown in Fig. 8


Fig. 8. Ratio of chloride to sulfate at the Cache Creek at Rumsey site and at other selected sites within the Cache Creek basin.

(Domagalski et al., 2000b). Also, shown in thatfigure are ClySO ratios from select sites using4

data collected during the present study. The ClySO ratio is mostly dependant on whether Clear4

Lake or the Indian Valley Reservoir is the primarysource of water(typically during the irrigationseason and wet season), or whether Bear andSulphur creeks dominate the flow(typically duringlate fall to early winter). Because of the higherchloride content in the geothermal springs withinthe Sulphur Creek drainage, the ClySO ratio4

increases as the percentage of water from SulphurCreek increases. The Harley Gulch samples havea low ClySO ratio because of their higher con-4

centrations of sulfate derived from the mine waste.As a result of the higher sulfate concentrations,the ClySO ratio is relatively low in Harley Gulch4

water. Winter storm water runoff indicates a mixedsource. The water chemistry changes significantlyin the fall, with large changes in the ClySO ratio.4

As the irrigation season ends, and flows fromClear Lake and Indian Valley Reservoir aredecreased, the percentage of Sulphur Creek and

Bear Creek water in Cache Creek increases, result-ing in the large increase in the ClySO ratio.4

It was hypothesized that systematic changes inthe differences in select element ratios in water,from upstream mining and geothermal sources, todownstream locations, might be useful as tracersof which locations(abandoned mines or geother-mal springs) are important sources of mercury ormethylmercury to downstream locations. Geo-chemical tracing of mercury transported to down-stream areas to specific source areas would providea powerful tool for mine re-mediation and clean-up efforts. Erosion of geologic material at the mineor geothermal sites, for example, might differ fromthat of the surrounding geologic material at otherlocations such that runoff from the mine or geo-thermal sites would have characteristic signatureswith respect to element ratios. An analysis ofwhole-water samples was completed by computingthe ratios of the amounts of various elements inwhole water samples to the amount of aluminum.Aluminum has low solubility, but is a majorconstituent of the rock types found in the Cache


Fig. 9. Boron to aluminum mass ratios at selected sites of theCache Creek Basin and Yolo Bypass.

Creek basin. Therefore, it was reasoned that nor-malizing the element concentrations in whole-water samples to that of aluminum might provideuseful chemical signatures.Boron concentrations were found to differ

among various locations throughout the CacheCreek Basin. A plot of the ratios of boron toaluminum(ByAl) for mining and other sites with-in the Cache Creek Basin is shown in Fig. 9.Many of the mining and geothermal sites, such asthose of the Abbott and Turkey Run mines, theSulphur Creek mines, sites that are downstream ofmines, such as Harley Gulch, and sites downstreamof geothermal streams, such as Sulphur Creek,generally have higher ByAl ratios relative to othernon-mining or non-geothermal sites. The outflowfrom Clear Lake(Cache Creek at Lower Lake)and the Upper Bear Creek(Bear Creek aboveSulphur Creek) has relatively low ByAl ratios.The median values of the ByAl ratio for the

outflow from Clear Lake and the Upper BearCreek did not differ statistically, according to theMann–Whitney non-parametric test. The upperBear Creek and the lower Bear Creek(Bear Creekabove Holsten Chimney Canyon) sites are different(Ps0.0001) by the same statistical test. Waterfrom Sulphur Creek probably has the greatestimpact on the ByAl ratio at the Bear Creek aboveHolsten Chimney Canyon site. The ByAl ratio forthe Bear Creek above Holsten Chimney Canyonand the Sulphur Creek at Wilbur Springs site aresimilar and well above that of the Bear Creekabove Sulphur Creek site.The ByAl ratios for Sulphur Creek, the Sulphur

Creek mines, Bear Creek above Holsten ChimneyCanyon, Harley Gulch near Wilbur Springs, andthe Abbott–Turkey mine sites were statisticallysimilar (P)0.05). The ByAl ratio of Indian ValleyReservoir was higher than that of Clear Lake andthat had an influence on the chemistry of theCache Creek. Although relatively few water sam-ples were collected at the outlet of Indian ValleyReservoir(North Fork Cache Creek near ClearlakeOaks), the ByAl ratio appeared to be higher thanthat of Clear Lake, and this higher ratio probablyaffected the chemistry of the downstream site ofCache Creek at Rumsey. The ByAl ratios weresimilar at the Cache Creek sites near Highway 505and Rumsey, but the ratio at the Cache CreekSettling Basin was much lower and similar to thatat Clear Lake. Therefore, the mine and geothermalchemical signature of boron to aluminum wasdistinctive in the upper part of the watershed, butwas lost before Cache Creek discharges into theYolo Bypass.Although the ByAl ratio in water is similar at

the geothermal and mining sites, boron and lithiummight co-vary, and the concentrations of bothboron and lithium might be highest in geothermalwater (Goff et al., 1993a,b). A plot of boron andlithium concentrations in water from the studysites is shown in Fig. 10. There is a very goodrelation between these two elements and theregression coefficient(r ) for all sites was 0.99.2

As expected, the concentrations of boron andlithium were highest in water from the SulphurCreek at Wilbur Springs and nearby Sulphur Creekmine sites. There was considerable overlap in


Fig. 10. Plot of boron and lithium concentrations at selected sites within the Cache Creek watershed.

ranges of concentrations between Sulphur Creekand waters from Abbott and Turkey Run Mine andHarley Gulch. Because of the discharge of SulphurCreek into Bear Creek, the boron and lithiumconcentrations of Bear Creek above Holsten Chim-ney Canyon water samples partially overlappedwith those of Sulphur Creek or Sulphur Creekmine samples. The boron and lithium concentra-tions of water from Bear Creek upstream of itsconfluence with Sulphur Creek, which is abovethe mining and geothermal influence, were thelowest of all samples. As with the ClySO ratio,4

the concentrations of boron and lithium tend to below in water from Clear Lake and Indian ValleyReservoir, which partially dilute the concentrationsfrom mine waste or geothermal water in CacheCreek. Boron and lithium were elevated at the

Cache Creek at Rumsey, relative to the concentra-tions in either Clear Lake outflow or the NorthFork of Cache Creek, but it is unclear whether thesource is from geothermal or the mine sites.Similar relations were found for boron and chloride(Donnelly-Nolan et al., 1993; Goff et al., 1993a,b).The relation among boron, chloride and sulfate

is also shown with a ternary plot(Fig. 11). In thisplot, the relative amounts are shown on a molarbasis. A mixing relation of Sulphur Creek water,Bear Creek at Holsten Chimney Canyon, and insome cases, the Cache Creek at Rumsey, and theCache Creek near Highway 505, is apparent fromthe plot. The water of the Cache Creek at Rumseyhad a similar chemistry to that of Sulphur andBear Creeks, as seen by the ratio of chloride tosulfate in water, when the outflows from the Indian


Fig. 11. Ternary plot of boron, chloride and sulfate for selected sites within the Cache Creek watershed.

Valley Reservoir and Clear Lake were low. Thathappens in the fall, after the irrigation season andbefore the rainfallyrunoff season. At other timesof the year, the water of the Cache Creek atRumsey is more similar to that of either IndianValley Reservoir, or Clear Lake. The water atCache Creek had a lower relative amount of boron,or higher chloride and sulfate, relative to that ofthe Indian Valley Reservoir or Clear Lake. TheAbbott–Turkey Run mine and Harley Gulch watersamples plot along a range of chloride and sulfatelevels. It is not possible to distinguish any mixingtrend of the Abbott–Turkey Run Mine and Harley

Gulch waters with downstream sites, such as theCache Creek at Rumsey.Another useful signature is the stable isotopic

composition of water. Hydrogen and oxygen iso-tope ratios were measured in samples collectedfrom most of the sites of the present study(Fig.12). Stable isotope signatures of the geothermalwaters have also been previously reported(Goffet al., 1993a,b). Many of the water samples col-lected during this study had isotopic distributionsthat plot away from the global meteoric water line,which is based on worldwide stable isotope pat-terns (d O and dD) in rainfall. Stable isotope18


Fig. 12. Plot of stable isotopes(d deuterium andd O) for selected sites within the Cache Creek watershed.18

ratios of Oy O and Hy H in rain become18 16 2 1

progressively smaller as air masses leave the oceanand move inland and towards the poles. By defi-nition, ocean water has values ofd O and dD18

equal to 0.0. Water that plots away from the globalmeteoric water line usually indicates some type ofisotopic fractionation such as may occur duringevaporation or certain types of water–rock inter-actions(Drever, 1982).The waters from the Sulphur Creek and the

Sulphur Creek mines had the greatest deviationfrom the global meteoric water line. The samplesfrom the Sulphur Creek mine sites were mostenriched in O. The samples from the Bear Creek18

above Holsten Chimney Canyon plotted along amixing line from the Sulphur Creek waters. Thelarge deviation from the global meteoric water linewas a unique geochemical signature for the watersof this study. In contrast, the waters from theAbbott–Turkey Run mines and those from Harley

Gulch were more depleted in O and plotted18

closer to the global meteoric water line. The runofffrom the Abbott–Turkey Run mines and the waterin Harley Gulch are generally not affected bygeothermal discharge, and therefore their isotopicdistribution is more typical of rain.A second prominent feature of the isotope plot

shown in Fig. 12 is the regression line for theClear Lake outflow. The isotopic signature fromthat site was similar to those for Cache Creek atRumsey, Cache Creek near Highway 505, and theCache Creek Settling Basin. The Clear Lake mix-ing line is indicative of the isotopic compositionof Clear Lake water, which resulted from the long-term evolution of lake water caused by evaporationand local geothermal input over geologic time.The water that is most depleted in the heavierisotopes is that of the Yolo Bypass. That waterplots on or just below the global meteoric line.Much of the water in the Bypass was from the


Sacramento River, which is depleted in the heavierisotopes, and also plots on the global meteoricwater line (Domagalski et al., 2001). Therefore,the isotopic patterns of the geothermal waters arevery distinct in the small streams in the upper partof the Cache Creek Basin, but the signature ofClear Lake water dominates at locations on CacheCreek downstream of the mining and geothermalsites.Plots of chemical constituents and stable iso-

topes of water molecules can be used to showmixing relations and to evaluate whether or notconstituent transport is conservative. Plots of lith-ium vs. d O, and total mercury vs.d O are18 18

shown in Fig. 13, for sites along a flow path fromSulphur Creek through Bear Creek. The watersamples from Sulphur Creek with the highestenrichment in the heavier isotope of oxygen werethe result of mine drainage or a higher percentageof geothermal discharge into Sulphur Creek. Ele-vated concentrations of both lithium and mercuryin Sulphur Creek water plotted across a range ofoxygen isotope signature. Dissolved lithium showsa continuous mixing line from the Sulphur Creekmines to the Sulphur Creek near Wilbur Springswaters, for the samples that are most enriched inO (d O)y2). In contrast, for mercury, there18 18

is no suggestion of a continuous mixing line fromthe Sulphur Creek mines to the water of SulphurCreek near Wilbur Springs. Mercury in the minewaters probably sorbs to sediment particles andprecipitates to the streambed. Lithium is probablytransported more or less conservatively in thesewaters because it is dissolved, and does not pre-cipitate as a mineral along this flow path orbecome absorbed to other sediment particles. Thisnon-conservative transport of mercury limits ourabilities to trace mercury deposited in downstreamareas to source areas using geochemical or isotopictracers.

4. Summary and conclusions

A 17-month study of mercury and methylmer-cury concentrations and loads was completed inthe Cache Creek watershed. Tributaries to CacheCreek located downstream of abandoned mercurymines and in proximity of geothermal discharges

were sampled for mercury and methylmercury andother aqueous constituents. Other major tributariesto Cache Creek and Cache Creek itself were alsosampled in several locations, as was the YoloBypass, which receives water from Cache Creekand the Sacramento River during flood conditions.The study period was one of relatively low streamdischarge in this watershed compared with histor-ical records, because of relatively low rainfall.Consequently, observed loads of mercury andmethylmercury were probably less than that occurduring years of normal or above-normal precipi-tation. Geothermal springs were the source of thehighest loads of mercury during this study. Thiswas attributed to the lower than normal rainfall,which failed to produce large run-off events capa-ble of eroding mine wastes. The largest instanta-neous loads of mercury and methylmercuryoccurred in the winter months in response torainfall-induced runoff. Loads of mercury andmethylmercury were generally low in the summermonths because of low stream discharge. Releaseof water from either Clear Lake or Indian ValleyReservoir, for the purpose of either flood controlor to supply irrigation water to downstream farms,may also increase the loads of mercury and meth-ylmercury by re-suspension of previously deposit-ed streambed sediment containing elevatedamounts of mercury or methylmercury. Althoughthe loads of mercury and methylmercury can below during the dry season, concentrations can beelevated at other times of the year. This wasparticularly true for methylmercury, which tendsto have elevated concentrations in April–May,July–August, and January.Water from the geothermal and mining locations

had relatively unique geochemical signatures,especially for stable isotopes of water and otheraqueous constituents such as boron, chloride, sul-fate and lithium. The ratio of chloride to sulfatein water samples from Cache Creek at Rumseyshows strong seasonal variations that can be attrib-uted to different sources of water in the watershed.The aqueous constituents are also useful as tracersfor geothermal sources of water and for evaluationof the extent to which mercury is transportedconservatively. Concentrations of lithium, a con-servatively transported ion, correlate well with


Fig. 13. Plots of lithium and total mercury vs.d O for selected sites within the Cache Creek watershed.18

oxygen isotopes along a mixing and dilution flowpath from Sulphur Creek through Bear Creek,indicating that all of these constituents are trans-ported conservatively. In contrast, total mercury

does not correlate well with oxygen isotopes orthe other aqueous constituents, indicating that mer-cury transport is non-conservative. It is hypothe-sized that dissolved mercury from the geothermal


sources is largely adsorbed onto fine-grained sed-iments in Sulphur Creek and lower Bear Creek, orfurther downstream in Cache Creek. Mercurytransport in the tributaries dominated by geother-mal sources is highly episodic, with much oftransport related to the re-suspension of previouslydeposited sediment. Mercury transport in tributar-ies dominated by mining sources such as HarleyGulch is also related to sediment transport mech-anisms, as the main form of mercury is hypothe-sized to be particles of mercury sulfide(cinnabarand metacinnabar).

References

Alpers CN, Taylor HE, Roth DA, Cain DJ, Ball JW, UnruhDM, Dileanis PD. Study design: field and laboratory meth-ods. In: Alpers CN, Taylor HE, Domagalski JL, editors.Metals Transport in the Sacramento River, California, 1996–1997: US Geological Survey Water-Resources InvestigationsReport 99-4286, 2000; pp. 19–38.

Anderson SW, Rockwell GL, Smithson JR, Friebel MF, Web-ster MD. Water Resources Data California Water Year 2000,vol. 4. Northern Central Valley Basins and the Great Basinfrom Honey Lake Basin to Oregon State Line, Water-DataReport CA-00-4, 2001.

Bigeleisen J, Perlman ML, Prosser HC. Conversion of hydro-genic materials to hydrogen for isotopic analysis. AnalChem 1952;24:1356–1357.

Bradley EM. Quicksilver resources of the state of California.California State Mining Bureau Bull 1918; 78.

Domagalski JL. Occurrence and transport of total mercury andmethyl mercury in the Sacramento River Basin, California.J Geochem Exploration 1998;64:277–291.

Domagalski JL. Mercury and methylmercury in water andsediment of the Sacramento River Basin, California. ApplGeochem 2001;16:1677–1691.

Domagalski JL, Dileanis PD. Water-Quality Assessment of theSacramento River Basin, California—Water Quality of FixedSites, 1996–1998: Water-Resources Investigations Report00-4247, 2000.

Domagalski JL, Knifong DL, Dileanis PD, Brown LR, MayJT, Connor V, Alpers CN. Water Quality in the SacramentoRiver Basin, California, 1994–1998: US Geological SurveyCircular 1215, 2000a.

Domagalski JL, Dileanis PD, Knifong DL, Munday CM, MayJT, Dawson BJ, Shelton JL, Alpers CN. Water-QualityAssessment of the Sacramento River Basin, California:Water-Quality, Sediment and Tissue Chemistry, and Biolog-ical Data, 1995–1998: US Geological Survey Open-FileReport 00-391, 2000b, http:yyca.water.usgs.govysac_nawqaywaterindex.html.

Domagalski JL, Zhou X, Lin C, Zhi D, Fan Lan C, Xu K, Lu¨Y, Luo Y, Liu S, Liu D, Guo Y, Tian Q, Liu J, Yu W,

Shedlock R, Knifong D. Comparative Water-quality Assess-ment of the Hai He River Basin in the People’s Republicof China and Three Similar Basins in the United States: USGeological Survey Professional Paper 1647, 2001.

Donnelly-Nolan JM, Burns MG, Goff FE, Peters EK, Thomp-son JM. The Geysers-Clear Lake Area, California: thermalwaters, mineralization, volcanism, and geothermal potential.Econ Geol 1993;88:301–316.

Drever JI. The Geochemistry of Natural Waters. EnglewoodCliffs, New Jersey: Prentice-Hall, Incorporated, 1982.

Edwards TK, Glysson DG. Field Methods for Measurementof Fluvial Sediment: US Geological Survey Open-FileReport 86-531, 1988.

Foe CG, Croyle W. Mercury concentrations and loads fromthe Sacramento River and from Cache Creek to the Sacra-mento–San Joaquin Delta Estuary: California RegionalWater Quality Control Board, Central Valley Region, 1999.

Goff F, Adams AI, Trujillo PE, Counce D, Mansfield J.Geochemistry of thermalymineral waters in the Clear LakeRegion, California, and implications for hot dry rock geo-thermal development. LA-12510 HDR, UC-251, Los Ala-mos, New Mexico: Los Alamos national laboratory, 1993a.

Goff F, Kenney BM, Adams AI, Trujillo PE, Counce D.Hydrogeochemical evaluation of conventional and hot dryrock geothermal resource potential in the Clear Lake Region,California, Geothermal Resources Council Transactions, vol.17, 1993b.

Kendall C, Coplen TB. Multisample conversion of water tohydrogen by zinc for stable isotopic determination. AnalChem 1985;57:1437–1440.

O’Neil JR. Theoretical and experimental aspects of isotopicfractionation. In: Valley JW, Taylor HP, O’Neil JR, editors.Stable Isotopes in High Temperature Geological Processes.Reviews in Mineralogy, 1986; 16: pp. 1–40.

Puckett HM, van Buuren BH. Quality assurance project planfor the CALFED Project: an assessment of ecological andhuman health impacts of mercury in the Bay–Delta water-shed. Monterey, CA: California Department of Fish andGame, 2000.

Roth DA. Ultratrace analysis of mercury and its distributionin some waters of the United States. Ph.D. Dissertation,Colorado State University, Fort Collins, Colorado, 1994.

Rytuba JJ. Cenozoic metallogeny of California. In: CoynerAR, Fahey, PL editors. Geology and Ore Deposits of theAmerican Cordillera. Geological Society of Nevada Sym-posium Proceedings, RenoySparks, Nevada, April 1995,1996; pp. 803–822.

Shelton LR. Field guide for collecting and processing stream-water samples for the National Water-Quality AssessmentProgram, US Geological Survey Open-File Report 94-455,1994.

Suchanek TH, Slotton DG, Nelson D, Ayers SM, Asher C,Weyand R, Liston A, Eagles-Smith C. Mercury Loadingand Source Bioavailability from the Upper Cache CreekMining Districts: CALFED—Cache Creek Study(Task 5A:final Report), 2002, http:yyloer.tamug.tamu.eduycalfedyFinalReports.htm, accessed 3 December 2003.


US Environmental Protection Agency. Method 1631: Mercuryin Water by Oxidation, Purge and Trap, and Cold VaporAtomic Fluorescence Spectrometry. US Environmental Pro-tection Agency, Office of Water, Draft, 821-R-96-012, 1996.

US Geological Survey. National Field Manual for the Collec-tion of Water-Quality Data, Chapter A4. Collection of Watersamples, Techniques of Water-Resources Investigations,

Book 9, Handbooks for Water-Resources Investigations,1999.

Ward JR, Hair CA. Methods for Collection and Processing ofSurface-water and Bed-material Samples for Physical andChemical Analyses: US Geological Survey Open-FileReport 90-140, 1990.

Documents

Mercury and methylmercury concentrations and loads in the ... · is Harley Gulch (Fig. 1), which drains an aban-doned mercury mine complex (the Turkey Run and Abbott mines ). Another