Journal of Petrology Volume 41 Number 1 Pages 149-174 2000
© Oxford University Press 2000
Rhyolite Thermobarometry and the Shallowing of the Magma Reservoir, Coso Volcanic Field, California
1DEPARTMENT OF GEOLOGY, UNIVERSITY OF NORTH CAROLINA, CHAPEL HILL, NC 27599-3315, USA
2US GEOLOGICAL SURVEY, 345 MIDDLEFIELD ROAD MS-910, MENLO PARK, CA 94025-3591, USA
Received June 17, 1998; Revised typescript accepted July 5, 1999
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
The compositionally bimodal Pleistocene Coso volcanic field is located at the western margin of the Basin and Range province
60 km north of the Garlock fault. Thirty-nine nearly aphyric high-silica rhyolite domes were emplaced in the past million years: one at 1 Ma from a transient magma reservoir, one at 0·6 Ma, and the rest since 0·3 Ma. Over the past 0·6 My, the depth from which the rhyolites erupted has decreased and their temperatures have become slightly higher. Pre-eruptive conditions of the rhyolite magmas, calculated from phenocryst compositions using the two-oxide thermometer and the Al-in-hornblende barometer, ranged from 740°C and 270 MPa (2·7 kbar; 10 km depth) for the 0·6 Ma magma, to 770°C and 140 MPa (1·4 kbar; 5·5 km) for the youngest (0·04 Ma) magma. Results are consistent with either a single rhyolitic reservoir moving upward through the crust, or a series of successively shallower reservoirs. As the reservoir has become closer to the surface, eruptions have become both more frequent and more voluminous. KEY WORDS: Al-in-hornblende; caldera; eruption; geothermal; rhyolite
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
Many large-volume, shallow crustal magma reservoirs have extended histories that may culminate in caldera-forming pyroclastic eruptions. Changes over geologic time in the depth, volume, and magmatic temperature of these systems are poorly known and are not amenable to study by geophysical methods. Silicic lava domes fed from such reservoirs can yield partial records of the early histories of their magmatic systems (e.g. Long Valley Glass Mountain, Metz & Mahood, 1991
), and they potentially provide insight into magma chamber growth processes before caldera collapse. The Coso volcanic field of eastern California is such a system in which phenocryst thermobarometry of highly differentiated rhyolite provides the opportunity to determine the depth and thermal history of the silicic magma reservoir(s) over a span of nearly 0·6 My. Although we cannot know whether the Coso volcanic field will eventually produce a large-volume eruption, the processes of accumulation of rhyolitic magma and its apparent migration to shallow depth are believed to be similar to the early stages of development of a Long Valley-like system.
The Pliocene–Pleistocene Coso volcanic field (Fig. 1) is located at the western margin of the Basin and Range province just east of the Sierra Nevada and 60 km north of the Garlock fault. About 39 high-silica rhyolite domes and lava flows (hereafter simply termed domes), between 70 and 170 m thick and totaling 1·6 km3, were emplaced in the western part of the field in the past million years, all but two since 0·3 Ma (Duffield et al., 1980). Explosive eruptions that preceded extrusion of many of the domes produced 0·3 km3 of pyroclastic deposits that form partial to complete rings around some domes (Duffield et al., 1980; Bacon et al., 1981). These deposits blanket the pre-eruptive topography and mantle some of the older domes. The dome field is flanked on three sides by monogenetic basaltic volcanoes, and several rhyolite domes carry enclaves of hybrid basaltic andesite (Bacon & Metz, 1984). An active geothermal system centered within the rhyolite field sustains at present four electrical generating stations built during the 1980s and 1990s. These stations produce a total of 260 MW of electricity (Duffield et al., 1994).
|
The Coso magmatic system appears to be maintained by intrusion of basaltic magmas facilitated, in part, by continuing deformation of the region, which lies in a transition region between the Basin and Range province (undergoing approximately WNW–ESE extension) and the San Andreas right-slip province. The tectonic situation has led to block faulting and intense seismicity (Weaver & Hill, 1979; Roquemore, 1980). Extension favors intrusion of basalt into the crust, providing the heat and mass flux to sustain a long-lived (4 My) locus of magmatism (Bacon et al., 1981; Bacon, 1982; Novak & Bacon, 1986). New Sr and Nd isotopic analyses (Miller et al., 1996) indicate that much of the mass in the Coso rhyolites is derived from mantle basalts, either by partial melting of underplated basalts or by differentiation of mantle melts. The highly evolved character of the major and trace element signatures of the rhyolites reflects significant crystal fractionation from the parent basalt or more-evolved intermediate magmas, with concomitant but relatively minor assimilation of upper crust (Miller et al., 1996; J. S. Miller, in preparation).
The volcanic field is the product of a small magmatic system that has neither produced ignimbrites nor undergone caldera collapse; the roof of its magma reservoir has thus remained intact throughout its active history. Dome extrusions over the past 1 My provide information—through bulk compositions, mineral assemblages and mineral compositions, and magmatic enclaves—about the internal state of the magmatic system at the times of the eruptions. Phase assemblages necessary for thermometry (two Fe–Ti oxides; two feldspars; hornblende and plagioclase) and Al-in-hornblende barometry (hornblende and at least seven other minerals, plus free water vapor) are present in domes of ages that span that of the system’s lifetime, allowing the thermal and depth history of the top of the Coso rhyolitic magma reservoir to be estimated.
|
COMPOSITION AND MINERALOGY |
|---|
|
TOPABSTRACTINTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
Lava compositions and dome groups
All the Pleistocene Coso domes consist of high-silica rhyolite (Bacon et al., 1981
; Macdonald et al., 1992) with a rather uniform major element composition, but with differing minor and trace element contents. Bacon et al. (1981) used trace element contents and K–Ar ages to divide the 39 domes into seven groups (Fig. 2) they thought reflected the eruption of discrete batches of magma. All domes fed by an individual magma batch did not necessarily erupt simultaneously but may have been tapped over some longer period of time. Bacon et al. (1981) assigned weighted mean ages to each of the dome groups (Fig. 2) based on the dates of individual domes (Fig. 3).
|
|
|
Mineralogy
Except for three domes, including the two oldest ones, the Coso rhyolites are nearly aphyric. Typical crystal contents range from 1 wt % to <0·001 wt % (Bacon et al., 1981). Thin sections are almost useless to observe and study the minerals in such rocks. Grain mounts must be used, but at the expense of knowing the detailed petrographic relationships among grains. This is especially significant in thermobarometry; without contact relations, equilibrium among grains must be assessed compositionally and thermodynamically. The mineral assemblage table (Fig. 3) shows the observed mineral phases in each dome and indicates whether each is thought to be present as phenocrysts or xenocrysts or both.
Although the Coso rhyolites are nearly aphyric, they contain a great number of mineral phases that appear to be phenocrysts, plus up to seven that probably are xenocrysts. At Coso, sources of xenocrysts include wall rock around the reservoir and the conduit walls, as well as andesitic enclaves found in several of the domes. The enclaves are thought to have formed by hybridization between silicic and underlying basaltic magmas; these hybrids were subsequently dispersed into the rhyolitic magma (Bacon & Metz, 1984). Multiple populations of ferromagnesian phases are common, and it is not always obvious which were in equilibrium and which were not.
Even for those phases not thought to be xenocrysts, recent studies have suggested how little we may know about the nucleation, growth, and subsequent history of the crystals commonly called phenocrysts in nearly aphyric high-silica rhyolites. Evidence from laboratory experiments and from other volcanic systems indicates that crystallization may preferentially occur at the walls of the reservoir, where heat loss to the country rock enhances the thermal gradient (Chen & Turner, 1980; Turner & Gustafson, 1981). Crystals may grow partly attached to the reservoir’s wall; the liquid not incorporated into the crystals becomes enriched in H2O and other incompatible species, and its lower density allows it to rise along the boundary layer (Turner & Gustafson, 1981) and accumulate at the top of the reservoir. As the magma continues to cool in this new location, additional crystals may nucleate and grow (Manley, 1996). An ideal eruption would bring to the surface a rhyolite liquid with a small population of euhedral crystals. Complications arise when eruptive stirring mixes in some ’phenocrysts’ that grew near the reservoir margins (Nakada et al., 1994; Wolff et al., 1999); although such crystals may not have actually grown within the liquid enclosing them in the erupted rock, they grew from it when it was in the boundary layer near the reservoir wall. The Coso rhyolites represent highly fractionated yet crystal-poor magmas that must have separated from a more crystal-rich environment. The magma’s content of dissolved water, and the number of phases with which it is saturated, can thus be unrelated to the magma’s actual crystal content.
Salic phases
Quartz occurs in 16 of the 39 domes. The oldest dome (group 1) and the group 5 domes appear to contain no quartz, but quartz has been found in one or more domes in each of the other dome groups. Unzoned sanidine phenocrysts (Table 1) are present in 29 of the 39 domes, and in all the dome groups. Compositions of sanidine phenocrysts from domes of groups 2–7 fall in the narrow range Or55 to Or59. Sanidine found in the oldest dome (group 1) is Or67. Xenocrystic potassium feldspar grains with compositions that range from Or91 to Or93 are found in three domes from groups 3, 6, and 7. Unzoned plagioclase (Table 1) is found in all but one of the 39 domes; that dome has 0·001 wt % crystals. Plagioclase grains that are clearly phenocrysts have compositions from An7·7 to An10, but an absolute cut-off between phenocryst and xenocryst compositions is not apparent. Most xenocrysts have compositions An26 to An79, but grains of An13, An14, An15, An18, An19, and An22 are present, and these are probably xenocrysts also.
|
Mafic phases
Ilmenite and/or magnetite (Table 2) were found in 35 domes from all dome groups. Hornblende (Table 3) was found in 32 domes from all dome groups. Apparent phenocrysts have mg-numbers [100xMg/(Mg + Fe)] that range from five to 17 or possibly 39. Probable xenocrysts have mg-numbers from 45 to 77. Pyroxene (Table 4) was found in 29 domes from all dome groups, although the oldest dome appears to have only xenocrystic clinopyroxene. Pyroxene mg-numbers increase with time, but this is not true of hornblende or biotite. Orthopyroxene phenocrysts have mg-numbers that range from 19 to 32; probable xenocrysts have mg-numbers from 29 to 51. Clinopyroxene phenocrysts have mg-numbers from 30 to 48; probable xenocrysts have mg-numbers from 64 to 77. Fourteen domes appear to have pyroxene phenocrysts, and 24 domes display pyroxene xenocrysts. Biotite (Table 5) is found in 29 domes of all dome groups. The biotite is probably phenocrystic; biotite mg-numbers range from 24 to 56. Olivine (Table 6) is found in 20 domes, of all dome groups except group 1. Apparent olivine phenocrysts have mg-numbers that range from four to 20; probable xenocrysts have mg-numbers from 27 to 44.
|
|
|
|
|
Accessory minerals
Allanite is present in grain mounts of 15 of the 39 domes, only from dome groups 3, 6, and 7. Discrete apatite crystals have been identified with certainty in grain mounts from six of the domes and tentatively in 20 other domes, from all dome groups except groups 1 and 2. Apatite needles are common inclusions in phenocrysts from nearly all dome groups. Titanite occurs in grain mounts of four domes, of groups 1, 3, and 7. Zircon crystals have been identified with certainty from 14 of the domes (groups 1, 3, 4, 6, and 7) and tentatively in 11 other domes, including those of group 5. Zircon inclusions are common in phenocrysts in nearly all dome groups. Xenocrystic Cr-spinel has been found in only two of the domes, both from dome group 7.
|
THERMOBAROMETRY |
|---|
|
TOPABSTRACTINTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
Analytical methods
Pieces of obsidian or pumice weighing several kilograms were crushed, and minerals separated from the glassy matrix by magnetic and heavy liquid techniques. The resultant mineral separates were weighed to estimate crystal content of the domes (Fig. 3). Compositions of phenocrysts and xenocrysts were determined by wavelength-dispersive electron microprobe analysis of polished grain mounts. Hornblende analyses were performed by C.R.M. on a four-spectrometer Cameca MBX instrument at Duke University using a Kakanui kaersutite standard and 15 nA and 15 kV analytical conditions. Other phases were analyzed by C.R.B. at the US Geological Survey, Menlo Park, CA, with an ARL EMX instrument under analytical conditions described by Bacon & Duffield (1981)
. Reported mineral compositions are averages of two or more points per grain, or are averages of two or more grains, with two or more analyzed points on each. In cases of zoned grains, rim and core compositions are averaged separately.
Two-oxide temperatures and oxygen fugacities
Compositions of apparently equilibrated, coexisting Fe–Ti oxide minerals were used in the QUILF 4.1 software of Andersen et al. (1993) to determine equilibrium temperatures and estimate the oxygen fugacity of the rhyolitic magmas (Fig. 4). QUILF assesses equilibria among oxide minerals, pyroxenes, olivine, and quartz, and uses improvements to previous solution models by Lindsley and coworkers (Lindsey & Andersen, 1983; Andersen & Lindsey, 1988; Davidson & Lindsley, 1989; Andersen et al., 1991). All oxide mineral pairs used for temperature calculation pass the Mg/Mn partitioning test of Bacon & Hirschmann (1988). Because Fe–Ti oxides are thought to re-equilibrate more rapidly than do silicates after a change in P–T–X conditions (Gardner et al., 1995), they seem the most likely ’phenocryst’ phases to record pressure and temperature conditions before eruption (Buddington & Lindsley, 1964; Frost et al., 1988). Nonetheless, only seven Coso domes contained Fe–Ti oxide pairs that pass the partitioning test (Fig. 4); many oxide grains may be xenocrysts that were stirred into the magma so soon before the eruption that re-equilibration was not possible (Nakamura, 1995).
|
QUILF assesses how close the input mineral compositions are to equilibrium, and we used the software in the following manner. Two-oxide temperatures were calculated without assuming any other phases (i.e. pyroxene or olivine) were in equilibrium with the oxides. Compositions of apparently equilibrated, coexisting Fe–Ti oxide minerals were input as electron microprobe major oxide analyses. The software recasts these analyses to four compositional parameters for each phase and calculates temperatures based on the solution model of Andersen et al. (1991). Temperatures and oxygen fugacities were first calculated from the input compositions, then magnetite NMg (number of Mg atoms per formula unit) was allowed to vary. In almost every case, varying NMg lowered the mathematical uncertainties in the fit. In some cases, the adjustment in NMg was excessive, and the reported temperature was calculated from the original composition. In most cases, temperatures calculated with and without adjusting NMg vary less than the stated uncertainty of the model (about ±25°C; Andersen et al., 1993).
Two-feldspar temperatures
Sanidine and plagioclase phenocrysts apparently coexisted in equilibrium in many of the Coso rhyolite magmas. Equilibration temperatures were determined with the SOLVCALC 1.0 software of Wen & Nekvasil (1994) using the feldspar site mixing model of Fuhrman & Lindsley (1988), which has stated uncertainties of ±30°C. Like the MTHERM3 software of Fuhrman & Lindsley (1988), SOLVCALC varies the input compositions of the feldspars to achieve better fits. To calculate two-feldspar temperatures, one must assume a value for pressure. Because pressure and temperature are directly proportional, calculated temperatures should increase smoothly with pressure (although the rate of temperature increase with pressure is model dependent). In practice, however, the software’s adjustments of composition lead to jumps (to both higher and lower values) in the calculated temperatures. To produce a smooth temperature plot as a function of pressure, the following procedure was used. Temperatures were calculated for pressures between 50 and 1000 MPa (0·5 and 10 kbar) at a compositional uncertainty of 0·020 (molar endmember composition). For this range of pressures, temperature typically varied by 100°C. If temperature did not vary smoothly, successively smaller uncertainties were specified, leading to smaller adjustments in composition and less variability in temperature. The result is a generally smooth curve with temperature rising with pressure at about 7–11°C per 100 MPa (per 1 kbar).
Two-feldspar temperatures [referenced to 200 MPa (2 kbar)] range from 680 to 775°C, save for the oldest dome at 630°C. Two-feldspar temperatures show no correlation with dome age or group; this further suggests that some or much of the feldspar may not be strictly ’phenocrystic’. At present, we consider the two-oxide thermometer in QUILF more reliable for these rocks than any of the two-feldspar thermometers.
Two-pyroxene temperatures
Temperatures were calculated for several domes using the two-pyroxene thermometer in the QUILF 4.1 software of Andersen et al. (1993). Uncertainties are ±30°C (Frost & Lindsley, 1992). These temperatures are nearly always higher than the other temperature determinations, often unrealistically so (up to 900°C) with respect to the mineralogy. In domes with multiple populations of pyroxenes, temperature determinations using pyroxene pairs with high mg-numbers are 40–100°C higher than those using the low mg-number pyroxene pairs; the high mg-number pairs apparently grew in a more mafic, higher-temperature magma.
Al-in-hornblende pressures
We used the Anderson & Smith (1995) T-dependent formulation of the Al-in-hornblende barometer. The original Al-in-hornblende barometer (Hammarstrom & Zen, 1986) was formulated for calc-alkalic plutons with near-solidus phase assemblages of seven minerals. Johnson & Rutherford (1989) revised the barometer on the basis of laboratory experiments and extended it to include silicic volcanic rocks as well as plutonic ones. Application to high-silica rhyolites is often complicated by the fact that although a ’near-solidus’ mineral assemblage may exist, the rock is so nearly aphyric that some rarer components of the assemblage will probably not be observed even if a larger than normal sample is crushed and separated for mineral grains.
The equilibrium phase assemblage quartz + sanidine + plagioclase + hornblende + biotite + magnetite or ilmenite + titanite + melt + fluid is required by the Al-in-hornblende barometer. Two Fe–Ti oxide minerals can take the place of one oxide and titanite, and the Ti contents of the melt and hornblende will still be buffered (Johnson & Rutherford, 1989).
Pre-eruptive H2O contents and the presence of a vapor phase
A free vapor phase is one of the requirements of the Al-in-hornblende barometer. Evidence from melt inclusions suggests that magmas of at least three dome groups were saturated with an H2O-rich fluid. Four pristine, quartz-hosted melt inclusions from the Coso rhyolite samples have been analyzed for pre-eruptive H2O and CO2 contents by Fourier transform infrared spectroscopy (FTIR) (Blouke, 1993; Newman et al., 1993). CO2 was below the 100 ppm detection limit for all the inclusions and bubbles (S. Newman, personal communication, 1998). One inclusion from dome 10 (group 3) contained 6·2 wt % total H2O, which would be saturated at lithostatic pressures less than 270 MPa [2·7 kbar; program of Holloway & Blank (1994), which utilizes the Burnham model for H2O and the Stolper model for CO2 solubilities]. Two inclusions from the relatively porphyritic (2% crystals) dome 5 (group 4) contained 4·5 and 5·2 wt % total H2O [saturated at pressures less than 140–180 MPa (1·4–1·8 kbar)], and both also contained large bubbles equal to 7–10% of the inclusion volume. An inclusion from dome 4 (group 7) contained 6·4 wt % total H2O [saturated at pressures <290 MPa (2·9 kbar)] and a large bubble. The bubbles showed significant water but no detectable CO2 (Blouke, 1993). Melt inclusions from dome 28 (group 2) were also examined; many are crystalline, but of those still glassy, some contain bubbles that make up as much as 10% of the inclusion’s volume. The large bubble volumes observed in the samples from group 2, 4, and 7 domes, compared with shrinkage bubbles of only a few volume percent (Roedder, 1979), imply that a free vapor phase was probably present during trapping of the melt inclusions.
Circumstantial evidence that a vapor phase is now present in the Coso magma reservoir was provided by the 1992 Landers 7·3 Mw earthquake in the Mojave Desert of California. This earthquake triggered long-lived, small-magnitude local seismicity at a number of active volcanic areas in the western USA, including Coso, Long Valley caldera, The Geysers, and Yellowstone caldera (Hill et al., 1993; Roquemore & Simila, 1994). Linde et al. (1994) suggested that the details of the long-lived seismicity and inflation at Long Valley are best explained by the shaking loose of existing vapor bubbles. These bubbles then rose (advected) through the magma reservoir and expanded because of the decreasing confining (lithostatic) pressure; this would have increased the pressure on the walls of the magma reservoir and/or increased the volume of the magma reservoir (Sahagian & Proussevitch, 1992). At Long Valley, the number of earthquakes triggered by the Landers event, and the amplitude of the deformation measured at stations inside and outside the caldera (Linde et al., 1994), imply that a significant volume of existing bubbles was involved, which indicates that the entire system contains a free vapor phase. On the basis of similarities in triggered seismicity at Coso and Long Valley, Roquemore & Simila (1994) concluded that the triggered seismicity beneath the Coso volcanic field also was due to advective overpressure, suggesting that the Coso magmatic system is at present vapor saturated.
Oxygen fugacity
The plutons used by Hammarstrom & Zen (1986) for their empirical barometer are known to have crystallized at fairly high fO2, close to that of the nickel–nickel oxide (NNO) buffer. The experiments performed by Johnson & Rutherford (1989) and Schmidt (1992), on which the present version of the barometer (Anderson & Smith, 1995) is based, were also buffered at NNO. Anderson & Smith (1995) discussed the effect of lower fO2 levels using data from a suite of plutons that crystallized at fO2 conditions below NNO and closer to FMQ (fayalite–magnetite–quartz). These are the Proterozoic anorogenic granites that occur in a broad swath running from the southwestern to the northeastern USA. These granites have the required mineral assemblage (although dissolved water contents are unknown) for applying the barometer, but their calculated Al-in-hornblende equilibrium pressures are both unreasonably large and much greater than pressures indicated from barometry on country rock in contact with the plutons. On that basis, Anderson & Smith (1995) suggested that silicic magmas with oxygen fugacity much below NNO are not amenable to Al-in-hornblende barometry. The bulk compositions of anorogenic granites (especially high total Fe) may also play an important role, as might the probable low H2O contents of their magmas. Although the Coso magmas apparently equilibrated at fO2 conditions at and just below FMQ they were low in temperature and in total Fe, unlike the anorogenic granites. The hornblende phenocrysts were apparently fully buffered (see below) in a system with elevated H2O contents at water-saturated conditions. A low oxygen fugacity by itself does not appear to preclude the conditions necessary for Al-in-hornblende barometry to be successful in these rocks.
Hornblende populations
The Coso rhyolites contain two populations of hornblende, one relatively rich in Fe (in dome groups 2, 3, and 7), the other relatively rich in Mg (all dome groups except 2). The two populations are seen most easily if plotted against Fe/(Fe + Mg), where a value of 0·6 separates them (Fig. 5). Several domes contain hornblende grains from both populations.
|
Other major element systematics are also different for the two populations. Covariation plots of all major elements against Fe/(Fe + Mg) show that most Fe-rich hornblendes have restricted compositional ranges (Fig. 5a–d), suggestive of compositional buffering by other phases, which is required for barometry. This is not seen in the Mg-rich hornblendes, where analyses of a number of hornblende grains from a single dome, and sometimes multiple analyses of a single grain, define linear trends showing significant variation in one or more elements (Fig. 5a). These grains were not buffered: they grew in a liquid the bulk composition of which was strongly controlled by the growth of other phases, such as feldspars. Many tie-lines are subparallel, reflecting control by the same phase or assemblage. This information also shows that the Mg-rich hornblendes did not grow in the Coso rhyolite magma, which varies little in major element composition across the dome groups (Duffield et al., 1980).
Back-scattered electron (BSE) imaging (Figs 6 and 7) reveals other differences between the populations. Mg-rich hornblendes can be zoned or unzoned, euhedral, embayed, or glomerocrystically intergrown with other phases; most are also moderately to intensely fractured. Fe-rich hornblendes are unzoned and euhedral; only a few are fractured; some have partially skeletal exterior morphologies (Fig. 7d) and/or irregular glass inclusions (Figs 6f, 7b and d) that imply, respectively, rapid growth (Roedder, 1979; Swanson & Fenn, 1986) and no extended residence time before eruption (Manley, 1996).
|
|
On the basis of the compositions, zoning features, and morphologies, the Mg-rich hornblendes are interpreted as xenocrystic to the rhyolite magma, probably syneruptively stirred-in from an andesitic or dacitic magma or from a relatively early crystallized mush (Nakada et al., 1994), presumably beneath the rhyolite magma in the reservoir. The Fe-rich hornblendes appear to be phenocrysts that grew in the rhyolite magma now represented by the Coso domes. That the compositionally restricted subset of Fe-rich hornblendes (Fig. 5d) are buffered phenocrysts is corroborated by the Al content of coexisting biotite (Table 5), which is about twice that in the hornblende. This relationship was also seen in experimental runs (Johnson & Rutherford, 1989) used to confirm and extend the Al-in-hornblende barometer (M. J. Rutherford, personal communication, 1998). In addition, the apparent late growth of the hornblende phenocrysts implies that the other observed phases were already present to buffer the hornblende during growth. The bimodality in Fe/(Fe + Mg) of the buffered Fe-rich hornblende phenocrysts, where a value of 0·83 separates the two groups (Fig. 5d), must reflect slight differences in the environment of crystal growth—perhaps near the reservoir wall vs floating free in the magma.
Hornblende–plagioclase temperatures
Temperatures were also calculated with the pressure-dependent hornblende–plagioclase thermometer of Blundy & Holland (1990) and Holland & Blundy (1994). In three cases, hornblende-plagioclase P–T trends lie at higher pressures and temperatures than indicated by the two-oxide and Al-in-hornblende methods, but in two other cases the results from all three methods converge within the uncertainties.
Intensive parameters by dome group
Individual two-oxide temperature determinations using the QUILF software range from 705 to 790°C, and oxygen fugacity values [referenced to 200 MPa (2 kbar) pressure] range from -14·7 to -16·9 log units—on and just below the FMQ buffer at these temperatures (Fig. 4). Bacon & Metz (1984) reported temperatures between 775 and 825°C, also at and just below FMQ, for a few oxide pairs from the Coso rhyolites, using the graphical method of Buddington & Lindsley (1964).
Pressures were calculated with the Al-in-hornblende barometer at temperatures calculated by two-oxide thermometry. These pressure estimates range from 140 to 270 MPa (1·4–2·7 kbar; 5·4–10·2 km depth) and decrease from the older to younger domes. Depths were calculated from the pressure values assuming a basement (granitic and metamorphic rock) density of 2700 kg/m3 (Plouff & Isherwood, 1980), which translates to 100 MPa (1 kbar) per 3·78 km depth.
Combining Al-in-hornblende pressures with calculated two-oxide temperatures yields estimated variations in pressure and temperature of the uppermost portion of the Coso rhyolite magma reservoir over the last 0·6 My. The magma tapped by any given eruption would have resided at or very near the reservoir roof, because the dome eruptions were low-energy, small-volume events, and draw-down of the magma in the reservoir would have been minimal. Thermobarometric determinations have the combined uncertainties of microprobe analyses and those inherent in the thermodynamic solution models (examples in Fig. 8). Uncertainties (accuracies) for individual pressure determinations are ±50 MPa (±0·5 kbar); uncertainties for temperature determinations from the two-oxide thermometer are ±25°C, and from both the two-feldspar and two-pyroxene thermometers are ±30°C. Relative temperature differences between individual domes are significantly less than these values, particularly for the Fe–Ti oxide thermometer, as the precision of the analyses can be as much as an order of magnitude better than the reported uncertainties in some well-constrained cases.
Dome Group 1; 1·04 ± 0·02 Ma
Dome 38 is the only group 1 dome; it is cut by faults, highly eroded, and displays no non-hydrated glass. Two-feldspar thermometry yields 628°C [referenced to 200 MPa (2 kbar)], which seems unreasonably low.
Dome Group 2; 0·59 ± 0·02 Ma
The sole group 2 dome is the Devils Kitchen dome (dome 28). It has the greatest crystal content of all the Coso domes (15%). QUILF two-oxide thermometry indicates a magma temperature of 740°C; at this temperature, Al-in-hornblende barometry indicates a pressure of 270 MPa (2·7 kbar), which corresponds to 10·2 km depth (Fig. 8a). Two-feldspar and two-pyroxene temperatures agree well with these P–T estimates; hornblende–plagioclase temperatures are consistent within reasonable limits of uncertainties.
Dome Group 3; 0·24 ± 0·03 Ma
Group 3 consists of 12 rhyolite domes. Domes 14 and 22 yield two-oxide temperatures of 743 and 746°C, respectively (Fig. 8b). Dome 22 contains phenocrystic hornblende which at 746°C gives a pressure of 270 MPa (2·7 kbar; 10·0 km); five other group 3 domes contain compositionally identical hornblende. At 270 MPa, two-feldspar temperatures from domes 13 and 14 agree fairly well with the oxide temperatures, but feldspars from dome 22 give much lower temperatures. Hornblende–plagioclase temperatures for domes 13 and 22 are in reasonable agreement with other thermometers.
At 270 MPa and the pre-eruptive H2O content of 6·2 wt % (see above; Blouke, 1993; S. Newman, personal communication, 1998), the magma would have been saturated with an H2O-rich vapor phase with at least 0·20 wt % F and 0·10 wt % Cl (Macdonald et al., 1992) and less than 100 ppm CO2 (S. Newman, personal communication, 1998).
Dome Group 4; 0·17 ± 0·01 Ma
The three group 4 domes all contain xenocrystic hornblende. One dome contains both Fe–Ti oxides, but they do not pass the Mg/Mn partitioning test. Melt inclusions with large bubbles from dome 5 yield pre-eruptive H2O contents of 4·5 and 5·2 wt % and less than 100 ppm CO2 (see above; Blouke, 1993; S. Newman, personal communication, 1998), which indicates probable saturation at pressures less than 140–180 MPa (1·4–1·8 kbar; depths of 5·3–6·8 km). The pre-eruptive melt also contained at least 0·25 wt % F and 0·05 wt % Cl (Macdonald et al., 1992).
Dome Group 5; 0·16 ± 0·03 Ma
Group 5 consists of four domes and a lava flow; none of these contain phenocrystic hornblende or yield two-oxide or two-feldspar temperatures that are compatible with those of other dome groups.
Dome Group 6; 0·09 ± 0·01 Ma
There are three group 6 domes or flow–domes, but none of them contain phenocrystic hornblende, so pressure estimates are not possible. Domes 24 and 25 provide two-oxide temperatures of 764°C and 751°C, respectively (Fig. 8c). Two-pyroxene temperatures for these domes are higher than the two-oxide temperatures, and the trends do not intersect.
Dome Group 7; 0·06 ± 0·01 Ma
Group 7 consists of 14 domes, including Sugarloaf Mtn (dome 26), which is the largest dome and has yielded the youngest K–Ar date in the Coso volcanic field (although it is littered with tephra from a still-younger eruption; Duffield et al., 1980). At a two-oxide temperature of 768°C, hornblende barometry yields a pressure of 140 MPa (1·4 kbar), equivalent to 5·4 km (Fig. 8d). Two-feldspar, two-pyroxene, and hornblende–plagioclase thermometry agree well with the oxide temperature estimate at this pressure.
Although dome 21 contains hornblende with the same Al content as that in dome 26, and the hornblende–plagioclase temperature is similar, the two-feldspar temperature is 100°C lower than the Sugarloaf Mtn value.
At a pre-eruptive H2O content of 6·4 wt % and less than 100 ppm CO2 (see above; Blouke, 1993; S. Newman, personal communication, 1998), the group 7 magma would have been vapor saturated at pressures of 280 MPa (2·8 kbar) or less. Although this is consistent with vapor saturation at the pressure indicated by hornblende barometry, it appears to indicate crystallization of quartz and trapping of melt inclusions at a depth greater than that recorded by the hornblende. This would be a logical consequence of magma undergoing further crystallization as a result of moving to a shallower level in the crust (M. J. Rutherford, personal communication, 1998). The pre-eruptive melt also contained at least 0·13 wt % F (Macdonald et al., 1992).
|
EVOLUTION OF THE MAGMA RESERVOIR |
|---|
|
TOPABSTRACTINTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
Our thermobarometric results suggest that the top of the Coso rhyolite magma reservoir was at a depth of about 10 km between
0·6 and 0·3 Ma, and had risen to 5·4 km by about 0·04 Ma (Fig. 9). The 4·6 km difference in depth indicated by hornblende barometry is significantly greater than the uncertainty associated with the barometer (±1·9 km) and persists even if the temperature difference indicated by the two-oxide thermometer is less than the 30°C we calculate. This is because the hornblende Al contents differ so much between the early and late dome groups.
|
M
Present-day geophysical evidence is consistent with a 5·4 km depth to the top of the magma reservoir. The Coso hydrothermal system (the main source of microseismicity) lies within the upper 5 km of the subsurface beneath the geothermal area, where it enhances the attenuation of seismic and microseismic events (Young & Ward, 1980; Wu & Lees, 1996, 1999). Below 5 km, seismicity tapers off (Fig. 9; Walter & Weaver, 1980). In three dimensions, earthquake foci determined for several years (J. M. Lees, personal communication, 1997) define a concave-downward region that lacks seismicity and is located roughly 1 km east to southeast of Sugarloaf Mtn, approximately at the center of the geothermal field (J. M. Lees, personal communication, 1997). A strong seismic reflector is present about 5–6 km beneath the central part of the geothermal field; this is probably the roof of the magma reservoir (P. Malin, personal communication, 1997). Data from teleseismic P waves, recorded by an array of seismograph stations within the Coso geothermal area, reveal an intense low-velocity body—consistent with a zone of partial melt—at depths between 5 and 20 km, centered beneath Devils Kitchen, where the present-day heat flow is highest (Reasenberg et al., 1980).
The apparent change in reservoir depth may reflect a single magma reservoir that moved either continuously or in discrete events upward through the crust. Alternatively, the change may reflect a series of reservoirs that were established at successively shallower depths. Because no basaltic magma younger than 0·6 Ma has vented through the center of the system (Fig. 2), it seems unlikely that the rhyolitic reservoir solidified between eruptive pulses. Surface heat flow (Combs, 1980) became elevated before 0·3 Ma, when a decrease in fluid discharge and an increase in the temperature (to approximately present-day levels) of the hydrothermal system occurred (M. Adams, personal communication, 1997). Before about 0·25 Ma, discharge of siliceous sinter-producing hydrothermal waters occurred along a north–south basement fault just east of the dome field (Duffield et al., 1980). Hydrothermal circulation upward along such faults would be the most efficient means for transfer of heat from deep magmatic reservoirs to the surface (Combs, 1980; Duffield et al., 1980). The distinct Pb, Sr, and Nd isotopic compositions of the 1·04 and 0·6 Ma rhyolite domes (Bacon et al., 1984; Miller et al., 1996; J. S. Miller, in preparation) imply that each was fed by an independent and transitory reservoir. In contrast, the Pb isotopic ratios of domes erupted after about 0·3 Ma are similar (Bacon et al., 1984), which is consistent with a single reservoir source for all of them. An abundance of enclaves of mafic magma in the 0·6 Ma dome suggests that its reservoir was small in volume (Duffield et al., 1980; Bacon & Metz, 1984).
If a single, rising magma reservoir was present at Coso, the 4·5 km change in depth over 0·26 My (from 0·3 to 0·04 Ma) is equivalent to an average ascent rate of >1·5 cm/yr. Stoping of blocks from the reservoir roof may be the most likely way reservoir migration could occur as a continuing or intermittent process. Opening of progressively more shallow sills or other intrusions in the country rocks also would move the reservoir closer to the surface. Stretching of the roof of the magma reservoir by tectonic extension (Weaver & Hill, 1979; Roquemore, 1980) probably would be accommodated by dike injection (Bacon, 1982), perhaps focused in one or a few locations on the roof, contributing to upward migration of the reservoir.
New Sr and Nd isotopic data reveal additional details of the interaction of the rhyolitic magma and the surrounding crust. The Sr-poor (<10 ppm) Coso rhyolites show elevated 87Sr/86Sr ratios, indicating limited contamination of the magma reservoir by the surrounding country rock (J. S. Miller, in preparation). That this is not attributable to in situ decay of 87Rb is shown by an inverse relationship between 87Sr/86Sr and Rb/Sr (J. S. Miller, in preparation; Wolff et al., 1999). Evidence of contamination is also seen in the Nd data. Coso basalts have 
Nd values between +3·0 and +7·5, whereas local crustal values range from -12 to -6 (Miller et al., 1996). The 1·04 Ma dome has an 
Nd of -2·6, indicating significant upper-crustal assimilation during its petrogenesis; this is distinct from all subsequent domes. Average Nd values of the groups 2–7 rhyolites are nearly all lower than the basalt values, and decrease from a high of +3·6 at 0·6 Ma to +1·8 at 0·04 Ma (Miller et al., 1996; J. S. Miller, personal communication, 1998) indicating slight increases in the proportion of crustal Nd with time. Both the Sr and Nd contamination imply that the dome eruptions probably did tap the top of the reservoir, where magma would have been in contact with country rock comprising the reservoir roof (e.g. Duffield et al., 1995). For contamination to occur, the magma–country rock interface must not have been armored by the products of crystallization caused by cooling of a static chamber. The possible slight increase in magmatic temperature indicated by the Fe–Ti oxide thermometry further suggests that the reservoir was not simply cooling in place. Both stoping and intrusive rise along dikes are viable methods by which the magma reservoir could have moved upward through the crust.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
Evolution of Coso rhyolitic magmatism involved emplacement of two small domes over a period of
0·75 My, after which the frequency of eruptions increased. The single dome erupted at 1·04 Ma reflects a short-lived reservoir unrelated to the younger domes. The reservoir that fed the porphyritic 0·6 Ma dome was probably also short lived, but its location at the center of the geothermal area, and the subsequent high heat flow and absence of basaltic eruptions in the center of the rhyolite field, imply that it marked the onset of high temperatures, if not magma, beneath the area. Domes erupted at and after 0·3 Ma show uniform major and trace element contents and Pb and Nd isotopic compositions (Bacon et al., 1984; J. S. Miller, personal communication, 1998) that suggest the presence of a long-lived rhyolitic magma reservoir.
Several other changes occurred in the Coso rhyolite system since 0·6 Ma. Volumetric eruptive flux has increased, and magma temperatures may have increased slightly; the youngest dome group is the most voluminous of the magma batches and is possibly up to 30°C hotter than rhyolite erupted at 0·6 Ma. The bulk compositions of the dome group magmas also shifted slightly, with contents of Fe and other compatible trace elements generally increasing (Bacon et al., 1981). This may result from earlier tapping of the most evolved, coolest magma from near the reservoir top at rates faster than it could form, thus leading to venting of deeper, warmer, less evolved magma. Alternatively, an increase in the rate of basaltic magma input into the roots of the Coso system may be supplying heat faster than it can be lost to circulation in the hydrothermal system.
Perhaps most significantly, the top of the rhyolitic magma reservoir(s) apparently became shallower by roughly 4·5 km, from a depth of about 10 km between 0·6 and 0·3 Ma to about 5·5 km depth at 0·04 Ma. Future studies may show whether such shallowing is a common feature of pre-collapse silicic magmatic systems such as Coso or pre-Bishop Tuff Long Valley. As the reservoir becomes shallower, dike intrusion events will be more likely to breach the surface. The observed increase in rhyolite eruptive flux at Coso may result partly from this. Ultimately, shallowing of the reservoir also should be a major factor in the timing of failure of the reservoir roof and of caldera collapse. In some systems, reservoir rise may directly control when caldera collapse occurs.
|
ACKNOWLEDGEMENTS |
|---|
In different decades, Wendell Duffield, and Andy Sabin and Jonathan Miller helped with field work and logistics. Discussions with Allen Glazner, Jonathan Miller, Rodney Metcalf, Mike Adams, Jonathan Lees, and Peter Malin helped focus these ideas. Alan Boudreau and Donna Whitney offered advice on microprobe analysis and troubleshooting on the Duke University facility. Sally Newman graciously shared unpublished data on melt inclusion analyses. Recommendations about oxygen buffers from Ronald Frost are appreciated. We thank Frank Monastero and the Navy Geothermal Program Office at China Lake Naval Weapons Station for soliciting and facilitating the present study, and the US Navy and California Energy for allowing access to the geothermal area. Work through UNC was made possible by Navy Contract N68936-95-C-0398 to Allen F. Glazner.
|
FOOTNOTES |
|---|
*Corresponding author. Present address: 16170 NE 11th Street, Bellevue, WA 98008-3527, USA.Telephone: +1-425-401-8783. e-mail: crmanley{at}mindspring.com
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONCOMPOSITION AND MINERALOGYTHERMOBAROMETRYEVOLUTION OF THE MAGMA...CONCLUSIONSREFERENCES |
|---|
Andersen, D. J. & Lindsley, D. H. (1988). Internally consistent solution models for Fe–Mg–Mn–Ti oxides: Fe–Ti oxides. American Mineralogist 73, 714–726.[Abstract]
Andersen, D. J., Bishop, F. C. & Lindsley, D. H. (1991). Internally consistent solution models for Fe–Mg–Mn–Ti oxides: Part II. Fe–Mg–Ti oxides and olivine. American Mineralogist 76, 427–444.[Abstract]
Anderson, J. L. & Smith, D. R. (1995). The effects of temperature and fO2 on the Al-in-hornblende barometer. American Mineralogist 80, 549–559.[Abstract]
Andersen, D. J., Lindsley, D. H. & Davidson, P. M. (1993). QUILF: a PASCAL program to assess equilibria among Fe–Mg–Mn–Ti oxides, pyroxenes, olivine, and quartz. Computers and Geosciences 19, 1333–1350.
Bacon, C. R. (1982). Time-predictable bimodal volcanism in the Coso Range, California. Geology 10, 65–69.[Abstract]
Bacon, C. R. & Duffield, W. A. (1981). Late Cenozoic rhyolites from the Kern Plateau, southern Sierra Nevada, California. American Journal of Science 281, 1–34.
Bacon, C. R. & Hirschmann, M. M. (1988). Mg/Mn partitioning as a test for equilibrium between coexisting Fe–Ti oxides. American Mineralogist 73, 57–61.[Abstract]
Bacon, C. R. & Metz, J. (1984). Magmatic inclusions in rhyolites, contaminated basalts, and compositional zonation beneath the Coso volcanic field, California. Contributions to Mineralogy and Petrology 85, 346–365.[ISI]
Bacon, C. R., Macdonald, R., Smith, R. L. & Baedecker, P. A. (1981). Pleistocene high-silica rhyolites of the Coso volcanic field, Inyo County, California. Journal of Geophysical Research 86, 10223–10241.
Bacon, C. R., Kurasawa, H., Delevaux, M. H., Kistler, R. W. & Doe, B. R. (1984). Lead and strontium isotopic evidence for crustal interaction and compositional zonation in the source regions of Pleistocene basaltic and rhyolitic magmas of the Coso volcanic field, California. Contributions to Mineralogy and Petrology 85, 366–375.
Blouke, K. J. (1993). Volatile compositions of melt inclusions in Coso Range rhyolite (abstract). Geological Society of America, Abstracts with Programs 25, 11.
Blundy, J. D. & Holland, T. J. B. (1990). Calcic amphibole equilibria and a new amphibole–plagioclase geothermometer. Contributions to Mineralogy and Petrology 104, 208–224.
Buddington, A. F. & Lindsley, D. H. (1964). Iron–titanium oxide minerals and synthetic equivalents. Journal of Petrology 5, 310–357.
Chen, C. F. & Turner, J. S. (1980). Crystallization in a double-diffusive system. Journal of Geophysical Research 85, 2573–2593.[ISI]
Combs, J. (1980). Heat flow in the Coso geothermal area, Inyo County, California. Journal of Geophysical Research 85, 2411–2424.
Davidson, P. M. & Lindsey, D. H. (1989). Thermodynamic analysis of pyroxene–olivine–quartz equilibria in the system CaO–MgO–FeO–SiO2. American Mineralogist 74, 18–30.[Abstract]
Duffield, W. A. & Bacon, C. R. (1981). Geologic map of the Coso volcanic field and adjacent areas, Inyo County, California. US Geological Survey Miscellaneous Investigations Map I-1200.
Duffield, W. A., Bacon, C. R. & Dalrymple, G. B. (1980). Late Cenozoic volcanism, geochronology, and structure of the Coso Range, Inyo County, California. Journal of Geophysical Research 85, 2381–2404.
Duffield, W. A., Sass, J. H. & Sorey, M. L. (1994). Tapping the Earth’s Natural Heat. US Geological Survey Circular 1125, 63 pp.
Duffield, W. A., Ruiz, J. & Webster, J. D. (1995). Roof-rock contamination of magma along the top of the reservoir for the Bishop Tuff. Journal of Volcanology and Geothermal Research 69, 187–195.
Friedman, I. & Obradovich, J. (1981). Obsidian hydration dating of volcanic events. Quaternary Research 16, 37–47.
Frost, B. R. (1991). Introduction to oxygen fugacity and its petrologic importance. In: Lindsley, D. H. (ed.) Oxide Minerals: Petrologic and Magnetic Significance. Mineralogical Society of America, Reviews in Mineralogy 25, 1–9.
Frost, B. R. & Lindsley, D. H. (1992). Equilibria among Fe–Ti oxides, pyroxenes, olivine, and quartz: Part II. Application. American Mineralogist 77, 1004–1020.[Abstract]
Frost, B. R., Lindsley, D. H. & Andersen, D. J. (1988). Fe–Ti oxide–silicate equilibria: assemblages with fayalitic olivine. American Mineralogist 73, 727–740.[Abstract]
Fuhrman, M. L. & Lindsley, D. H. (1988). Ternary-feldspar modeling and thermometry. American Mineralogist 73, 201–215.[Abstract]