Oxygen isotopic and geochemical evidence for a short-lived, high-temperature hydrothermal event in...

pumalofvokanoio~ andgeohemdremh

ELSEVIER Journal of Volcanology and Geothermal Research 73 (1996) 213-244

Oxygen isotopic and geochemical evidence for a short-lived, high-temperature hydrothermal event in the Chegem caldera,

Caucasus Mountains, Russia

Carey Gazis a3 * , Hugh P. Taylor Jr. a, Ken Hon b, Andrei Tsvetkov ’ a Division of Geological and Planetary Sciences, 170-25. California Institute of Technology, Pasadena, CA 91125, USA

b U.S. Geological Survey, MS 903, Federal Center, Denver, CO 80225, USA

’ Institute of Geology of Ore Deposits, Petrography, Mineralogy and Geochemistry, Russian Academy of Sciences, Moscow 109017, Russia

Received 29 August 1995; accepted 11 February 1996

Abstract

Within the 2.8 Ma Chegem ash-flow caldera (11 X 15 km), a single cooling unit of rhyolitic to dacitic welded tuff more

than 2 km thick is exposed in deep valleys incised during recent rapid uplift of the Caucasus Mountains. The intracaldera tuff is mineralogically fresh and unaltered, and is overlain by andesite lavas and cut by a resurgent granodiorite intrusion. Major- and trace-element compositions for a 1405-m stratigraphic section of intracaldera tuff display trends of upwardly increasing Na,O, CaO, Al,O,, total Fe, MgO, TiO,, Sr and Zr and decreasing SiO,, K,O and Rb. This mafic-upward zoning (from 76.1 to 69.9% SiO,) reflects an inverted view of the upper part of the source magma chamber. Oxygen isotope studies of 35 samples from this 1405-m section define a striking profile with “normal” igneous PO values (+7.0 to + 8.5) in the lower 600 m of tuff, much lower 6180 values ( - 4.0 to + 4.3) in a 700-m zone above that and a shift to high 6’*0 values (+4.4 to + 10.9) in the upper 100 m of caldera-fill exposure. Data from two other partial stratigraphic sections indicate that these oxygen isotope systematics are probably a caldera-wide phenomenon. Quartz and feldspar phenocrysts everywhere have “normal” igneous 6180 values of about +8.5 and +7.5, respectively, whereas groundmass and glass 6180 values range from - 7.7 to + 12.3. Consequently, the 6’*0 values of coexisting feldspar, groundmass and glass form a

steep array in a plot of &eldspar vs. 6groundmass,glass. Such pronounced disequilibrium between coexisting feldspar and groundmass or glass has never before been observed on this scale. It requires a hydrothermal event involving large amounts of 10w-‘~O H,O at sufficiently high temperatures and short enough time (tens of years or less) that glass exchanges thoroughly but feldspar does not. The most likely process responsible for the I80 depletions at Chegem is a very high temperature (500-6OO”C), short-lived, vigorous meteoric-hydrothermal event that was focused within the upper 750 m of intracaldera tuff. Mass balance calculations indicate fluid fluxes of = 6 X 10m6 molcm-2s-‘. We believe that the closest historical analogue to this Chegem hydrothermal event is the situation observed in the Valley of Ten Thousand Smokes

* Corresponding author. Present address: Department of Earth Sciences, Dartmouth College, Hanover, NH, 037.55 USA. Fax: 603-646-3922. E-mail: [email protected].

0377-0273/96/$15.00 Copyright 0 1996 Elsevier Science B.V. All rights reserved. PII SO377-0273(96)00023-6

214 C. Gazis et ul./Joumal of Volcanology and Geothermal Research 73 (19%) 213-244

(Alaska, USA), where hundreds of steam fumaroles with measured temperatures as high as 645°C persisted for 10 to 15

years in the much smaller welded ash-flow tuff sheet ( = 200 m thick) produced by the 1912 Katmai eruption.

Kewvordst oxygen isotope; ash-flow tuff; water-rock interaction; hydrothermal event: fumarole; Chegem caldera; Caucasus

1. Introduction

Because they represent a massive, rapid emptying

of the upper parts of a large magma chamber, thereby producing an “upside-down” version of the original

magma system (e.g., Smith, 1979; Hildreth, 1979,

1981), ash-flow tuffs and other pyroclastic deposits

erupted from major calderas provide a “snap-shot”

of a specific stage in the evolution of silicic crustal

magma chambers. Most previous geochemical stud-

ies of such calderas have focused on outflow sheets

rather than the thick intracaldera deposits (which are

more susceptible to post-depositional alteration).

However, Fridrich et al. (1991) demonstrated that

intracaldera tuffs can provide useful information

about magma chamber evolution, even in mid-Ter-

tiary calderas. Here, we present results of a geo-

chemical study of a much younger and better pre-

served caldera located in the northern Caucasus

Mountains (Russia).

Principally because of its combination of youth,

exposure and simplicity, the Chegem caldera pre-

sents a unique research opportunity. Although it is

only 2.8 million years old (Gazis et al., 1995), more

than a 2-km-thick section of caldera fill is now

exposed as a result of recent rapid uplift and deep

erosion in the Caucasus Mountains. In addition, the

Chegem caldera has a relatively simple structural and eruptive history: the caldera fill consists of a

single cooling unit of welded tuff with sub-horizontal bedding; there is no evidence for pre-caldera

volcanism, nor has there been extensive post-caldera

hydrothermal alteration. In other words, this particular “snap-shot” of the upper part of an inverted magma chamber is exceptionally well exposed and well preserved.

Oxygen isotope studies are useful for determining both the petrogenesis of an igneous rock and the extent of its interaction with water. Whole-rock 6”O values for unaltered igneous rocks are normally between +6 and + 10 (Taylor, 1968). Interaction and

exchange with water can either lower (e.g., at high

temperatures) or raise (e.g., at low temperatures)

these 6’*0 values, depending on the degree of equilibration, the temperature of exchange, and the iso-

topic composition and amount of water. Assimilation

of country rock by a magma or exchange of isotopes with the country rock (usually mediated by an aque-

ous fluid) can also raise or lower the 6’*0 value of a

body of magma.

Previous oxygen isotope studies of ash-Bow

calderas (e.g., Lipman and Friedman, 1975; Lambert

and Epstein, 1980; Hildreth et al., 1984; Larson and

Taylor, 1986a,b; Johnson et al., 1990; Smith and Suemnicht, 1991) have documented a variety of

interactions with water. For example, Hildreth et al. (1984) observed ” 0 depletions in quartz and feldspar

from several generations of post-collapse lavas at

Yellowstone. Because the measured lx O/ I60 fractionations between coexisting quartz and feldspar in

these lavas recorded equilibration at magmatic tem-

peratures, Hildreth et al. (1984) concluded that these rocks crystallized from low- I8 0 magmas and that the

“0 depletions were caused by direct interaction

between the magma and meteoric water. However,

Taylor (1985, 1987) proposed that these IRO depletions were the result of assimilation or melting of

low-‘*0 country rocks that had previously ex-

changed oxygen with meteoric-hydrothermal fluids, and in a more recent study Hildreth et al. (1991) now also concur in this general model.

Bacon et al. (1989) present direct evidence that, prior to ash-flow eruption, there was a lowering of

6’*0 in the magma at Crater Lake, Oregon, by assimilation of 10w-‘~O country rocks. They observed partially fused low- “0 granitoid blocks that were ejected during the climactic eruption and that the low- I80 quartz, plagioclase, and glass in partially fused blocks were in high-temperature isotopic equilibrium. Lipman and Friedman (1975) in their study of the Timber Mountain caldera in the South- west Nevada caldera complex, also demonstrated

C. Gazis et al./ Journal of Volcanology and Geothermal Research 73 (1996) 213-244 215

that “‘0 depletions of very large sized batches of

ash-flow tuff magma had occurred just prior to erup-

tion. Oxygen isotopic studies of a number of other

calderas have revealed no evidence for pre-eruptive

“0 lowering. For example, the Questa, New Mex-

ico, and Grizzly Peak, Colorado, calderas (Johnson

et al., 1990), the Central Nevada caldera complex

and numerous systems in the San Juan Mountains,

Colorado (Larson and Taylor, 1986b), and the Super-

stition-superior volcanic area, Arizona (Stuckless

and O’Neil, 1973), all have erupted ash-flows and

lavas with essentially unaltered igneous isotopic sig-

natures. Others, like Lake City, Colorado (Larson

and Taylor, 1986a). Long Valley, California (Smith and Suemnicht, 19911, and Valles, New Mexico

(Lambert and Epstein, 19801, have experienced ex-

tensive post-collapse, sub-solidus “0 depletions dur-

ing hydrothermal alteration.

Chegem igneous rocks; such data are essential in

understanding the stratigraphy and eruption history of a set of ash-flow deposits. Inasmuch as most (if

not all) large-volume silicic magma chambers be-

come compositionally zoned in major and trace ele-

ments prior to eruption, the compositions of the

pyroclastic products can in principle be used to infer

their original positions within the magma chamber,

most successfully done if pumices rather than whole

rocks are studied (e.g., Lipman, 1967). The com-

monly observed mafic-upward zoning in tuff se-

quences [e.g., Long Valley, California (Hildreth,

1979) and Grizzly Peak, Colorado (Fridrich et al.,

199111 is believed to represent progressive down-

ward tapping of a compositionally layered magma

chamber in which the top was the most silicic part of

the system.

Thus, examination of oxygen isotope systematics

has identified basically three end-member types of

situations that are associated with ash-flow tuff

calderas: (1) extensive sub-solidus hydrothermal ‘* 0 exchange, particularly along ring fractures and adja-

cent resurgent intrusions; (21 essentially “dry” sys-

tems where the primary magma 6’*0 values are all

largely preserved (except where they have been lo-

cally modified by low-temperature hydration or de-

vitrification of volcanic glass); (3) lo~-‘~O magmas

and coexisting low- ‘* 0 phenocrysts; the develop-

ment of such magmas is commonly associated with

large meteoric-hydrothermal systems, which also give rise to situation (1).

2. Geologic setting

In this paper, we present oxygen isotope analyses

of 134 samples from within and around the Chegem caldera, including 38 samples collected from a 1405

m stratigraphic section of intracaldera tuff. Within

this section we identified a zone of low- “0 rocks which display extreme isotopic disequilibrium be-

tween phenocrysts and coexisting groundmass. As will be discussed below, we now believe that this

represents a fourth end-member type that should be

added to the above list. We shall also discuss why

this fourth type may be very difficult to recognize in nature, except in historical eruptions or in very well preserved caldera complexes.

The 2.8 Ma Chegem caldera (Bogatikov et al.,

1992; Lipman et al., 1993) is exposed in an 11 X 1.5

km area of the north-central Caucasus Mountains, a

cratonic collision zone in southern Russia. It is the

only known ash-flow caldera in the region, consisting of a single complex cooling unit of densely

welded tuff which ranges in composition from rhy-

olitic at the lowermost exposure to dacitic at the top.

There are no obvious erosional breaks in this unit,

but there are horizons of more poorly welded tuff

indicating that there were some short hiatuses in the chain of eruptions that created this deposit. However,

cooling joints and structures pass across these hori-

zons, indicating that the entire mass of intracaldera

tuff cooled as a unit.

The intracaldera tuff is erosionally overlain by

glacial till, which in turn is overlain by andesitic lavas, preserved in the southern and western parts of

the caldera. In the northeast quadrant of the caldera, the tuff is cut by and in fault contact with a granitic

to granodioritic porphyry intrusion that has pushed

up in piston-fashion a large block of the tuff. Out-

flow tuff remnants as large as 25 X 7 km are exposed up to 50 km to the north of the northern caldera margin.

In addition to the ‘*O/ I60 data, we also present 40Ar/ ‘9Ar total fusion ages of biotite and sanidine major- and trace-element compositions for the from the base, top and interior of the intracaldera

216 C. Gazis et al./Joumal of Volcanology and Geothermal Research 73 (1996) 213-244

tuff, and from the outflow tuff, are analytically indis-

tinguishable at 2.82 f 0.02 Ma (Gazis et al., 1995).

Sanidine from the cross-cutting porphyry intrusion

yielded a similar age (2.84 f 0.03 Ma) as did an

incrementally heated whole-rock andesite (2.82 f

0.02 Ma), indicating that the Chegem Tuff was

erupted, deposited, intruded, eroded and covered by

andesite flows in a relatively short time, probably

less than 50,000 yr (Gazis et al., 199.5).

3. Analytical methods

Unweathered samples (= 1 kg) were collected

from within and around the Chegem caldera (Fig. 1).

As the terrain is rather rugged, sampling was mainly

confined to the more accessible parts of the caldera

(i.e., along the Chegem River). Efforts were made to

sample a complete stratigraphic section of caldera

fill, as well as to collect samples from all identifiable

lithologies and near all major structural features.

The main stratigraphic section discussed in this

paper (Section 1, Fig. 1) is located on the southeast

side of the caldera near Mt. Kum Tyube, and spans a

vertical range of 1405 m of compositionally zoned

tuff, locally overlain by 310 m of andesitic lavas. As

the sampled section is traversed radially inward from

the margin of the caldera, it samples progressively

upward in the stratigraphic column. Thus, the strati-

graphically higher samples are also more distant

/

0 1 2 3 4km i ‘)c-,

/

1 r /

,

I

I

\

L

,,’

I I

\

Pliocene Deposits of the

Upper Chegem River Area:

IUB Andesite Flows

cl Glacial Deposits

q Granodiorite

Chegem Tuff:

Upper Dacite

cl Lower Rhyolite

-r_.. . Fault

(dashed where approximately

located, dotted where concealed)

---- Contact

(dashed where approxlmateiy located)

River

- Foliation in welded tuff

. Sample location

Fig. 1. Geologic map of the Chegem caldera (modified after Lipman et al., 1993). 0 = sample locations; sections 1, 2 and 3 are

stratigraphic sections which were sampled in detail, and are discussed in the text.


from the caldera margin, conceivably injecting a bias into the geochemical data. Two other partial stratigraphic sections were sampled, one on the east side of the caldera (Section 2, Fig. 1) up a side drainage of the main Dzhilgusu drainage, and one in the upper dacitic tuff horizons southwest of Kum Tyube Peak (Section 3, Fig. 1).

Whole-rock samples selected for oxygen isotope analysis were crushed in a mortar and pestle. Mineral separates were obtained by hand-picking = 50 mg of the given mineral from a 35-50 mesh fraction of the ground sample. The fluorination technique used for oxygen isotope analyses is essentially that described in Taylor and Epstein (1962). Sample gases were analyzed at Caltech with either a McKinney- Nier mass spectrometer or a Finnegan MAT 252 mass spectrometer and compared to the laboratory Rose Quartz standard. Oxygen isotope ratios are reported in the familiar &notation as per mil difference from SMOW (standard mean ocean water). &values on this scale for NBS-28 and the Caltech Rose Quartz standard are + 9.60 and + 8.45, respectively. Analytical precision is better than f 0.15%0 (la).

In preparation for major- and trace-element analysis, lithic fragments were removed from whole-rock samples, and these or pumice samples were pulver- ized to = 200 mesh in a ceramic shatterbox. The

200 $ 0 70 72 74 76 78

Wt%

powders were fused into glass disks for major-element analysis and pressed into pellets for trace-element analysis at the U.S. Geological Survey. Major- element oxides were analyzed using wavelength X- ray fluorescence spectrometry, as described by Tag- gart et al. (1987). Accuracy for most of the major elements in concentrations found in the Chegem rocks is f 1%. The trace elements (Sr, Rb, Zr) were determined by energy dispersive X-ray fluorescence with an accuracy of +5% for Rb and than 10% for Zr (Yager and Quick, major- and trace-element data tables sented in Gazis et al. (in prep.).

Sr and better 1992). These will be pre-

4. Results

4.1. Mineralogy and major- and trace-element compositions

4.1.1. Intracaldera stratigraphy The intracaldera Chegem Tuff is best character-

ized by the sampled section on the east side of Mt. Kum Tyube (Section 1, Fig. 1) that ranges from 2080 to 3485 m in elevation. Although this section spans more than 1400 m, it fails to expose the base of the intracaldera tuff, which is probably at least 2 km thick. The intracaldera ash-flow sequence con-

Fig. 2. Major-element compositions (in wt.%) of samples from Section 1 (located in Fig. 1). Fe,O; = total Fe expressed as Fe,O,. Dashed

horizontal lines represent the boundaries between zones A, B and C (described in text).

218 C. Gazis et al./Joutnal of Volcanology and Geothermal Research 73 (1996) 213-244

sists mainly of densely welded, devitrified tuff that is

zoned upward from rhyolite to dacite. Crude, mas-

sive layers are defined by flattened pumices, and

these also grade upwards from rhyolitic to dacitic.

Densely welded, vitrophyric rhyolite tuff occurs

along the margins of the caldera fill, usually low in

the stratigraphic sequence. The upper dacitic tuffs

contain at least two, and possibly three, distinct

cooling sub-units that are 30-50 m thick. Cooling

breaks are defined by 2-3 m zones of flattened

vitrophyric pumice at the base of each sub-unit.

The Chegem Tuff generally contains = 20% phe-

nocrysts, varying from predominantly quartz and

sanidine in the basal rhyolites to predominantly bi-

otite, augite and plagioclase in the upper dacitic

horizons. Corresponding to these systematic changes

in the modal phenocryst abundances, there are dis-

tinct trends in major- and trace-element composition.

Major-element compositions of the whole-rock

samples in Section 1 are shown in Fig. 2: SiO,

ranges from 76.1 wt.% at the base to 69.9 wt.% just

below the andesite flows, decreasing upward in a

generally monatonic fashion. There is a correspond-

ing upward decrease in K,O and increase in Al,O,,

Na,O, CaO, and MgO. Total Fe (as Fe,O,) also

increases upward, but in a very erratic fashion. The

Na,O content drops abruptly in the uppermost dacitic

sub-unit. Trace elements display the same overall strati-

graphic trend (Fig. 3): Sr ranges from 30 ppm (base

of section) to 275 ppm (top); Zr from 74 ppm (base) to 185 ppm (top); Rb displays a less systematic

upward decrease from 177 ppm (base) to 109 ppm

(top). There are two significant reversals in these

trends; the smaller of the two is at 2870 m (sample

CG-C2-4B) and the more pronounced reversal oc-

curs at 3225 m (CG-C2-4F). These reversals are best

seen as anomalous spikes in the trends of SiO,,

K,O, CaO, Al,O,, TiO,, Sr and Rb. Although

principally defined by the samples at 2870 and 3225

m, a sample taken at 3280 m (CG-C-44G) displays intermediate compositions lying between the normal

trend and the anomalous samples. Note, however,

that the Rb anomalies document a more gradual

change between the normal trends and the reversals.

The paucity of “reversed” points is probably a

sampling artifact; the vertical distances between

samples ( = 50 m) are roughly the size of the anoma-

lous zone. A second stratigraphic section (Section 2, Fig. 1)

spans a vertical range from 1775 to 3 124 m and also varies from high-silica rhyolite at the base to dacite

(69.8% SiO,) at the top. The base of the dacite unit

in Section 2 is at 3080 m, some 250-300 m lower

that the base of the dacitic tuffs in Section 1. How-

ever, at this locality erosion has removed the distinc-

tive cooling sub-units, and their partial vitrophyric

members, which mark the upper zone of the dacitic

tuffs in Section 1. Rb decreases slightly upward in

Section 2, whereas Sr and Zr contents increase. One sample defines a slight spike in Sr and Zr, but the

anomaly is not corroborated by the Rb data and is of

opposite sense to the anomalies observed in Section

I.

-Section 1

- D -Section 2

0 100 200 300 50 100 150 200 250 300 60 100 140 160 220

9 kvm) Rb (wm) Zr @pm)

Fig. 3. Trace-element compositions (in ppm) of samples from sections I and 2 (located in Fig. 1). Dashed horizontal lines represent the

boundaries between zones A, B and C (described in text).


The mafic-upward zoning displayed in the

Chegem Tuff probably represents downward tapping

of the upper levels of a compositionally layered

magma chamber during eruption, as has been in-

ferred for many other ash-flow tuffs (e.g., Smith and

Bailey, 1966; Smith, 1979; Hildreth, 1979, 1981;

Fridrich et al., 1991). However, note that the whole-

rock major- and trace-element compositions dis-

cussed here are bulk-tuff values that represent the

average mixtures of erupted pumice compositions.

Analyses of twenty pumices collected from

throughout the sections of intracaldera and outflow

tuff show the same range of variation (69.0-76.6

wt.% SiO,) as the whole-rock tuff samples. Pumice

fragments in the high-silica rhyolites are typically

extremely small and difficult to collect; nevertheless,

two such pumices showed a range from 76.6 to 74.5

wt.% SiO,, although part of the variation may be

error related to the small sample sizes. The low-silica

rhyolite tuffs contain the complete spectrum of

pumice compositions from 75.4 to 69.0 wt.% SiO,. In contrast, the pumices from the dacitic tuffs display

a restricted range of SiO, (69.2-72.9 wt.% SiO,).

Thus, the anomalous compositional reversals ob-

served in the low-silica rhyolites in Section 1 appear to reflect changes in the mixture of erupted pumices,

probably related to eruptions from separate vents

and/or to fluctuations in the level of tapping of the sub-caldera magma chamber.

4.1.2. Outjlow stratigraphy The outflow tuff sheet, whose remnants are ex-

posed 10 to 50 km away from the Chegem caldera,

contains only part of the sequence observed in the intracaldera section. Lower parts of the outflow sheets

are typically covered by talus and poorly exposed, and the tops of the outflow tuff remnants are invari-

ably eroded. As with the intracaldera Chegem Tuff,

phenocrysts consist of quartz, feldspar and biotite, and the outflow tuff also becomes more crystal-rich

and more biotite-rich upward. The two most com-

plete sections of outflow tuff are about 200 m thick

and correlate with the lower part of the intracaldera stratigraphic section in their major- and trace-element contents (SiO, = 73.0-76.6 wt.%; Sr = 12-161

ppm; Zr = 66-132 ppm). These ranges correspond most closely with the lower 600 m of caldera fill (zone A, described below) although some of the

anomalous spikes in the central part of Section 1 also

have comparable major- and trace-element composi-

tions.

4.1.3. Resurgent intrusion and post-collapse lavas The porphyry intrusion in the northeastern quad-

rant of the Chegem caldera contains 30-40% phe-

nocrysts of plagioclase (andesine), with lesser

amounts of resorbed quartz, sanidine, biotite, augite

and magnetite in a microcrystalline, granophyric

groundmass of quartz and feldspar. It ranges from

fine-grained granite at its borders to coarser-grained

granodiorite in its interior; this granodiorite is similar

in both major- and trace-element composition to the

dacitic tuff units (SiO, = 70.3-71.9 wt.%, Sr =

193-239 ppm and Zr = 141-188 ppm).

The post-collapse lavas which cap the intracaldera

Chegem Tuff range from andesite to mafic dacite

with 1 O-40% phenocrysts of plagioclase (andesine-

labradorite), pyroxene (mainly hypersthene), amphi-

bole, and rare biotite. Major and trace elements of the andesites indicate that they are much more mafic

than any of the Chegem ash-flow tuffs, with SiO, =

63.0-64.4 wt.%, Sr = 221-350 ppm and Zr = 149-

221 ppm.

4.2. Oxygen isotopes

Oxygen isotopic compositions for the igneous rocks of the Chegem caldera and some of the nearby

country rocks are presented in Table 1.

4.2.1. Intracaldera stratigraphy Measured whole-rock, quartz, feldspar, glass, and

groundmass 6’*0 values for the 1405 m of Section 1

(Fig. 4) define an elevation profile in which the

lower half and top have relatively high 6l*O values, whereas the whole-rock and groundmass samples of

the intermediate zone exhibit very low 6l*O values.

Quartz phenocrysts and all but two of the feldspar

phenocrysts throughout the caldera have relatively

constant St80 values of + 8.5 k 0.2 and + 7.7 + 0.4,

respectively (mean, std. dev.). In contrast, whole-rock 6”O values range from -4.0 to + 12.1, and measured 6180 values of separated groundmass and glass span an even wider range ( - 7.7 to + 10.9). The two samples of feldspar phenocrysts that are slightly I80 depleted (at El80 = +5.2 and +5.9) both come

220 C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 (19961213-244

Table I Oxygen isotope results

Sample Rock type, location and elevation 6’80 (%oo)

(if) Wr a Qtz Feld b Other ’

Stratigraphic sections: Section I-SE side qf caldera CG-C2-4S

CG-C2-4T

CG-C2-4Q,R

CG-C-44A

CG-C2-4P

CG-C-44B

CG-C-44C

CG-C-44D

CG-C2-40

CG-C-44E

CG-C2-4N

CG-C2-4M

CG-C-44F

CG-C2-4E

CG-C2-4D

CG-C2-4C

CG-C2-4B

CG-C2-4A

CG-C2-4L

CG-C2-4K

CG-C2-4K2

CG-C2-4J

CG-C2-41

CG-C2-4H

CG-C2-4G

CG-C2-4F

CG-C-44G

CG-C-44H

CG-C-441

CG-C-44J

CG-C-44K

CG-C-44L

CG-C-44M

CG-C-44N

CG-C-440

CG-C-44P

CG-C-44Q

vitrophyre, contact w/metamorphic rocks, 2080 m

vitrophyre, 2080 m

vitrophyre, 2 120 m

welded rhyolitic tuff, 2 180 m

white altered tuff, 2220 m

welded tuff, 2340 m

welded tuff, 2380 m

welded tuff, 2400 m

fresh grey tuff, 2450 m

welded tuff, 2490 m

partly altered tuff, 2550 m

fresh red-brown tuff, 2630 m

fresh red-brown tuff, 2680 m

welded tuff, 27 10 m

welded tuff, 2750 m

welded tuff w/pumices, 2780 m

welded tuff, 2870 m

welded tuff, 29 15 m

grey partly altered tuff, 2965 m



fresh grey tuff, 3 100 m

fresh grey tuff, 3 145 m

white poorly welded tuff, 3 160 m

fresh grey tuff, bio-rich, 3 165 m

fresh grey tuff, bio-rich, 3225 m



partial vitro., 3400 m


welded tuff, 3420 m


welded tuff, 3450 m

welded tuff, 3465 m

poorly welded tuff, 3485 m

andesite flow, 3490 m

andesite flow, 3775 m

8.4

8.6

8.7

7.6

6.7, 6.9

8.0

7.9

7.9

7.9

8.3

7.8

8.0

7.8

7.0

4.0

6.1

2.9

2.2

3.0

- 1.9

- 1.0

2.5

-2.1

- 3.9

-4.1

8.6(P)

8.5

8.5 7.6 8.4(GM)

8.3 7.5

7.9, 8. I 5.2

- O.S(GM)

- 0.4(GM)

7.1 5.9

8.4

- 7.7, - 7.O(GMJ

- 4.8, -7.6(P),

6SLF)

-0.3

-0.5

2.7. 3.0

4.3

7.7, 8.2

4.4, 4.6

6.5, 6.5

9.5

8.2 8.1

8.5 7.2

8.4

9.4

10.3, 10.9

7.9

7.5

8.5 7.6

4.5(G)

I .4, 2.5(G)

5.6(P)

9.3(P), 7.8(G),

9.8(GM)

8.7(GM)

12.3(GMJ, 5.6 (B)

Partial stratigraphic sections: Section 2-E side of caldera KH91-2B bio-rich tuff, local summit, 3 I24 m

KH91-3 tuff, NW of KH91-2, = 3120 m

KH9 1-4 darker tuff, 3080 m

KH9 l-9 tuff, close to fault, 3030 m

CG-C2-35 slightly altered tuff, 2900 m

CG-C2-34 tuff, 2780 m

CG-C2-38 tuff, saddle E of fault, 2500 m

3.0

I .4

2.5

1.7

7.8

8.4

6.8

C. Gazis et al./ Journal of Volcanology and Geothennal Research 73 (1996) 213-244 221

Table 1 (continued)

Sample Rock type, location and elevation 8’80 (“loo)

(#I Wr a Qtz Feld b Other L

Partial stratigraphic sections: Section 2-E side of caldera CG-C2-39 altered tuff, 2430 m 7.9

CG-C2-40 tuff, 2200 m 8.0

CG-C2-28 tuff, base of columnar outcrop, 1940 m 8.1

CG-C2-29 tuff, 90 m from contact, 1870 m 8.2

CG-C2-31A vitrophyre, contact w/limestone, 1775 m 8.9

CG-C2-30A tuff, 50 m from contact, I845 m 7.8

Partial stratigraphic section in dacitic tu$ Section 3-SW of Kum Tvubey CG-C2-67F poorly welded tuff, 3530 m 10.8

CG-C2-6lE partial vitro., 3500 m 8.1. 8.6

CG-C2-67D partial vitro., 3495 m 5.9

CG-C2-67H partial vitro.. 3425 m 5.1

CG-C2-67G partial vitro., 3415 m 8.6

CG-C2-67C welded tuff, 3390 m 4.6

CG-C2-67B welded tuff, 3340 m 6.5

CG-C2-67A welded tuff, 3300 m 6.6

Vitrophyres and glasses CG-C-20 vitrophyre. Bashil dam (south)

CG-CZ- 1 vitrophyre.same as C-20

KH91-OlBP glassy pumice, same as C-20

9OL- 109 vitrophyre, road near Bulungu

CG-C-34B glassy pumice, dacitic, SW side of caldera

KH91-llP1 glassy pumice, dacitic, SW side of caldera

CG-C2-5 1 vitrophyre, base of Kektash

CG-C2-4Q,R vitrophyre, base of SE section

CG-C2-4S vitrophyre, base of SE section

CG-C2-4T vitrophyre, base of SE section

CG-C2-3 1 A vitrophyre, base of Section 2

CG-C-42B vitrophyre, Sary-su dike

CG-C-50 vitrophyre, bottom(?) of outflow

9.7

8.8

8.8, 8.8

8.5

8.1

10.1

8.7

8.4

8.6

8.9

9.2

6.9, 6.7

Welded tuff samples CG-C-34A welded tuff, top on west side

CG-C-5 1 welded tuff, deep interior

CG-C2-24B partial vitrophyre, E of Kyugenkaya, = 3300 m

CG-C2-7OC partial vitrophyre, SW of Kum Tyube, 3445 m

CG-C2-70B partial vitrophyre, 34 15 m

CG-C2-70A welded tuff, 3370 m

9OL-114 tuff w/pum., float, Dzhilgusu

CG-C2-74 welded tuff, Hobetayeen block

CG-C2-75 welded tuff, Hobetayeen block

KH91-IOB welded tuff, saddle S. of Likarlgi

KH91-6 fault breccia, saddle near Likarlgi

KH91-5 welded tuff, between faults

CG-C2-37B brecciated tuff near granite

CG-C2-5 welded tuff, Cheg.-Jung. 2000 m CG-C2-52 welded tuff, base of Kektosh

8.0, 7.9

5.8, 6.0,

6.0

6.6

7.0

7.2

6.5

7.0

4.4

7.1

5.5

4.9

2.8

2.5

7.2

8.6

8.6 7.5(G)

8.6 8.2 7.3(G)

7.2 4.8(G)

8.5 7.7

8.5

8.6, 8.7

8.6 7.5(G)

8.7 7.5(G)

8.5 7.1

8.3. 8.5, 8.1

8.7

8.5 6.0(G)

6.8(G)

9.1(G)

8.5(G)

6.1(B)

8.6(P)

6.0(P)

222

Table 1 (continued)

C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 11996) 213-244

Sample Rock type, location and elevation 8’8 0 (%Joc)

(#) Wr ‘l

Welded tuff samples from block near intrusion CC-C-53B welded tuff, next to intrusion 6.8

CG-C-58B welded tuff, next to intrusion 1.3, 1.6,

CG-CZ 14 welded tuff, 3110 m 5.9

CC-C2-I 3 welded tuff, south of Likarilgi 7.0

KH91-8 welded tuff, contact w/granite 4.4

CG-CZ-15 welded tuff, 3330 m 2.0

KH91-1OC welded tuff, 3030 m 5.5

CG-C2-12 welded tuff, 3400 m 7.5

Section from Likarilgi westward to saddle (fault) near Mt. Kygenkaya

Qtz

8.5

Feld b Other ‘

CG-C2-10 welded tuff, Likarilgi, 3430 m 7.9

CG-C-54 welded tuff 6.5

CG-C-55 welded tuff 5.2

CC-C2-18 welded tuff 4.2

CG-C2-17A welded tuff, 3360 m -0.3

CG-C2-17B black tuff, float 3360m -4.6, -4.5


CG-C2-27 welded tuff 1.6

CG-C2-26 welded tuff, 3365 m 1.9, 1.5


CG-C2-24C welded tuff, 3200 m, saddle 1.4

CG-C2-24A welded tuff, 3205 m, saddle 1.8

outjlow tuff CG-C-47A

CG-C-47B

CO-C-50

9OL-117A

9OL-I 17B

9OL- 117c

9OL- 127C

9OL- 119P4

9OL- 12OP7

9OL-119A

CG-C-46

tuff, Lechnikaya, top of exposure

tuff, 100 m below 47A

vitrophyre, base of Teraklan

tuff, lowest-Lechnikaya

tuff, Lech., 3 m above A

partial vit., Lech., higher up

basalt, base(?) of outflow

pumice, Lechnikaya

pumice, Lechnikaya

tuff, Lechnikaya

tuff, Kaminka quarry

5.7

8.2

6.9, 6.7

5.5

4.4

6.4

7.3

5.3

Granodiorite porphyr)’ intrusion CC-C-53A granodiorite, edge of intrusion

CG-C-57 granodiorite, away from edge

CG-C-58A granodiorite, edge of intrusion

CG-C2-I9A granite, border, near inclus.

CG-C2- 19C granite, border

CG-C2-19D granite, 100 m from contact

C2-19B andesite incl., border

CG-C2- 1 I granodiorite, 2960 m (low)

0.X-56 granodiorite, 3020 m

CG-C2-37A granite, quenched near fault

4.8

5.0

3.8, 3.5

1.3

6.5

4.6

5.9

6.2

3.9

1.6

Andesite, diorite KH91-16 andesite flow, Kyugenkaya

CG-C2-2 diorite, inclusion in float

CG-C-44P andesite flow, SE section 3490 m

1.5

6.6

7.9

5.5 7.2, - 6.9(GM)

8.4 7.2 5.8,5.6(P)

8.7

8.4

8.1

8.5

85(xeno)

6.8(S)

7.8(S)

8.1(S)

8.2 7.0

8.6 6.8

8.5 7.0

8.3 7.3

8.2 6.6

7.4

7.5(G)

4.6(GM)

3.8(GM)

6.O(GM), 7.4(G)

2.7(GM)

6.O(GM)

5.9(GM)

3.5(GM)

C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 (1996) 213-244 223

Table 1 (continued)

Sample Rock type, location and elevation s’*o (%0)

(#)

Andesite, diorite CG-C-44Q CG-CZ-68D

andesite flow, SE section 3775 m

andesite, little Kum Tyube

Wr a

7.5

1.2

Qtz Feld b

7.5

Other ’

Country rock CG-C-40

CG-C-4 1

CG-C-42D

CG-C-60C

CG-C-60B

CG-C-60A

CG-C-63B

CG-C2-31B

CG-C2-32

CG-C2-33

granite, several km west of caldera

gneiss, several km west of caldera

granite, near Sary-su dike

limestone, 4 m from ring fault



limestone, far from fault

limestone, caldera margin, fault

limestone, 100 m away from fault

limestone , = 1 km from fault

8.7

8.1

8.2

29.1 613C = 3.8

29.4 6’3C = 3.9

29.9 6’3C = 3.9

28.5 6’3C = 3.7 27.8 6’3C = 2.7

30.2 6’3C = 4.0

27.4 6°C = 3.9

All 6r”O values are expressed as per mil differences from Standard Mean Ocean Water. GM = groundmass; P = pumice (whole, including

phenocrysts); B = biotite; LF = lithic fragment; 613C expressed as per mil difference from PeeDee Belemnite (PDB) standard.

a Wr = whole rock.

b S = sanidine.

’ G = glass.

from the zone of intense “0 depletion in the ground-

mass (Zone B, discussed below). Also shown in Fig.

4 are the calculated material-balance 6180 values for

the groundmass samples, based on the measured

whole-rock values, and assuming for simplicity that each rock consists of 80% groundmass and 20%

3500

z 3000

s ‘G

c

y 2.500

2000

Fig. 4. Measured oxygen isotope compositions for whole rock,

quartz, feldspar, glass/groundmass in Section 1. Also included are calculated material-balance groundmass S180 values based on

whole-rock values, assuming rock consists of 20% phenocrysts

with combined 8r*O = + 7.9 and 80% groundmass. The shaded

zone is the range of calculated 6’*0 values for magma in equilib-

rium with phenocrysts.

phenocrysts; for this model calculation, the phe-

nocrysts (quartz + feldspar) are assumed to have a

combined aI80 of $7.9 in all samples, The shaded

zone represents the range of 6”O values for a

silicate melt in equilibrium with the quartz phe-

nocrysts and most of the feldspar phenocrysts, utiliz-

ing the empirical observation (e.g., Taylor, 1968)

that granitic melts are typically 0.2 to 0.4%~ higher in “0 than their coexisting feldspars.

In order to determine whether the low- 180 zone

in Fig. 4 represents a caldera-wide, low-“0 stratigraphic horizon or simply a local anomaly on the

southeast side of the caldera, two partial stratigraphic

sections (sections 2 and 3; Fig. 1) were also studied.

The whole-rock a’*0 results for sections 2 and 3, shown in Fig. 5, indicate that this is probably a

caldera-wide phenomenon.

For purposes of description, we have divided the ash-flow tuff into three zones on the basis of

“O/ I60 and petrology. Zone A (from 1775 to 2700

m in Section 1) is a section of “normal” al80 values, in which most of the measured whole-rock 6’*0 values clearly correspond to the calculated range for magmas in equilibrium with the analyzed phenocrysts. Chemically, the rhyolite tuff within this interval shows a relatively smooth upward transition

224 C. Gazis et al/Journal of Volcanology and Geothermal Research 73 (19961213-244

3200

E 2800 .,o ;;j i w 2400

i

Zone A

Fig. 5. Comparison of whole-rock 6’“O values in three strati-

graphic sections (sections 1, 2 and 3; located in Fig. 1). Zones A, B and C are described in the text.

to more mafic compositions. The tuff in this zone is

densely welded and forms massive, columnar-jointed

cliffs. Zone B (2700-3300 ml is the low-“0 sec-

tion: whole-rock 6”O values are as low as -4.0,

and the measured groundmass for the lowest I80

sample is -7.7. The measured whole-rock a”0

values are typically much lower than the magma 6’*0 values calculated from the phenocrysts. Note

that Zone B also contains both of the anomalous

reversals in major- and trace-element compositions

SiOz (wt%) l

66 68 70 72 14 76

-5 0 5 10 15

PO q

Fig. 6. Comparison of whole-rock 6180 values and SiO, content

(wt.%) for Section I. Zones A, B and C are described in the text.

defined by samples from Section 1 (Fig. 6). The tuff

throughout this zone is densely welded except for the

lowest “0 sample, which is poorly welded. Zone C

(3300-3465 ml is the upper dacitic part of the

stratigraphic section, and it shows very large varia-

tions in whole-rock 6’*0 over a very small strati-

graphic interval, extending to high 6”O values

(+ 10.9). This zone is also visually distinct from the

lower two zones because of the partial vitrophyre

horizons (containing black glassy pumices) that define welding breaks between cooling sub-units.

4.2.2. Resurgent block

In samples collected along the ridges that traverse

the uplifted block of tuff in the northeast quadrant of

the caldera (Fig. 71, the whole-rock 6”O values tend

to be higher ( + 6 to + 8) on the northeast, closest to

the resurgent intrusion. The a’s0 values decrease to

0 to +2 to the southwest, near the fault that bounds

this block. Two anomolously ‘XO-depleted samples,

CG-C2-17A and CG-C2- 17B, were collected at a

single location in the middle of this block. Sample 17A is typical grey devitrified tuff, whereas 17B is

from a large, dark-grey, devitrified pumice block

found in float; in thin section this sample is entirely crystalline and lacks the pyroclastic texture of the

rest of the tuff in the resurgent block.

4.2.3. Vitrophyres and glassy pumices

All intracaldera vitrophyres consisting entirely of

phenocrysts and black glass have whole-rock 6”O

of +8.1 or higher (i.e., slightly enriched in ‘“0

relative to the range of calculated values for magma in equilibrium with the coexisting phenocrysts). A

single outflow vitrophyre, collected as float but considered to be from the base of the outflow sheet, has

the lowest whole-rock alsO value of the vitrophyres,

+ 6.5. Glass from glassy pumices exhibits a range of S’“0 values, from + 1.4 to + 9.1. All of these

glasses are hydrated, and contain = 3 to 4 wt.% H,O (Gazis, 1994).

4.2.4. OutjIow tuff The quartz and feldspar a”0 values of the out-

flow tuffs (Table 1; Fig. 81 are very uniform ( + 8.4 f 0.2 and + 7.5 f 0.5, respectively), similar to those from the caldera fill. Groundmass and glass separates are commonly lower in “0 than the phenocrysts. In


contrast to the major and trace elements, the whole- 4.25. Post-caldera igneous rocks rock oxygen isotopic compositions of the outflow The granodiorite intrusion displays the same kinds tuffs are most similar to those from the lower part of of oxygen isotope disequilibrium phenomena that are Zone B and the transition zone between zones B and observed in the middle section of Chegem Tuff, C. though the isotopic variations are less pronounced

b 6180 contour intervals (in per rnil)

Fig. 7. (a) Whole-rock 6”O values lettered in alongside sample localities (0) for the granodiorite porphyry intrusion and resurgent block,

both located in the northeast part of the Chegem caldera (Fig. 1). All tuff samples were collected along a set of ridges which range in

elevation from 3030 to 3430 m. Note that samples were also analyzed from the peaks of Vorlan and Likarilgi (triangles). (b) Interpretive

whole-rock ??I80 contour map of the resurgent block. Only welded tuff samples are included in this contour map.

226 C. Gazis et al. /Journal of Volcanology and Geothemal Research 73 (1996) 213-244

andesite flows

-10 -5 0 5 10

6'80 15

Fig. 8. Measured 6’*0 values for whole rock, quartz, feldspar and

groundmass in granodiorite porphyry intrusion, outflow tuffs and

andesite flows. The shaded zone represents the range of 8”O

values for magma in equilibrium with phenocrysta.

(Table 1; Fig. 8). Again, the quartz and feldspar

phenocrysts have S’*O = + 8.4 + 0.2 and +7.0 + 0.3, respectively, similar to all other Chegem sam-

ples. Groundmass S180 values range from +2.7 to

+ 6.0, and corresponding whole-rock S180 values

range from +3.7 to +7.3.

All measured andesites have whole-rock S180 between + 7.2 and + 7.9, within the range calculated

for a silicate melt that would have been in isotopic

equilibrium with the coexisting phenocrysts.

4.2.6. Country rock

The Chegem caldera is surrounded Proterozoic to

Paleozoic crystalline rocks (whole-rock S” 0 values = + 8.1 to + 8.7) and Jurassic limestones (S”O =

+ 27.4 to + 30.2).

5. Discussion

5.1. Contrast with other caldera complexes

It is useful to contrast the Chegem caldera with a

couple of other caldera complexes where oxygen isotope trends and meteoric-hydrothermal phenomena are well documented: the Yellowstone volcanic field (Wyoming) and Lake City caldera (Colorado).

The Yellowstone volcanic field has produced = 6000 km3 of rhyolite tuffs and lavas. These cover an area of = 17000 km’, and were associated with three caldera cycles, spanning a period of 2.2 million

years. Hildreth et al. (1984) observed a number of

lavas at Yellowstone in which the S’*O values of

quartz and feldspar phenocrysts were very low (down to S180 = - l), providing clear-cut evidence for re-

peated eruptions of strongly “O-depleted rhyolitic magma; these low-‘*0 magmas were found to be

particularly prominent in the cycles immediately following caldera collapse.

Unlike Yellowstone, Chegem represents a single

caldera-forming eruption with little if any record of

pre-caldera volcanism. The zone of low-‘*0 rocks at

Chegem occurs late in the ash-flow sequence itself

and not in the post-collapse lavas as in Yellowstone. Another fundamental difference between the two

systems is that in the low-“0 rocks at Chegem, it is

the groundmass that is strongly “0 depleted and far

out of equilibrium with coexisting quartz and feldspar phenocrysts, which have nearly “normal” igneous

S”O values; this observation indicates that (1) the

“O-lowering process at Chegem did not affect the magmas from which the phenocrysts crystallized,

and (2) the “O-depletion process was short-lived.

Compared to the enormous Yellowstone caldera complex, the 23 Ma Lake City caldera is only I 1 X 14

km, similar in size to the Chegem caldera. However,

unlike the Chegem caldera, the Lake City caldera

underwent an extended period of resurgence accompanied by significant sub-solidus hydrothermal activ-

ity and alteration. Larson and Taylor (1986a) found

no evidence for low-“0 magmas, but showed that permeable zones of mineralogically altered (serici-

tized and chloritized) rocks typically were found to

exhibit the greatest lRO depletions (to S’*O as low as - 2); at Lake City these effects were localized around

the central resurgent intrusion and along the caldera

ring fracture (where they are associated with several ring intrusions). In contrast, the Chegem caldera rocks are remarkably fresh mineralogically. Fig. 9 shows photomicrographs of feldspars in 180-depleted rocks from the these two calderas: the Chegem sample CG-C2-4H (Fig. 9b, S180 = - 4.0) has fresh

feldspar and biotite phenocrysts with no trace of calcite, chlorite or other hydrothermal minerals; in the Lake City sample (SC-71 ; Fig. 9a, S180 = + 0.4)

C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 (1996) 213-244 221

a

Fig. 9. Photomicrographs of low-‘*0 rocks from Lake City and Chegem calderas. (a) Sample SC-71 from Lake City caldera

(6’sO = +0.5): feldspar is altered to sericite and calcite; (h) Sample CG-C2-4H from Chegem caldera (6’*0 = - 4.0): feldspars are entirely

unaltered mineralogically.

the feldspars are turbid and altered to calcite in their

cores, and biotites are almost completely replaced by

sericite + quartz.

5.2. Caldera-wide geochemical correlations

Although sections 1 and 2 exhibit the same kinds of compositional systematics, a close comparison of

their trace elements shows that the fit between the

two sections, especially at the top and bottom, is

better when the Section 2 values are arbitrarily moved

up 300 m in elevation (Fig. lOa-cl. This offset could be due to lateral variation of silicic/mafic pumice

ratios within the caldera fill or to post-eruption verti-

cal offsets along faults or folds. If Section 2 has

indeed been displaced downward 300 m relative to Section 1, then restoring it to its original position improves the Sr and Zr correlations, but virtually destroys any correlation in oxygen isotope compositions (see below and Fig. 1Od). In fact, the fit

between the two oxygen isotope profiles is best if

Section 2 is moved down about 200 m (Fig. 10e).

Assuming that the major- and trace-element trends are magmatic, these offsets imply that the oxygen

isotope trends are not magmatic, an important obser-

vation that supports the general model that we pro-

pose below.

One could conceivably construct a multi-stage

scenario in which immediately following the ash-flow

deposition, Section 2 moved down 500 m relative to Section 1, followed by a hydrothermal event which

caused caldera-wide “0 depletions in the upper parts of the caldera-fill, especially along a horizontal

zone about 600 m thick; afterwards, further tectonic

movements such as resurgence caused Section 2 to

be displaced upward about 200 m relative to Section 1. An analogous type of late-stage collapse, followed by piston-like uplift, has been well documented at the Grizzly Peak caldera by Fridrich et al. (1991). The curved section of the Dzhilgusu River immedi-

228 C. Gazis et al./ Journal of Volcanology and Geothermal Research 73 (1996) 213-244

0 100 200 300 50 100 150 200 250 300 60 100 140 180 220

Sr (wm) Rb (wm) Zr (wm)

Fig. 10. (a, b, c) Comparison of trace-element (Sr, Rb, Zr) composition profiles in Section I and Section 2, if the values for Section 2 are

moved up in elevation by 300 m. (d) Comparison of whole-rock 6’*0 profiles for sections I and 2, if the values for Section 2 are moved up

in elevation by 300 m: (e) Comparison of whole-rock 6180 profiles for sections 1 and 2, if the values for Section 2 are moved down in

elevation by 200 m.

ately south of the resurgent block is a likely candi-

date for the inner ring fracture zone in the Chegem

caldera. If this multi-stage scenario is correct, it

implies that the hydrothermal I8 O-depletion event

must have occurred after deposition of the tuff but

prior to the resurgence of the granodiorite intrusion.

5.3. Cooling, hydration and devitrifcation

The cooling history of the Chegem Tuff, includ-

ing devitrification and hydration of the glasses, is reviewed here because it undoubtedly has significant bearing on the observed oxygen isotope profile. Quartz-feldspar oxygen isotope fractionations in the Chegem Tuff (e.g., As-r = 1.1%0, feld = An,,) correspond to typical rhyolite eruption temperatures (700-900°C; Cas and Wright, 1988). Because the Chegem tuff is welded, its emplacement temperature was also high (> 600-750°C; Cas and Wright, 1988).

After emplacement, the Chegem ash-flow tuff underwent compaction and welding, cooling slowly by a

combination of conduction, advection-convection of aqueous fluids, and radiation of heat. Riehle (1973)

developed an analytical heat-flow model for an ash-

flow tuff sheet, in which radiative heat transfer is

taken into account by modifying the coefficient of

thermal conductivity, and convective heat transfer by volatiles is assumed to be negligible. Scaling this

model to the 2-km thickness of intracaldera Chegem

Tuff (Fig. 111, one finds that it takes approximately 1300 yr to cool the top 100 m of tuff to below 3OO”C, and more than 66,000 yr to cool the entire thickness to less than 300°C. Obviously, any large- scale convection of aqueous fluids will accelerate this cooling process.

The uppermost section of intracaldera Chegem Tuff must have cooled quickly, as evidenced by the 20 m of nonwelded tuff at the top of this unit. The


TW

Fig. 11. Modeled cooling history of 2 km thick sheet of welded

tuff with initial temperature of 700°C. Modified from model by

Riehle (1973). The model assumes that cooling is mainly by

conduction.

dacitic tuff horizons also have well-preserved cooling sub-units with pristine glassy pumices at their boundaries. The margins of the intracaldera tuff mass also cooled quickly against the caldera walls, quenching to form densely welded marginal vitrophyres a few meters thick. At temperatures of 650°C a l-km-thick tuff sheet probably requires less than 1 year to produce dense welding, but this time can be reduced to less than one week if at least 0.5% H,O is present (Bierwith, 1982).

Some of the initially glassy tuffs at Chegem also underwent high-temperature devitrification. Sub- solidus crystallization of thermodynamically metastable glass produces a variety of devitrification products, depending on the cooling rates (Lofgren, 1970). Rapid cooling produces fibrous crystals of quartz and sanidine that rapidly nucleate to form spherulites and axiolites (Ross and Smith, 1961). Slower cooling rates produce more equigranular granophyric textures. For a 3-mm glass shard, experimental studies predict devitrification times of < 3 yr at 750°C to about 3000 yr at 300°C (Lofgren, 1970).

All of the analyzed glassy pumices from the intracaldera Chegem Tuff have undergone hydration, typically attaining water contents of 3-4 wt.% (Gazis, 1994). Glasses are rapidly hydrated, even at surface temperatures, when water diffuses into and through the glass structure, forming spheroidal (perlitic) cracks. Hydration rate experiments (Friedman and Smith, 1960; Friedman et al., 1966) predict that a 3-mm shard will be hydrated in < 3 h at 750°C or in

about 3 years at 300°C. These rates of hydration are 3-4 orders of magnitude faster than the rates of devitrification for the same size shard.

5.4. Significance of 10w-‘~O rhyolitic glasses in na-

ture

The presence of apparently pristine glasses with 6”O as low as + 1.4 presents an interesting dilemma. Inasmuch as glasses are essentially silicate melts quenched at sub-solidus temperatures, they should originally have exactly the same 6’*0 value as the melt phase of the magma (i.e., within the normal “magmatic” range of about +6 to + 10). In fact, in general they do not, because the low-temperature alteration and hydration that is commonly observed in older volcanic glasses in nature is invariably associated with an upward shift of the glass F”O value by several al., 1985). This i!

er mil (Taylor, 1968; Cerling et 0 shift is a result of pervasive

exchange between Earth-surface waters and such glasses over the course of several thousand years, together with the fact that the equilibrium oxygen isotope fractionations between water and silicate minerals or glass are very large at such low temperatures (e.g., at 25”C, A water_g,ass = 20%0: Taylor, 1968; Cerling et al., 1985). Thus, at ambient temperatures only extremely light meteoric waters in relatively cold climatic regions (e.g., 6180 < - 15) can sub- stantially lower the 6l*O value of a volcanic glass. On the other hand, water-glass oxygen isotope fractionations at higher temperatures (e.g., greater than 200°C) are sufficiently small that, even in temperate or subtropical climates, all meteoric water-glass interactions will lower the I80 content of the glass; however, such high temperature alteration would also be expected to devitrify and recrystallize the inherently metastable glass. For example, experiments by Lofgren (1970) and Marshall (1961) predict that at 200°C a lmm (radius) glass shard will be devitrified by pure water in = 3000 yr. At higher temperatures or with saline waters, these times will be orders of magnitude shorter.

Also, even if low- i*O glasses are formed by some process, any subsequent interaction with additional water at low temperature will in general produce a high- I80 overprint that is likely to obscure the original low-‘*0 signature. This could be the main reason

230 C. Gazis et al. / Journal of Volcanology and Geothermal Research 73 (1996) 213-244

why low- I80 glasses are only rarely observed, except in very young volcanic rocks in unique geologi-

cal environments (e.g., in Iceland, Muehlenbachs et

al., 1974). Smith and Suemnicht (1991) found two

glassy pumices at Long Valley caldera with 6”O

values of + 0.3 and + 0.4, and Hildreth et al. (1984)

found that glasses in the non-hydrated cores of some

vitrophyres at Yellowstone have S’s0 values as low

as +0.5.

5.5. Lack of hydrothermally altered country rocks

The oxygen isotopic compositions of the country

rocks surrounding the Chegem caldera, like the field

and petrographic observations of the tuff itself, argue

against the development of any substantial “normal”

(i.e., 200-400°C) hydrothermal system at this local-

ity. In particular, samples CG-C-60A to 6OC, which are limestones collected at distances of 4, 18 and 30

m from the main caldera ring fault, have normal

sedimentary 8”O values of +29.0 to + 29.9. If a

significant hydrothermal system had developed after

caldera collapse, as, for example, in the Lake City

caldera (Larson and Taylor, 1986a), one would ex-

pect that abundant fluids would have moved along this ring fault, a zone of higher permeability. Lime-

stones are sensitive indicators of the passage of

meteoric-hydrothermal fluids in the brittle-fracture regime of the Earth’s crust, because under such

conditions they exchange rapidly and they can be

very permeable; also, because they have very high

initial a”0 values, even a small degree of 180/ I60

exchange can be readily observed.

5.6. Constraints on the process responsible for ‘so

depletion

The dramatic ‘*O depletion and disequilibrium

described in this paper has not to our knowledge

been observed previously anywhere in the world, and

certainly not on this scale. It is possible that these types of “0 depletions in fact occur much more frequently in nature, but are overlooked because they are usually overprinted and obscured by the ubiqui- tous hydrothermal alteration or low-temperature hydration that is so characteristic of volcanic centers, particularly calderas. The Chegem caldera may have uniquely preserved the effects of this “0 depletion event because of its youth, and because the rapid

uplift and associated deep erosion in the Caucasus Mountains has minimized the development of any

long-lived hydrothermal stage.

Based on our data, we can place certain con-

straints on the process that caused the “0 depletion

at Chegem. In quantifying these constraints in the following discussion, assumptions must be made re-

garding temperatures of equilibration, the isotopic

compositions of meteoric waters and unaltered

Chegem Tuff, and diffusion mechanisms and coeffi-

cients for oxygen isotope exchange between H,O

and feldspar or glass. Where these parameters are

uncertain (as for temperature), a range of values is

used or the most conservative estimate is made (i.e., we utilize values that minimize the amount of water

required to produce the observed effects).

5.6.1. Involvement oj’high-temperature meteoric wa-

ters

Whatever the IsO-lowering process was, it must

have involved meteoric waters at relatively high

temperatures ( > 300°C). Earth-surface waters are the only waters isotopically light enough to cause the

observed I80 depletions, and high temperatures are

necessary to produce sufficiently favorable equilib-

rium ‘so fractionations between silicates and H,O. It is possible that some of the less extreme “0

depletions (such as those in the outflow sheets) were

caused by isotopically light waters at temperatures below 300°C and therefore we should perhaps con-

sider two-stage (or multi-stage) processes, or pro-

cesses involving a range of temperatures. In fact, if

the ‘so enrichments of up to 4%0 observed at the top

of the intracaldera tuff section were caused by mete-

oric waters (as seems likely), these must have occurred at low temperatures (< 100°C). Only at such

low temperatures are oxygen isotope fractionations between silicates and H,O large enough to raise the

S1’O values of the rocks. Also, the simple cooling

model (Fig. 11) indicates that the upper 100 m of tuff was cooled rapidly to < 100°C.

In addition, mineralogical evidence provides an important constraint on the temperatures of water- rock interaction. Feldspars and biotites are known to be rapidly altered in a “low-temperature” (20O-

400°C) hydrothermal environment, but the feldspars in the Chegem Tuff are entirely unaltered (Fig. 9a> and there is no sign of chloritization of biotites. Cole

C. Gazis et al. / Journal of Volcanology and Geothermal Research 73 (1996) 213-244 231

et al. (1992) performed hydrothermal experiments

with pure water on granites at temperatures of 170 to 3OO”C, and observed that chlorite formed in biotites

in all cases after only 160 hours. The presence of salts (NaCl + KCl) in the water increased the alter-

ation rate. Thus, the lack of “typical” hydrothermal

alteration minerals, such as chlorite and sericite,

implies that the hydrothermal “0 exchange that

produced the ‘*O depletions at Chegem had to take

place at very high temperatures ( = 500-6OO”C),

within the stability fields of the high-temperature

minerals.

The present-day Chegem River water, a combina-

tion of glacial meltwater and rain water, has a 6D value of -89 (Gazis, 1994) which corresponds to a

6”O value of - 12, assuming that it lies on the

meteoric water line (Craig, 1961). Modern precipita-

tion at the highest point in the Chegem caldera might

be expected to be several per mil lower in “0 than this river water. However, estimates of recent uplift

in the Greater Caucasus are quite high (up to 12

mm/yr; Philip et al., 1988) and it is likely that

erosion did not keep pace with uplift. Thus, the overall elevation of the central Caucasus Mountains

was lower and topography was probably less pronounced 2.8 million years ago than at present. In

such a scenario, all else being the same, one would expect that 2.8 Ma meteoric water in the Chegem

area would have had a higher Si80 value than

present-day meteoric water. In the discussion below,

we will consider a range of meteoric water isotopic compositions from 8180 = - 12 to - 16. The lower

6’*0 value will tend to minimize the amount of

meteoric water needed. The measured 8D values of

glass samples from the caldera fill are compatible

with low-temperature exchange with a meteoric wa-

ter lying within the above range of isotopic compositions (Gazis, 1994).

5.6.2. Mass-balance W/R ratios

Mass balance calculations can be used to estimate the amounts and fluxes of water necessary to cause

the observed “0 depletions at Chegem. We will

utilize the following assumptions: (1) The original magma and erupted rock both

had 6’*0 = + 8, a value representing a close approach to equilibrium with the coexisting phenocrysts.

(2) The = 7OO-m-thick ‘*O-depleted zone ob-

served in the stratigraphic section on the southeast

side of the caldera was present throughout the 11 X

15 km caldera and is equally intense throughout,

giving a flattened cylindrical volume of low-‘*0 rock equal to 1.16 X 10” m3. With a density of 2.4

g/cm3, this translates into about 2.8 X 10” g of

‘*O-depleted tuff. (3) The groundmass in each sample has ex-

changed thoroughly with (and attained isotopic equi-

librium with) the interacting meteoric waters; this

assumption minimizes the amount of water needed to

produce the observed ‘*O depletions.

(4) The groundmass 6’*0 value is estimated by

material-balance from the measured whole-rock 6’ 8 0

value, assuming that all of the samples contain 20%

phenocrysts having a bulk 6’*0 of +7.9.

(5) The oxygen isotope fractionation between wa-

ter and glass (or groundmass) is assumed to lie

halfway between the measured equilibrium fractiona-

tions for quartz-H,0 and albite-H,O (Clayton and

Kieffer, 199 1).

With the above assumptions, zero-dimensional box

models such as those described by Taylor (1977) are

useful for estimating the amounts of water required

to produce observed oxygen isotopic shifts. Material-balance water/rock (W/R) ratios can be

calculated and converted to integrated fluxes when

appropriate length scales and flow directions are

known. An advantage of such box models is that the

details of the hydrothermal “plumbing” system or

flow paths need not be known. On the other hand, a

disadvantage of such models is that the actual W/R

ratios in such systems in nature can only be calcu-

lated if the position of the rock in the flow path and

the conditions of flow and exchange are taken into

account (Norton and Taylor, 1979; Bowman et al.,

1994). Also, although the actual W/R ratios are dimensionless, they depend on the size of the rock

parcel (box) which is considered. The material-balance and actual water/rock ratios only become iden-

tical if the size of the rock parcel is increased to

essentially encompass the entire hydrothermal sys-

tem. Nevertheless, since neither the nature nor the direction of fluid flow in the Chegem caldera system

is known, whereas the size of the “O-depleted zone is well-defined, zero-dimensional models are appropriate in this case.

232 C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 (1996) 213-244

To minimize the amount of water needed to pro-

duce the observed “0 depletions, we consider an

open-system model, where a tiny parcel of water

exchanges isotopes with the rocks and then passes

out of the rock system forever. The material-balance

model water/rock ratio (i.e., the amount of water

oxygen/amount of rock oxygen) for an open system

where rocks have reached equilibrium with the inter-

acting water, integrated over the lifetime of the

hydrothermal system, is given by (Taylor, 1977):

W - = In R

[

&!I*0 + A - a,:,,

6’ -@;/A) u2o I

where i is the initial value and f is the final value

after exchange, and A = &Lck - ?jL20 is the equilibrium fractionation between water and rock at the

given temperature.

Because we do not a priori know the temperature

of exchange, we probably must consider a range of

temperatures, from 25 to 850°C. However, in apply-

ing the above expression one quickly discovers that the elevated 6180 values in the upper 40 m of the

stratigraphic section can only be produced at temper-

atures below 100°C from waters with 8L20 > - 13.

The lowest groundmass value (sample CG-C2-4H,

calculated 6 18 Ogm,= - 7) requires temperatures of at

least 350°C if 6;120 = - 12 and at least 250°C if

6’ H,O = - 16. Fig. 12 shows the calculated W/R

2000 d

0 0.5 1 1.5 2 2.5

Wk,,n

ratios vs. elevation for temperatures of 850 and

400°C (or 300°C) and for a1;2o = - 12 (or - 16).

The material-balance W/R ratios (e.g., for sample

CG-C2-4H) are reduced by as much as 45% at the

higher temperature (850°C) as opposed to the lower

temperature. Averaged over the entire I8 O-depleted

zone of the stratigraphic column, these W/R ratios

are overall about 30% lower at 850°C than at the

lower temperatures. These oxygen W/R ratios of 0.3

to 0.6 translate into water/rock ratios by weight of

about 0.16 to 0.32 (g of H,O per g of rock). This in

turn implies that about 0.5 to 1.0 X 10” g of H,O

exchanged with the = 700-m-thick section of tuff.

5.6.3. Timescale of I80 depletion at Chegem

The caldera-fill ash-flow tuff is not appreciably

mineralogically altered, and it also still contains

apparently pristine (but strongly hydrated!) volcanic

glass. As discussed earlier, the IgO depletions at

Chegem were definitely not caused by an extensive

long-lived hydrothermal system like the one which

existed at the Lake City caldera. Furthermore, the “O-lowering process at Chegem must have hap-

pened on a very short timescale because of the

extreme isotopic disequilibrium between groundmass and phenocrysts. A useful graphical mechanism for

discussing isotopic disequilibrium is provided by S-6 plots for coexisting quartz-feldspar or pyrox-

Fig. 12. Calculated open-system W/R ratios for Section 1 (Fig. 1). These were calculated using material-balance open system equation of

Taylor (1977) for waters with initial compositions of g’s0 = - 12 (a) and S”O = ~ 16 (b). The fractionation factors used are averages of

the fractionation factors for quartz and albite given in Clayton and Kieffer (1991). Two temperature extremes are shown for each water,

T= 850°C (magmatic temperature) and T= 400°C (6’*0 = - 12) or T= 300°C (6’80 = - 16). Exchange at temperatures below 350°C (6’sO = - 12) and 250°C (6180 = - 16) can not explain lowest 6”O sample. The averaged W/R ratios for the stratigraphic thickness from

2700 to 3400 m elevation are listed.

C. Gazis et al. /Journal of Volcanology and Geothemud Research 73 (1996) 213-244 233

15

a 10

2 B ‘c15 z

O

2 0

Ga

-5 /d

0 5 10 15 0 10 1s

PO quartz Sldb feldspar

Fig. 13. 6-6 plots of Chegem Tuff and granodiorite. (a) hquartz -

6 reldspar plot; symbols: 0 = intracaldera and outflow tuff; crossed

square = granodiorite porphyry intrusion. (b) i5reldlpar -

6 sroundmass,slass plot; symbols: l = intracaldera tuff (ground-

mass); 0 = intracaldera tuff (glass); crossed square = granodiorite

porphyry intrusion (groundmass).

ene-feldspar pairs (e.g., Gregory and Taylor, 1986; Gregory et al., 1989). For the Chegem caldera, the 6 quartz -6 feldspar plot (Fig. 13a) reveals only small degrees of disequilibrium in some samples, whereas the 6-6 plot for feldspar-groundmass/glass (Fig. 13b) displays extreme disequilibrium. This “mineral pair” is sensitive to disequilibrium at much shorter timescales than the Squartz-Sfeldspar mineral pair because at a given temperature, silicate glass (and very fine-grained groundmass?) exchanges oxygen isotopes with H,O much more rapidly than feldspar phenocrysts (Taylor, 1968); in turn, it has been shown in countless studies, both in nature and in the laboratory, that feldspar-H,0 exchange is much more rapid than quartz-H,0 exchange.

In order to quantify the necessary timescales, let us consider a particular Chegem sample, namely the one that displays the greatest degree of disequilibrium between phenocrysts and groundmass (CG-C2- 4H #so feldspar = + 5.9, ~‘80groundmass = -7.0). The kinetics of isotopic exchange greatly depend on the temperature, so here we will consider two temperature extremes: 300°C (typical hydrothermal) and 800°C (near magmatic). At either temperature, the isotope exchange process must have occurred over a long enough time scale that the glass or the groundmass could nearly fully equilibrate (> 90%) with the water, but yet over a short enough timescale that the feldspars are nearly unaltered ( < lo%), both isotopi- tally and mineralogically. The extent of “0 ex-

change for both feldspar phenocrysts and glass are dependent on the mechanism of exchange, either surface reactions (such as dissolution and reprecipita- tion) or diffusion of oxygen-bearing species such as H,O (Cole et al., 1983). Inasmuch as volume diffusion-controlled exchange is slower than exchange due to surface reactions by several orders of magnitude, this mechanism will provide a maximum estimate of the water-rock interaction time that conceivably could have elapsed without isotopically altering the feldspar.

Diffusion-controlled exchange is dependent on the diffusion coefficient for the phase of interest, the diffusion radius, the temperature and the mole ratio of oxygen in the water to that in the rock. Typical values of hydrothermal diffusion coefficients for feldspars at 300 and 800°C are 1.4 X lo-” and 6.4 X lo-l3 cm*/s, respectively (Giletti et al., 1978). Assuming the feldspar is a sphere of radius 300 pm and that the oxygen reservoir outside the sphere is well-mixed (i.e., that the system is fluid buffered), the solution of Crank (1975) can be applied: 10% exchange is achieved when Dt/a* = 0.001 (Fig. 14). This corresponds to timescales of approximately 2000 yr at 300°C and 0.04 yr at 800°C. At temperatures of 500 to 600°C likely temperatures for “0 exchange at Chegem, the maximum timescale is approximately ten years (Fig. 14).

The isotopic composition of glass or groundmass can be used as a constraint on the minimum time necessary to create the observed isotopic systematics.

10000

of hegem *tern

100 200 300 ‘loo 500 600 700 800 900

'U'C)

Fig. 14. Calculated time to achieve = 10% exchange in feldspars (An,,) assuming volume diffusion in a sphere of radius 300 pm

surrounded by a large, well-mixed reservoir of water, based on the

solution of Crank (1975). The diagonally ruled zone represents the

estimated range of temperatures for the Chegem hydrothermal

system (500-600°C).

234 C. Gazis et al./ Journal of Volcanology and Geothermal Research 73 (1996) 213-244

We assume that the groundmass S’s0 value was

attained before or during devitrification. Diffusion of

oxygen in rhyolite glass is thought to be controlled

by diffusion of water into the glass structure, and

diffusion coefficients for water in glass are about 5

orders of magnitude higher than hydrothermal diffu-

sion coefficients for oxygen in feldspar (Zhang et al.,

1991). This would imply timescales of less than a

year to achieve 90% exchange. Furthermore, if oxy-

gen exchange is caused by a mechanism other than

simple volume diffusion of H,O (which is especially

likely for the devitrified 10w-‘~O groundmass), then

the timescale will be even shorter.

Another approach to the timescale question is to

treat the plot of SFeldspar vs. Sslass,groundmass as an ‘ ‘oxygen isotope clock’ ’ , in the same way that Gre-

gory et al. (1989) regarded the Squartz-Sfeldspar plot

for fossil hydrothermal systems in nature. We will

consider our Sfe,dspar-Sg,ass,groundmass plot to be essen-

tially a SFeldspar -Sglass plot. Gregory et al. (1989)

pointed out that the slope of an array of data points

on a Squartz-Sfeldspar P lot can in principle indicate the

duration of a hydrothermal system. The array moves essentially from vertical (slope = m) to 45”(slope = 1)

through time as if it were the hand of an “oxygen

isotope clock”. In these terms, a Sfeldspar-Sglass plot is simply a clock which records much shorter

timescales than the Squartz-Sfeldspar clock, because glass exchanges oxygen extremely rapidly, and (rela-

tive to coexisting glass) feldspar can be regarded as the slow-exchanging mineral.

Criss et al. (1987) have shown that the slope of an

array of data points on a S-S plot is related to the

kinetic rate constants (k) for the two exchanging “minerals”, and to the fractional approach of that

mineral-water system to equilibrium (f). The rate

law used in this treatment is:

d R, - = k( aR, - RA)

dt

where R, and R, are the IsO/ I60 ratios in the

mineral and water, respectively, and cx is the mineral/water equilibrium isotopic fractionation factor

for the temperature of interest:

( at equilibrium) eq

The term ,f is defined as:

6” - s .f= 6” _ s’q 1 - exp( -kt)

where S” and Seq are the S-values initially and at

equilibrium, respectively. A mineral-water system is

initially completely out of equilibrium with f’= 0,

whereas f= 1 when the system has attained equilib-

rium.

For buffered open-system conditions, Criss et al.

(1987) showed that for spherical mineral grains, the

kinetic rate constant, k, is related to the diffusion coefficient by the equations:

I

exp( -kt) = 1 -S=exp[--3.5/Dt/u2)

for \iDt/u’ < 0.1

exp( - kt) = 1 -f = exp( - 9.87Dt/a2 + 0.498)

for /Dt/a2 > 0.2

Table 2 shows how these different parameters and

the slopes of a data-point array on a Sfeldspar-Sgla\\

plot change with temperature, time and grain radius. One can see immediately that glass exchanges oxy-

gen isotopes with water extremely quickly, and for

almost all of the time-temperature conditions repre- sented in the table, the glass has completely equili-

brated with the water. Even at 300°C a sphere of

300 km will exchange = 80% in 4 months. In

contrast, a sphere of feldspar of equal size is essentially unchanged (f= 0.004) under these conditions.

Therefore, the degree of attainment of isotopic equi-

librium in the entire feldspar-glass-water system

(and hence the slope of the data point array on a

S feldspar -~g,ass plot) is controlled by the fractional

approach to equilibrium of the feldspar-water system. That is, the glass-water system equilibrates so

rapidly that it plays no role in defining these exchange rate systematics.

Our estimate of the grain radius for both feldspar and glass at Chegem is 300 pm. For temperatures of 300, 600 and 800°C complete feldspar-water isotopic equilibrium (slope = 1) should be attained after

approximately 600,000, 300 and 30 yr, respectively. The measured slope of the Chegem data array on a

6f,ldspar-8glass,groundmass plot is approximately 7 (Fig.

Table 2

C. Gazis et al./Joumal of Volcanology and Geothemwl Research 73 (1996) 213-244 235

Calculated values of various isotopic reaction rate parameters for feldspar-glass pairs

t a(f) Feldspar Glass Slope

k D f kt k D f kt

T = 300°C 0.3 mm, 0.3 yr

0.3 mm, 30 yr

0.3 mm, 300 yr

0.3 mm, 3000 yr

0.3 mm, 30,000 yr

0.1 mm, 3000 yr

0.6 mm, 3000 yr

1 mm, 3000 yr

T = 600°C

0.3 mm, 0.3 yr

0.3 mm, 3 yr

0.3 mm, 30 yr

0.3 mm, 300 yr

0.1 mm, 30 yr

0.6 mm, 30 yr

1 mm, 30 yr

T = 800°C 0.3 mm, 0.03 yr

0.3 mm, 0.3 yr

0.3 mm, 3 yr

0.3 mm, 30 yr

0.1 mm, 3 yr

0.6 mm, 3 yr

1 mm, 3 yr

l.OOE+08 0.03 4.365E - 11 1.40E - 17 0.004 0.004 4E - 08 1.88E - 12 0.98 4.0lE + 00 225.4

1 .OOE + 09 0.03 1.38E - 11 1.40E - 17 0.014 0.014 1E - 08 1.88E - 12 1.00 1.38E + 01 72.9

l.OOE + 10 0.03 4.365E - 12 1.40E - 17 0.043 0.044 very fast 1.88E - 12 1.00 very fast 23.4

l.OOE+ 11 0.03 1.38E - 12 1.40E - 17 0.129 0.138 very fast 1.88E - 12 1.00 very fast 7.8


l.OJlE+ 11 0.01 4.14lE - 12 1.40E - 17 0.339 0.414 very fast 1.88E- 12 1.00 very fast 2.9

l.OOE+ 11 0.06 6.902E - 13 1.40E - 17 0.067 0.069 very fast 1.88E - 12 1.00 very fast 15.0

I.OOE+ 11 0.1 4.14lE - 13 1.40E - 17 0.041 0.041 very fast 1.88E - 12 1.00 very fast 24.7


l.OOE +08 0.03 1.662E- 09 2.03E - 14 0.153 0.166 very fast 3.17E - 09 1.00 very fast 6.5

1.00E + 09 0.03 5.256E - 10 2.03E - 14 0.409 0.526 very fast 3.17E - 09 1.00 very fast 2.4

l.OOE + 10 0.03 1.662E - 10 2.03E - 14 0.81 1.662 very fast 3.17E- 09 1.00 very fast 1.2




l.OOE + 06 0.03 6.315E - 08 2.93E - 13 0.061 0.063 very fast 4.46E - 08 I .OO very fast 16.3 l.OOE + 07 0.03 1.997E - 08 2.93E - 13 0.181 0.200 very fast 4.46E - 08 1.00 very fast 5.5 l.OOE + 08 0.03 6.315E - 09 2.93E - 13 0.468 0.632 very fast 4.46E - 08 1.00 very fast 2.1 l.OOE + 09 0.03 1.997E - 09 2.93E - 13 0.864 1.997 very fast 4.46E - 08 1.00 very fast 1.2

l.OOE + 08 0.01 1.895E - 08 2.93E - 13 0.85 1.895 very fast 4.46E - 08 1.00 very fast 1.2 l.OOEf08 0.06 3.1588 - 09 2.93E- 13 0.271 0.316 very fast 4.46E - 08 1.00 very fast 3.7 l.OOE + 08 0.1 1.895E -09 2.93E - 13 0.173 0.189 very fast 4.46E - 08 1.00 very fast 5.8

Abbreviations: t = time (s); a = effective grain radius km); k = kinetic rate constant (s _ ’ ). D = diffusion coefficient km* s- ‘1; , f = 1 - exp (- kt) (fractional approach to equilibrium; slope = fglara/ffeld).

13b). For 300, 600 and 8OO”C, this corresponds to

approximate timescales of 3000, 3 and 0.03 yr,

similar to the timescale calculated for sample CG-

C2-4H, because in both cases we predicted = 10%

feldspar exchange by volume diffusion in a fluid-

buffered system.

5.6.4. Cross-cutting relationships

The low-‘*0 zone is a = 700-m-thick strati-

graphic horizon located between elevations 2710 and 3390 m along the sample traverses in the southern

and eastern parts of the caldera. In the northeast, this

horizon was apparently offset by faults during late collapse and the faults were reactivated when the resurgent intrusion uplifted a block of tuff. The samples from the uplifted block of tuff appear to represent a portion of the low-‘*0 zone (possibly

that part between 2600 and 3000 m> that has been

both uplifted and tilted. Such an interpretation im-

plies that a significant part of the “0 depletion event

occurred prior to the intrusion of the granodiorite.

Though it is also possible that the apparent trunca-

tion of the low- “0 zone was due to preferential fluid flow in a direction of greater permeability at

that fault contact, the apparent vertical offsets be-

tween Section 1 and Section 2, which are in opposite

directions for the trace-element profile and the oxy-

gen isotope profile (as described above, Fig. lo),

also strongly suggest that the low-‘*0 zone was

formed after deposition of the tuff but prior to resurgence. 4oAr/ 39Ar studies (Gazis et al., 1995) constrain the resurgent intrusion to have been em- placed, crystallized, and cooled to 300°C within 50,000 years of eruption. These relationships, to-

236 C. Guzis et al. /Journal of Volcanology and Geothermal Research 73 f 1996) 213-244

gether with the requirement of high temperatures,

suggest that the low-“0 zone must have formed

immediately after eruption of the Chegem Tuff.

Outflow-facies tuff samples are similar to the

caldera fill, but they are more glass-rich and display

much less pronounced I80 lowering and lesser isotopic disequilibrium effects. Inasmuch as these IgO

signatures exist up to 50 km away from the caldera,

and because the outflow tuff was probably every-

where less than 400 m thick, it is inconceivable that

all of these “0 depletions could have been caused

by a single hydrothermal system. The only likely

alternatives are: (1) that the intracaldera tuff and the

outflow tuff interacted in a somewhat similar fashion

with separate batches of meteoric waters; or (2) that

the “0 lowering occurred immediately prior to (or

during) eruption, a possibility that we do not favor

(see the discussion below).

The isotopic composition of the granodiorite por-

phyry intrusion further complicates the puzzling

cross-cutting relationships. Like the outflow tuff, the

granodiorite porphyry has a 10w-‘~O groundmass

even though its phenocrysts have “normal” igneous

a-values. Two pairs of samples of adjacent tuff and

granodiorite, collected several meters apart across

the intrusive contact (CG-C-53A, B; CG-C-58A, B),

consistently indicate that the tuff is 3 to 4%0 heavier

than the nearby granodiorite. However, the quartz is

isotopically identical in both the tuff and the gran-

odiorite (6”Oquartz = + 8.5 and + 8.61, strongly sug-

gesting that both kinds of rock initially formed from

isotopically similar magmas. Three plausible expla-

nations of this cross-cutting relationship are: (1)

meteoric water interacted briefly with the granodiorite before it came into contact with the resurgent

block of tuff; (2) meteoric water interacted more

extensively with the granodiorite than with the tuff,

either because the intrusion was hotter or more per-

meable, or because its groundmass was more prone

to oxygen exchange (perhaps because it was still glassy, while the tuff had already been devitrified

and homfelsed by the intrusion); or (3) both samples originally suffered “0 depletion, but superposition of a late, lower-temperature alteration event increased the S180 value of the fine-grained tuff much

more than the more coarsely crystalline granodiorite. The post-caldera andesite lavas do not show any

signs of interaction with meteoric water. In fact,

there is an abrupt change in whole-rock 6”O across

the unconformity at the top of the intracaldera tuff in

Section 1 (below the andesites). Because this horizon

also contains glacial outwash gravels, and thus repre-

sents a significant time interval, it seems certain that

the process of water-rock/glass interaction that

caused the change in 6”O of the tuff must have

ceased on the surface by the time the andesites were

erupted.

5.7. Possible ‘80-depletion scenarios

Given the above, we here discuss several possible

scenarios of IgO depletion at Chegem.

5.7.1. Assimilation and / or melting of low- I80 rocks

This process has been suggested to explain low-

‘a0 magmas in the Yellowstone volcanic field

(Taylor, 19871, Crater Lake (Bacon et al., 1989) and

the Calabozos caldera complex (Grunder, 1987). Ei-

ther (1) a rather substantial amount of hydrothermally altered rock foundered into the upper part of

the magma chamber and exchanged with this magma

or became assimilated by the magma prior to erup-

tion, or (2) hydrothermally altered rocks were par-

tially melted in the roof or walls of the magma

chamber. Neither of these possibilities is likely at

Chegem because: (1) The most striking “0 depletions are observed right in the middle of this series

of tuffs, so the 10w-‘~O effects cannot be the result

of assimilation in the upper part of the magma chamber. (2) In contrast to the above-described

caldera complexes, there are no low-“0, hydrother-

mally altered country rocks at Chegem. (3) There is no compositional or mineralogical evidence for large

amounts of assimilation. (4) Massive assimilation is incompatible with the extremely short timescales

(< 6 months) that are necessary to preserve the

observed oxygen isotope disequilibrium between

feldspar and coexisting silicate melt.

5.7.2. Direct water-magma interaction

Direct interaction between meteoric water and magma has been proposed to explain other occur- rences of low- “0 magmas (Lipman and Friedman,

1975; Hildreth et al., 1984; Harris and Erlank, 1992). A criticism of this mechanism has always been that, except for possible buoyant uprise of bubbles in a

C. Gazis et al./Joumal of Volcanology and Geothermal Research 73 (1996) 213-244 237

H,O-saturated melt, magmas have a very low or nonexistent permeability and are under lithostatic pressures, whereas the meteoric waters in the country rocks are under much lower (hydrostatic) pressures, and are separated from the magma chamber by a zone of hot, plastic (unfractured) rock. Therefore, the large amounts of H,O that needs to be cycled into and then out of such magma chambers will be in large part controlled by the very slow diffusion of H,O through ductile rock or silicate melt, and there is simply not enough time available for this to happen (Taylor, 1987). At Chegem, this problem is compounded by the fact that the feldspars have hardly exchanged at all with the meteoric waters, and this should have been relatively rapid, occurring in days (e.g., Gregory et al., 1989).

5.7.3. Water-glass interaction deep in vent system

during or just prior to eruption

At Chegem, we know that low- ‘* 0 volcanic rocks, 10w-‘~O intrusive rocks, and 10w-‘~O glasses all exist. However, there is no proof that any of these were actually derived from 10w-‘~O melts, although the mere existence of the low- I80 glasses implies that these materials were depleted in 180 while they were in the liquid, undercooled, or glassy state, and, as discussed above, this must have occurred at high temperatures ( > 250-350°C). One possible way that glass and water might interact at elevated temperatures is for meteoric water to infiltrate deep into the vent system of a caldera-forming system just before or during eruption. The extreme pressure differen- tials and the higher permeability provided by the interconnected fractures that develop during the cracking of the rapidly quenched porphyritic glass might make large-scale fluid infiltration more plausible in this environment. For example, Sheridan et al. (1981) propose a somewhat analogous mechanism during the late stages of the A.D. 79 eruption of Vesuvius (Italy).

In addition, there is a possible correlation between the caldera-wide low- “0 zone and the second-order fluctuations in major- and trace-element compositions, and as we suggested earlier (Gazis et al., 1992), these compositional variations may be related to eruption dynamics. Perhaps fluctuating eruption rates or eruptions from different vents disrupted the progressive downward tapping of a compositionally

zoned magma chamber. This conceivably could per- mit the entrainment of hot meteoric fluids into the fragmented magma or highly fractured glass deep within the vent conduit system.

The problem with this mechanism is that very large integrated fluxes of water are required by the oxygen isotope systematics, and these must be sup- plied over a very short period of time. That is, in days to months (?), a large volume of water similar to the volume of the ‘*O-depleted zone itself would need to penetrate deep into the vent system, and some efficient physical mixing process would then be needed to produce intimate exchange of oxygen between these large volumes of materials. Further- more, the possible structural offsets in sections 1 and 2 of the trace-element and oxygen isotopic trends (discussed above) are in opposite directions, arguing against our earlier suggestion (Gazis et al., 1992) that both kinds of trends were formed at the same time.

5.7.4. A short-lived, high-temperature hydrothermal

system with efJicient recharge in the upper 7.5’0 m of

caldera fill

If the caldera-wide low-“0 zone at Chegem was formed by a flow system that developed after the tuffs were deposited, then the W/R profiles of Fig. 12 clearly imply that this was a very heterogeneous flow system containing horizons of variable permeability and fluid flow. In this scenario, the horizon of greatest fluid flux would be at the level of sample CG-C2-4H, which was in fact observed to be more poorly welded than the surrounding tuff. Further- more, the al80 profile is most easily explained if the flow direction is nearly horizontal. However, it is unlikely that the flow system was perfectly horizontal, because under such conditions it is difficult to initiate convection, particularly if the isotherms are also sub-horizontal. Even a slight tilt (= 5”), of either the boundaries of the flow system or the isotherms is sufficient to drive circulation (Criss and Hofmeister, 1991).

One drawback of attributing the “0 depletions to a hydrothermal system is the absence of any hydrous alteration minerals replacing the feldspar, biotite and groundmass. Consequently, a practically unique combination of unusually high hydrothermal temperatures, a very short time interval, and large fluxes of

238 C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 C 1996) 213-244

water are required, and all of these features must act

in concert. The flow system must be essentially shut

down before the temperatures drop into the range where hydrous alteration minerals are stable (i.e.,

< 400°C). Also, it is unclear why this hydrothermal

system would be so short-lived and not have contin-

ued on to produce a full range of mineralogical and

isotopic alteration as is observed at many other

calderas (e.g., Lake City, Valles, Long Valley). Per-

haps the difference is that the Chegem hydrothermal

event is mainly related to simple cooling of a single

welded tuff unit, whereas the others involve addi-

tional heating during emplacement of a variety of

ring intrusions, as well as a large central resurgent

dome intrusion.

This scenario, the existence of a short-lived,

“perched” hydrothermal system at Chegem immedi-

ately after eruption, depicted in Fig. 15, is our

preferred mechanism for producing the isotopic

changes in the Chegem Tuff. We propose an intense

fumarolic event similar to that observed at the Valley of Ten Thousand Smokes (discussed below) to ex-

plain the observed oxygen isotope trends. The obser-

vations described in this paper can all be fitted into a

geologic history encompassing the following se-

quence of events:

(1) Caldera-forming eruption and deposition of

tuff (Fig. 15a), accompanied by slight tilting either

of intracaldera stratigraphy or of isotherms (by cooling). The tuff of Section 2 may have been faulted

downward = 500 m relative to Section 1 during a

late-stage intracaldera collapse.

(2) Immediately after deposition, there was vigor-

ous hydrothermal interaction between large amounts

of meteoric water and rock in the upper 750 m of

caldera fill while the tuff was still very hot and was undergoing welding in the center (> 500 to 600°C)

but had cooled significantly in the upper 50 m (Fig.

15b). Water circulated in a myriad of small convec-

tion cells, flowing into the hot ash through fractures and fissures, being heated and then being ejected upward as hot steam through fumaroles. Flow was especially intense along a zone of higher permeability about 350 m below the surface, and above the zone of intense welding. The outflow tuff probably was affected in a somewhat similar fashion at the same time, though with much smaller amounts of water and perhaps at lower temperatures (however,

3 Immediately after eruption: hot-ash flow tuff deposited

3 Immediately after deposition: short, intense hydrothermal event (= 10 years)

rnNP”W WatLr

C Resurgence

Fig. 15. Preferred ‘*O-depletion scenario at Chegem. (a) Immedi-

ately after eruption: ash-flow tuff is deposited and is near horizon-

tal. The emplacement temperatures are approximately 600-700°C.

(b) Immediately after deposition: A short-lived hydrothermal sys-

tem is active in the upper 750 m of intracaldera tuff where the tuff

is still glassy and very permeable. Temperatures were high (500-

600°C) and hot meteoric waters circulated vigorously in small

hydrothermal convection cells. This fluid flow was focused in a

more permeable zone of poorly welded tuff at = 350 m depth and

did not pass below 750 m depth, where the tuff was more densely

welded. This hydrothermal system was active for about ten years.

Glaciation and erosion began some time after this. (c) Resurgence:

a small granodiorite porphyry intruded the intracaldera tuff on the

north side of the caldera along the ring fault. This pushed up and

tilted a block of intracaldera tuff. Small amounts of meteoric

hydrothermal waters circulated locally around the intrusion. An-

desite flows may have erupted at this time.

note that the outflow tuff remnants are also welded and thus were undoubtedly very hot). This water- rock interaction was accompanied by hydration and profound “O/ I60 alteration of glasses and devitrified groundmass of the core of the caldera fill. The hydrothermal event continued for = 10 to 25 yr and abruptly ended when the upper section of tuff had been sufficiently cooled and devitrified. At this point,

C. Gazis et al. /Journal of Volcanology and Geothenal Research 73 (1996) 213-244 239

the fractures and other hydrothermal conduits had been partially sealed by mineral deposition (e.g., quartz) such that the permeability of the tuff body was lower and hydrothermal circulation was no longer possible.

(3) Glacial erosion occurred some time after the deposition of the tuff. The timing of this glaciation relative to the hydrothermal event and the intrusion of the granodiorite is poorly constrained.

(4) Resurgence of granodiorite porphyry in the northeast part of the caldera (Fig. 1%) possibly accompanied by uplift of Section 2 relative to Sec- tion 1 (= 200 m). The intrusion also pushed up and tilted the small resurgent block of isotopically altered tuff. The vigorous high-temperature hydrothermal activity had largely ceased at this point but some meteoric water was still present at depth, producing a small, high-temperature hydrothermal system in this part of the caldera, which interacted with the (glassy?) groundmass of the intrusion as it cooled and devitri- lied, but not with the immediately adjacent tuffs.

(5) Eruption of andesite flows after hydrothermal activity had ceased at the surface. The eruption of andesites could actually be synchronous with the intrusion of the granodiorite if fluid flow had ceased at the surface before it had stopped at depth.

This model implies that the dacitic tuffs of Zone C were once low-‘*0 rocks also. Their present high Si*O values were probably caused by later low-temperature interaction with meteoric water near the surface before significant erosion of the caldera fill had occurred.

The source of the meteoric waters and geometry of the flow system is poorly constrained by this study. More work will need to be done to sort out these details. Sr isotope studies (Gazis, 1994) indicate that the waters carried radiogenic strontium which could only have come from the crystalline basement to the south of the caldera. However, this crystalline basement is also present within the intracaldera tuff sequence in as slide debris in the form of megabreccias and lithic fragments, particularly along the margins of the caldera, perhaps providing a local, internal source for this radiogenic strontium. One possible source for the meteoric water is a lake or glacier near the caldera fill.

Given the large amounts of water that have to be heated to produce these “O/ I60 effects, tempera-

tures of 500 to 600°C probably could not be sus- tained for more than lo-25 yr, because of the constraints of heat balance. The following simple heat balance calculation illustrates this point: The W/R ratio for the hydrothermally altered zone of tuff probably falls between 0.3 and 0.6 (see earlier discussion), which converts to 0.17 to 0.34 in weight units (g H,O/g rock). The average heat capacity of water/steam over the temperature range from 25 to 500°C is 0.6 calg -’ “C-l (Haar et al., 1984). Thus, the heat needed to raise the temperature of water from 25 to 500°C is 285 Cal/g H,O. In addition, the water will be vaporized. The enthalpy of vaporiza- tion of water at 1 atm and 100°C is 540 Cal/g H,O (Marsh, 1987), so the total heat needed to bring the water to 500°C and vaporize it is = 825 Cal/g H,O. Using the W/R ratio to convert this to rock weight, we find that the heat needed is about 140 to 280 Cal/g rock. Rhyolite tuff has a heat capacity of about 0.22 Cal/g (Riehle, 1973). If such a rock is cooled from 800 to 5OO”C, it will produce 66 Cal/g of heat. Although this is not sufficient to heat and vaporize the water, an additional 30-70 Cal/g of latent heat is given off as the glass crystallizes or devitrifies. Thus the total heat budget of the rock is 96 to 136 Cal/g, somewhat less than to the heat budget of the water. The additional heat required in this model can be readily inferred to come from the cooling of the thick mass of welded tuff below the permeable zone. This simple calculation suggests that hot, devitrifying Chegem Tuff had just enough heat to sustain the high temperatures and water-rock ratios in the scenario we propose, but that after the welded tuff had devitrified (crystallized), it could no longer sustain these high temperatures.

Since we know the timescales of this hydrothermal event and the W/R ratios, we can calculate fluxes of water through the system if we assume some geometry of fluid flow. The flux of water or of an aqueous fluid (in moles per unit cross-sectional area per unit time) through an elemental volume of rock (a cube with each side length d in cm) is given by (e.g., Taylor, 1996):

F bV4(NwR)

H,O = (Wt)

where W/R is the material-balance open-system

240 C. Gazis et al./ Journal oj’Valcanology and Geothemal Research 73 (1996) 213-244

water rock ratio, t is the duration of fluid flow in s,

h/d is the total flow length (A) divided by the size

of each elemental volume, and m is a factor which

converts the W/R ratio to moles of HZ0 per unit

elemental volume. If we arbitrarily choose an ele-

mental volume of 1 cm3 (d = 1 cm), then m is

approximately 0.076 moles. The W/R ratio for the

system probably falls between 0.3 and 0.6 (see ear-

lier discussion), and so for a length scale of 750 m

(i.e., the side of an equant convection cell, defined

by the thickness of the hydrothermal zone) and a

timescale of 10 to 25 yr, we have fluxes of 2.2 X

10e6 to 1.1 X lop5 molcm-*s-I (equivalent to

4.0 X lo-* to 2.0 X 10e7 lcm~‘s-’ for liquid

H,O). A final question is whether this high flux of

aqueous fluid is feasible. That is, can these fluxes be

attained with geologically reasonable permeabilities.

We can make an order of magnitude estimate of the

permeability needed by using a simple version of

Darcy’s Law (Dullien, 1979):

-k L’=-vp*

77

where I? is the Darcy velocity of the fluid

(cm3 cm- * s- ’ ), k is the permeability in cm2, -Q is

the fluid viscosity (poise), and P * is the piezometric

pressure (dyn/cm’). Thus, since the Darcy velocity

is related to the flux we calculated by:

where M is the molar weight of the aqueous fluid

and p is its density, we can cast the permeability k

in terms of these variables:

k= -41,o *(:)*!a We must make estimates of the water density and viscosity, and the piezometric pressure gradient. At 500°C and a pressure sufficient to maintain com- pressed water (= 100 bar), the density of water is 0.8 g/cm” and the viscosity of water is about 1 X lo-’ poise. Because this is a near-surface environment, we will assume that the water is at hydrostatic pressure in hydrothermal cells = 750 m in diameter and that when the water rises as steam, it exerts no

pressure. This is equivalent to a pressure gradient of

980 dyn/cm3. These values predict permeabilities of 5.0 X lo-” to 2.5 X lo-“’ cm* or 5 to 25 milli-

darcy. Though these values are fairly high, measure-

ments of permeabilities in ash-flow tuff from Yucca

Mountain, Nevada (Nelson and Anderson, 19921,

have revealed comparable values. In fact, glassy

sections are commonly several orders of magnitude more permeable than devitrified tuff, with values as

high as 100 millidarcy (Nelson and Anderson, 1992).

Devitrification or mineralogical alteration tends to

reduce the permeability of such a rock. Also, the

permeabilities at Yucca Mountain were measured on

borehole samples and do not reflect the bulk perme-

ability of the rock. Any fracture permeability will

increase these values, and so our estimated perme-

abilities for the Chegem Tuff would seem to be

entirely reasonable, and well capable of allowing

circulation of the required amounts of H,O.

5.8. The Valley of Ten Thousand Smokes-A histori-

cal analogue?

In searching to understand the process which caused the extreme I80 depletion and disequilibrium

at Chegem, and in trying to develop a realistic model

of this system, we have also sought a modern ana-

logue. We suggest that the fumarolic activity at the

Valley of Ten Thousand Smokes in Alaska may be

just such an analogue, though at a different scale,

and in a somewhat different geologic setting (namely

in a much smaller eruption and in the thinner out-

flow facies of a welded ash-flow sheet rather than in

the thick caldera-fill facies as at Chegeml. The Valley of Ten Thousand Smokes is located at

the foot of Mt. Katmai on the Alaska Peninsula.

During the 1912 eruption of Novarupta caldera, = 20 km3 of air-fall tephra and 1 l- 15 km’ of ash-flow

tuff were ejected, much of which filled the nearby

valley with a = 200-m-thick sheet (Hildreth, 1983). This was one of the most voluminous historic eruptions, and was unique in that the pyroclastic flows

came to rest on land and were subsequently welded. The valley was discovered in 1916 by Robert F.

Griggs, the head of a National Geographic Society expedition, and so named because of the many fissures and fumaroles that were vigorously emitting steam. In a subsequent expedition in 1917, Shipley

C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 (I 996) 213-244 241

(1920) collected gases and attempted to measure

fumarole temperatures, but as he only had a mercury

thermometer, he could not measure temperatures above 357°C (the boiling point of mercury). In 1919,

Allen and Zies (1923) succeeded in measuring fuma-

role temperatures as high as 645°C using Hoskins

thermocouples. Several other fumaroles with temper-

atures above 500°C were measured and several oth-

ers were observed to melt lead or zinc. Allen and Zies (1923) also determined that the gases were

made up almost entirely of steam. Details of their

studies of the fumarolic gases and the near-surface mineral deposits are given in Allen and Zies (1923)

and Zies (1924a,b,1929)). Griggs (1922) and Fenner (1920) also published descriptions of those early

expeditions.

The hydrothermal activity at the Valley of Ten

Thousand Smokes was vigorous, but short-lived.

Within 15 yr, the high-temperature activity had

ceased and the fumaroles became hot springs and

mud pots before dying out altogether (Zies, 1924b), except in the vent region and one locality outside the

vent region (Keith, 199 1).

By comparison, the eruption of the Chegem

caldera produced pyroclastic deposits an order of

magnitude greater in volume than those at the Valley

of Ten Thousand Smokes. Furthermore, the intracaldera fill is at least two km thick, as opposed to the

200-m thickness at the Valley of Ten Thousand

Smokes. Thus, one would expect that a similar hy-

drothermal event at Chegem would be much more

vigorous and involve larger volumes of water. This

is certainly consistent with our observations. This

correlation between the actual events at Katmai and

our hypothetical model of the Chegem system is

made even more compelling because we formulated

the basic parameters of our Chegem model (i.e., temperatures of 500-600°C a timescale of lo-25

yr, and exchange with large volumes of H,O) long before we realized that these same parameters are in

fact part of the historical record at the Valley of Ten

Thousand Smokes.

It would also be interesting to compare the dis-

charge rates at the Valley of Ten Thousand Smokes with the calculated fluxes of water in the Chegem system. Unfortunately, no careful measurements were made of discharge rates. However, Zies (1929) did make an estimate that the total volume of steam

‘ ‘exhaled” by the fumaroles was 26 X lo6 l/s

(calculated at 100°C and 1 atm). If the density of the

steam is 0.00059 g/cm3 and the total area through which it flows is the area of the valley ( = 120 km* ),

then this estimated discharge rate converts to a flux

of = 7 x lo-’ molcm-‘s-‘, within an order of

magnitude of our lowest calculated value. First-hand

descriptions of fumaroles at the Valley of Ten Thou-

sand Smokes also suggest that the flux of water

through these permeable tuffs was in fact compara-

ble to the calculated flux at Chegem. For example,

Griggs (1918) wrote: “Now [the valley] is traversed by hundreds of

fissures extending along its margin or criss-crossing

its floor. These fissures are the seat of several mil-

lions of volcanic vents of all sizes, from great volca-

noes pouring forth columns of vapour more than a

mile high, down to minute jets of gas which pass

unnoticed amongst their greater neighbours... For

here, continually rising quietly from the ground with-

out explosive action of any sort, is more vapour than is given off by all the rest of the world’s volcanoes

put together (except during a period of dangerous

eruption). ’ ’ One possible shortcoming of this analogy is that

the fumarolic activity at the Valley of Ten Thousand

Smokes is known to have been accompanied by

extensive mineralogical alteration and deposition of

fumarolic minerals (e.g., Zies, 1929; Papike et al.,

1991; Keith, 1991; Spilde et al., 1993). No evidence

has been found for fossil fumaroles or their deposits

in the Chegem intracaldera fill. However, this evi-

dence would only be expected in the near-surface portion of the caldera fill, which has been eroded

more extensively at Chegem than the lower part.

Also, the major fumaroles at Katmai were spaced

roughly 500-1000 m apart (fig. 1 in Zies, 1929) and

thus their deeper conduits might be difficult to locate at Chegem without doing careful fieldwork. More-

over, fumarolic encrustations are often metastable or

unstable under cooling fumarolic conditions, are sol-

uble in water or acids formed at the surface, and/or

are easily eroded (Keith et al., 1992); thus they are

not likely to be preserved in a 2.8 million year old caldera system.

This analogy between the hydrothermal event at Chegem and the fumarolic activity at the Valley of Ten Thousand Smokes can easily be tested by exam-

242 C. Gazis et al. / Journal of Volcanology and Geothermal Research 73 (1996) 213-244

ining samples from the Valley of Ten Thousand

Smokes. In particular, if the Valley of Ten Thousand

Smokes is ever drilled, a hole in the tuff would yield

a core which would be ideal for testing this model.

Acknowledgements

This research was part of the Caucasus-Western

USA cooperative project which was supported by the

Russian Academy of Sciences, the U.S. National

Academy of Science, and the U.S. Geological Sur-

vey. The authors are grateful to Peter Lipman for his

extremely helpful encouragement and guidance on

this project, as well as to Charles Bacon and an

anonymous reviewer for their helpful comments on

the manuscript. We also thank Valentine Beus, Ana-

toly Gurbanov, Nikolai Koronovsky, Igana Latifova,

and Peter Marchev for sharing in discussions, field

work and sample collection. Financial support for

this study was also provided by NSF grant EAR90-

19190, Division of Geological and Planetary Sci-

ences, California Institute of Technology, Pasadena,

CA, 91125. Contribution No. 5618, Division of Geological

and Planetary Sciences, California Institute of Tech-

nology.

References

Allen, E.T. and Zies, E.G., 1923. A chemical study of the

fumaroles of the Katmai region. Natl. Geogr. Sot. Contrib.

Tech. Pap., Katmai Ser., 2: 75-155.

Bacon, C.R., Adami, L.H. and Lanphere, M.A., 1989. Direct

evidence for the origin of 10w-‘~O silicic magmas: quenched

samples of a magma chamber’s partially-fused granitoid walls,

Crater Lake Oregon. Earth Planet. Sci. Lett., 96: 199-208.

Bierwith, P.N., 1982. Experimental welding of volcanic ash.

Ph.D. Thesis, Monash Univ.

Bogatikov, O.A., Gurbanov, A.G., Koronovsky, N.V., Kovalenko,

V.I., Lipman, P.W. and Tsvetkov, A.A., 1992. 2.5 Ma ash-flow

caldera at Chegem River in the northern Caucasus Mountains

(USSR), contemporaneous granites and associated ore bodies.

Izv. Russian Acad. Sci. Ser. Geol., 1: 5-21 (in Russian);

English Transl. in Int. Geol. Rev., 34: 131-147.

Bowman, J.R., Willett, S.D. and Cook, S.J., 1994. Oxygen iso-

topic transport and exchange during fluid flow: One-dimen-

sional models and applications. Am. J. Sci., 294: 1-55.

Cas, R.A.F. and Wright, J.V., 1988. Volcanic Successions. Unwin

Hyman, London, 528 pp.

Cerling, T.E., Brown, F.H. and Bowman, J.R., 1985. Low-temper-

ature alteration of volcanic glass: Hydration, Na, K, 180, and

Ar mobility. Chem. Geol., 52: 281-293.

Clayton, R.N. and Kieffer, S.W., 1991. Oxygen isotopic ther-

mometer calibrations. In: H.P. Taylor Jr., J.R. O’Neil and I.R.

Kaplan (Editors), Stable Isotope Geochemistry: A Tribute to

Samuel Epstein. Geochem. Sot. Spec. Publ., 3: 3-10.

Cole, D.R., Ohmoto, H. and Lasaga, A.C., 1983. Isotopic ex-

change in mineral-fluid systems. I. Theoretical evaluation of

oxygen isotopic exchange accompanying surface reactions and

diffusion. Geochim. Cosmochim. Acta, 47: 1681-1693.

Cole, D.R., Ohmoto, H. and Jacobs, G.K., 1992. Isotopic ex-

change in mineral-fluid systems: III. Rates and mechanisms

of oxygen isotope exchange in the system granite-H,O+ NaCl

+ KC1 at hydrothermal conditions. Geochim. Cosmochim.

Acta, 56: 445-466.

Craig, H., 1961. Isotopic variations in meteoric waters. Science,

133: 1702-1703.

Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press,

Oxford.

Criss, R.E. and Hofmeister, A.M., 1991. Application of fluid

dynamical principles in tilted permeable media to terrestrial

hydrothermal systems. Geophys. Res. Lett.. 18: 199-202.

Criss, R.E., Gregory, R.T. and Taylor Jr., H.P., 1987. Kinetic

theory of oxygen isotope exchange between minerals and

water. Geochim. Cosmochim. Acta., 5 1: 1099- 1108.

Dullien, F.A.L., 1979. Porous Media. Fluid Transport and Pore

Structure. Academic Press, New York, NY.

Fenner, C.N., 1920. The Katmai region, Alaska, and the great

eruption of 1912. J. Geol., 28: 569-606.

Fridrich, C.J., Smith, R.P., Dewitt, E. and McKee, E.H., 1991.

Structural, eruptive and intrusive evolution of the Grizzly Peak

caldera, Sawatch Range, Colorado. Geol Sot. Am. Bull., 103:

1160-1177.

Friedman, I. and Smith, R.L., 1960. A new dating method using

obsidian. Am. Antiq., 25: 347-352.

Friedman, I., Smith, R.L. and Long, R.L., 1966. Hydration of

natural glass and formation of perlite. Geol. Sot. Am. Bull..

77: 323-328.

Gazis, C.A., 1994. An isotopic study of the fluid flow and thermal

history of the 2.8 Ma Chegem ash-flow caldera and related

intrusive rocks (Caucasus Mountains, Russia). Ph.D. Thesis,

California Inst. Technol., 280 pp.

Gazis, C.A., Hon, K., Taylor Jr., H.P., Tsvetkov, A.A. and Beus,

V., 1992. Stratigraphically-controlled geochemical trends

within the Chegem caldera, central Caucasus Mountains, Rus-

sia. EOS, 73: 623.

Gazis, CA., Lanphere, M., Taylor Jr., H.P. and Gurbanov, A.,

1995. 40Ar/ 39Ar and ‘*O/ I60 studies of the Chegem ash-flow

caldera and the Eldjurta Granite: Cooling of two Late Pliocene

igneous bodies in the Greater Caucasus Mountains, Russia.

Earth Planet. Sci. Lett., 134: 377-391.

Gazis, CA., Hon, K., Taylor Jr., H.P. and Saleeby, J.S., in prep.

Geochemical and Sr isotopic studies of the Chegem igneous

rocks and the Eldjurta Granite (Caucasus, Russia): Petroge-

netic and hydrothermal signatures.

Giletti, B.J., Semet, M.P. and Yund, R.A., 1978. Studies in


diffusion-III. Oxygen in feldspars: an ion microprobe deter-

mination. Geochim. Cosmochim. Acta, 42: 4.5-57.

Gregory, R.T. and Taylor Jr., H.P., 1986. Non-equilibrium, meta-

somatic IsO/ I60 effects in upper mantle mineral assem-

blages. Contrib. Mineral. Petrol., 93: 124-135.

Gregory, R.T., Criss, R.E. and Taylor Jr., H.P., 1989. Oxygen

isotope exchange kinetics of mineral pairs in closed and open

systems: Applications to problems of hydrothermal alteration

of igneous rocks and Precambrian iron formations. Chem.

Geol., 75: l-42.

Griggs, R.F., 1918. The eruption of Katmai. Nature, 101: 497-499.

Griggs, R.F., 1922. The Valley of Ten Thousand Smokes. Natl.

Geogr. Sot., Washington, DC.

Grunder, A.L., 1987. Low 6l*O silicic volcanic rocks at the

Calabozos Caldera Complex, Southern Andes. Contrib. Min-

eral. Petrol., 95: 71-81.

Haar, L., Gallagher, J.S. and Kell, G.S., 1984. NBS/NRC Steam

Tables. Hemisphere Press, New York, NY.

Harris, C. and Erlank, A.J., 1992. The production of large-volume,

10w-8’~0 rhyolites during rifting of Africa and Antarctica:

The Lebombo Monocline, southern Africa. Geochim. Cos-

mochim. Acta, 56: 3561-3570.

Hildreth, W., 1979. The Bishop Tuff: Evidence for the origin of

compositional zonation in silicic magma chambers. Geol. Sot.

Am. Spec. Pap., 180: 43-75.

Hildreth, W., 1981. Gradients in silicic magma chambers: Implica-

tions for lithospheric magmatism. J. Geophys. Res., 86:

10,153-10,192.

Hildreth, W., 1983. The compositionally zoned eruption of 1912

in the Valley of Ten Thousand Smokes, Katmai National Park,

Alaska. J. Volcanol. Geotherm. Res., 18: l-56.

Hildreth, W., Christiansen, R.L. and O’Neil, J.R., 1984. Catas-

trophic isotopic modification of rhyolitic magma at times of

caldera subsidence, Yellowstone Plateau Volcanic Field. J.

Geophys. Res., 89: 8339-8369.

Hildreth, W., Halliday, A.N. and Christiansen, R.L., 1991. Iso-

topic and chemical evidence concerning the genesis and con-

tamination of basaltic and rhyolitic magma beneath Yellow-

stone Plateau volcanic field. J. Petrol., 32: 63-138

Johnson, C.M., Lipman, P.L. and Czamanske, G.K., 1990. H, 0,

Sr, Nd, and Pb isotope geochemistry of the Latir volcanic field

and cogenetic intrusions, New Mexico, and relations between

evolution of a continental magmatic center and modifications

of the lithosphere. Contrib. Mineral. Petrol., 104: 99-124.

Keith, T.E.C., 1991. Fossil and active fumaroles in the 1912

eruptive deposits, Valley of Ten Thousand Smokes, Alaska. J.

Volcanol. Geotherm. Res., 45: 227-254.

Keith. T.E.C., Thompson, J.M., Hutchinson, R.A. and White,

L.D., 1992. Geochemistry of waters in the Valley of Ten

Thousand Smokes region, Alaska. J. Volcanol. Geotherm.

Res., 49: 209-23 1.

Lambert, S.J. and Epstein, S., 1980. Stable isotope investigations

of an active geothermal system in Valles caldera, Jemez Mountains, New Mexico. J. Volcanol. Geotherm. Res., 8: 111-129.

Larson, P.B. and Taylor Jr., H.P., 1986a. An oxygen isotope study

of hydrothermal alteration in the Lake City caldera, San Juan

Mountains, Colorado. J. Volcanol. Geotherm. Res., 30: 47-82.

Larson, P.B. and Taylor Jr., H.P., 1986b. An oxygen-isotope study

of water-rock interaction in the granite of Cataract Gulch,

western San Juan Mountains, Colorado. Geol. Sot. Am. Bull.,

97: 505-515.

Lipman, P.W., 1967. Mineral and chemical variations within an

ash-flow sheet from the Aso caldera, Southwestern Japan.

Contrib. Mineral. Petrol., 16: 300-327.

Lipman, P.W. and Friedman, I., 1975. Interaction of meteoric

water with magma: An oxygen-isotope study of ash-flow

sheets from southern Nevada. Geol. Sot. Am. Bull., 86:

695-702.

Lipman, P.W., Bogatikov, O.A., Tsvetkov, A.A., Gazis, C., Gur-

banov, A.G., Hon. K., Koronovsky, N.V., Kovalenko, V.I. and

Marchev, P., 1993. 2.8 Ma ash-flow caldera at Chegem River

in the northern Caucasus Mountains (Russia), contemporane-

ous granites, and associated ore deposits. J. Volcanol.

Geotherm. Res., 57: 85-124.

Lofgren, G., 1970. Experimental devitritication rates of rhyolite

glass. Geol. Sot. Am. Bull., 81: 553-560.

Marsh, K.N. (Editor), 1987. Recommended Reference Materials

for the Realization of Physicochemical Properties. Blackwell,

Oxford.

Marshall, R.R., 1961. Devitritication of natural glass. Geol. Sot.

Am. Bull., 72: 1493-1520.

Muehlenbachs, K., Anderson Jr., A.T. and Sigvaldason, G.E.,

1974. Low-018 basalts from Iceland. Geochim. Cosmochim.

Acta, 38: 577-588.

Nelson, P.H. and Anderson, L.A., 1992. Physical properties of ash

flow tuff from Yucca Mountain, Nevada. J. Geophys. Res., 97:

6823-6841.

Norton, D. and Taylor Jr., H.P., 1979. Quantitative simulation of

the hydrothermal systems of crystallizing magmas on the basis

of transport theory and oxygen isotopic data. J. Petrol., 20:

42 I -486.

Papike, J.J., Keith, T.E.C., Silde, M.N., Galbreath, K.C., Shearer,

C.K. and Lam, J.C., 1991. Geochemistry and mineralogy of

fumarolic deposits, Valley of Ten Thousand Smokes, Alaska:

Bulk chemical and mineralogical evolution of dacite-rich pro-

tolith. Am. Mineral., 76: 1662-1673.

Philip, H., Cistemas, A., Gvishiani, A. and Gorshkov, A., 1988.

The Caucasus: An actual example of the initial stages of

continental collision. Tectonophysics, 161: l-21.

Riehle, J.R., 1973. Calculated compaction profiles of rhyolitic

ash-flow tuffs. Geol. Sot. Am. Bull., 4: 2193-2216.

Ross, C.S. and Smith, R.L., 1961. Ash-flow tuffs: Their origin,

geologic relations and identification. U.S. Geol. Surv. Prof.

Pap., 366: 1-81.

Sheridan, M.F., Barberi, F., Rosi, M. and Santacroce, R., 1981. A

model for Plinean eruptions of Vesuvius. Nature, 289: 282- 285.

Shipley, J.W., 1920. Some chemical observations on the volcanic

emanations and incrustations in the Valley of 10,000 Smokes,

Katmai, Alaska. Am. J. Sci., Ser. 4, 50: 141-153.

Smith, B.M. and Suemnicht, G.A., 1991. Oxygen isotope evi-

244 C. Gazis et al. /Journal of Volcanology and Geothermal Research 73 (1996) 213-244

dence for past and present hydrothermal regimes of Long

Valley caldera, California. J. Volcanol. Geotherm. Res., 48:

319-339.

Smith, R.L., 1979. Ash-flow magmatism. Geol. Sot. Am. Spec.

Pa,,., 180: 5-27.

Smith, R.L. and Bailey, R.A., 1966. The Bandelier Tuff: A study

of ash-flow eruption cycles from zoned magma chambers.

Bull. Volcanol., 29: 83-104.

Spilde, M.N., Brearley, A.J. and Papike, J.J., 1993. Alteration of

plagioclase and pyroxene phenocrysts in a fissure fumarole,

Valley of Ten Thousand Smokes, Alaska. Am. Mineral., 78:

1066-1081.

Stuckless, J.S. and O’Neil, J.R., 1973. Petrogenesia of the Super-

stition-superior volcanic area as inferred from strontium- and

oxygen-isotope studies. Geol. Sot. Am. Bull., 84: 1987-1998.

Taggart, J.E., Lindsay, J.R., Scott, B.A., Vivit, D.V., Bartel, A.J.

and Stewart, K.C., 1987. Analysis of geologic materials by

wavelength-dispersive X-ray fluorescence spectrometry. In:

P.A. Baedecker (Editor), Methods of Geochemical Analysis.

U.S. Geol. Surv. Bull., 1770: El-El9

Taylor Jr., H.P., 1968. The oxygen isotope geochemistry of

igneous rocks. Contrib. Mineral. Petrol., 19: l-71.

Taylor Jr., H.P., 1977. Water/rock interactions and the origin of

Hz0 in granitic batholiths. J. Geol. Sot. London, 133: 509-

558.

Taylor Jr., H.P., 1985. Igneous rocks: II. Isotopic case studies of

Circumpacific magmas. In: J.W. Valley, H.P.Taylor Jr. and

J.R. O’Neill (Editors), Stable Isotopes in High Temperature

Geological Processes. MSA Rev. Mineral., 16: 273-317

Taylor Jr., H.P., 1987. Comparison of hydrothermal systems in

layered gabbros and granites, and the origin of low-‘so

magmas. In: B.O. Mysen (Editor), Magmatic Processes:

Physicochemical Principles. Geochem. Sot. Spec. Publ.. 1:

337-357.

Taylor Jr., H.P., 1996. Oxygen and hydrogen isotope relationships

in hydrothermal mineral deposits. In: H.L. Barnes (Editor).

Geochemistry of Hydrothermal Ore Deposits. Wiley, New

York, NY (in press).

Taylor Jr.. H.P. and Epstein, S., 1962. Relationship between

o’s/0 I6 ratios in coexisting minerals in igneous and meta-

morphic rocks-Part 1. Principles and experimental results.

Geol. Sot. Am. Bull.. 73: 461-480.

Yager, D.B. and Quick, J.E., 1992. SUPERXAP Manual. U.S.

Geol. Surv., Open-File Rep. 92-l 3, 45 pp.

Zhang, Y., Stolper, E.M. and Wasserburg, G.J., 1991. Diffusion

of water in rhyolitic glasses. Geochim. Cosmochim. Acta, 55:

44 I -456.

Zies, E.G., 1924a. The fumarolic incrustations of the Valley of

Ten Thousand Smokes. Natl. Geogr. Sot. Contrib. Tech. Pap.,

Katmai Ser.. pp. 157-179:

Zies, E.G., 1924b. Hot Springs of the Valley of Ten Thousand

Smokes. J. Geol., 32: 303-310.

Zies, E.G., 1929. The Valley of Ten Thousand Smokes: I. The

fumarolic incrustations and their bearing on ore deposition. II.

The acid gases contributed to the sea during volcanic activity.

Natl. Geogr. Sot. Contrib. Tech. Pap., Katmai Ser., 4: l-79.

Oxygen isotopic and geochemical evidence for a short-lived, high-temperature hydrothermal event in...

Documents

Transcript of Oxygen isotopic and geochemical evidence for a short-lived, high-temperature hydrothermal event in...