Ann. Rev. Earth Planet. Sci. 1989. 17:13-53

GEOCHEMISTRY AND DYNAMICS OF THE YELLOWSTONE NATIONAL PARK HYDROTHERMAL SYSTEM!

Robert O. Fournier

US Geological Survey, Menlo Park, California 94025

INTRODUCTION

The magmatic-hydrothermal system at Yellowstone National Park is unique among presently active systems in regard to its unparalleled geysers, tectonic environment, and magnitude of caldera-forming volcanic activity. However, over geologic time, systems like Yellowstone have been common and appear to have been responsible for the formation of many important basc-metal and precious-mctal ore deposits. For these reasons, Yellow­ stone has received a great amount of scientific attention. Geologic inves­ tigations in Yellowstone began with the Hague expedition in the late 1800s, and the results of the first chcmical analyses of thermal waters from the Park's many hot springs and geysers were published in 1888 by Gooch & Whitfield (1888). Since then, numerous investigators have studied the hydrothermal system and its geologic environmcnt, and data collectcd over a span of a hundred years are available for comparison and by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. interpretation. Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org

THE GEOLOGIC AND GEOPHYSICAL SETTING

The geologic setting and geophysical characteristics of the Yellowstone National Park region have been described by many investigators. Par­ ticular attention is called to summaries by Smith & Christiansen (1980),

1 The US Government has the right to retain a nonexclusive royalty-free license in and to any copyright covering this paper. 13 14 FOURNIER

Christiansen (1984), Smith & Braile (1984), Fournier & Pitt (1985), and the Yellowstone National Park Task Group (1987). Yellowstone Park is situated on a high volcanic plateau that sits astride a region of active major crustal extension. This plateau and the adjacent region to the west have been the most seismically active area of the Rocky Mountains in historic time (Smith & Braile 1984). According to Christiansen (1984), volcanism in the Yellowstone region has been active for the last 2.2 Myr, and the erupted lavas and tuffs have been predominantly rhyolitic and sub­ ordinately basaltic. Major caldera-forming eruptions of silicic pyro­ clastics occurred about 2.0, 1.3, and 0.6 Ma (Figure 1), and each of these was preceded and succeeded by large eruptions of rhyolitic lava. The last eruption of rhyolitic lava was about 70,000 yr B.P. (Christiansen 1984). A variety of geophysical anomalies coincide with the 0.6-Ma Yellow­ stone caldera that, taken together, indicate that the magmatic system at Yellowstone is still vigorous and that temperatures in excess of 350°C exist at very shallow depths. These anomalies include low densities (Blank & Gettings 1974), large convective and conductive heat flows (Fournier et al 1976, Morgan et al 1977, White 1978), a magnetic low and shallow cal­ culated Curie-point isotherms (Smith et al 1974, Bhattacharyya & Leu 1975, Eaton et al 1975), low seismic velocities (Iyer et al 1981, Daniel & Boore 1982);high seismic attenuation (Benz & Smith 1984, Brokaw 1985), a sharp increase in electrical conductivity at a depth of about 5 km, shown by magnetotelluric soundings (Stanley et al 1977), and a lack of seismic focal depths deeper than about 3-4 km beneath much of the Yellowstone caldera (Fournier & Pitt 1985). In addition, there has been a spatial and temporal pattern of historical deformation-inflation followed by deflation(Pelton & Smith 1979, 1982, Jackson et a11984, Hamilton 1985, Dzurisin et al 1986, Meyer & Lock 1986, Dzurisin & Yamashita 1987)­ that is difficult to explain by processes that do not involve magmatic activity. The combined geologic and geophysical data support a model in which

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. magma or partially molten material underlies much of the O.6-Ma Yel­

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org lowstone caldera at depths ranging from about 5 to 10 km, and possibly as shallow as 3 km beneath portions of the eastern half of the caldera (Eaton et a1 1975, Iyer et a1 1981, Lehman et a1 1982, Smith & Braile 1984, Benz & Smith 1984, Brokaw 1985). A sharply bounded sei,smic low-velocity zone in the western part of the Park was reported by Lehman et al (1982) that would have been compatible with the presence of partially molten rock at a depth of 3-4 km in a N-S belt just to the west of Upper Geyser Basin and extending to Shoshone Geyser Basin. However, Lehman et al (1982) cautioned that this low-velocity zone might be fictive, the result of YNP HYDROTHERMAL SYSTEM 15

\ '-\.,"'- , ( '., ,-�. -"-. '/ '-- MH����------/ ''-. \ . .1 • 'TB ) f.q,. \'-" "- .�... a: I \ )CS ',\ >' . <'9 ; � \. \ � , I \. '\. (}�B \ \.,--. ./ . ..:) r�1.....1"\ . MJ._. ,f' I -, I , .r' J__ ,.-' ... \

10 20 30 40 50km

Figure 1 Locations of major rivers draining Yellowstone National Park (dot-dashed lines) and calderas of the Yellowstone Plateau volcanic field: northeast-slanting lines within the 2.0-Ma first-cycle caldera, northwest-slanting lines within the 1.3-Ma second-cycle caldera, and horizontal lines within the O.6-Ma third-cycle caldera. Crosshatched areas show regions of caldera overlap. Letter codes show river names and approximate locations of places mentioned in text: BC, Boundary Creek; CH, Crater Hills; CS, Calcite Springs; FAR, Falls River; FHR, Firehole River; GAR, Gardner River; GBR, Gibbon River; GC, Grand Canyon; HL, Heart Lake; HLB, Heart Lake Basin; HSB, Hot Springs Basin; JC, Josephs Coat Hot Springs; LB, Lower Basin; LL, Lewis Lake; LR, Lamar River; MB, Midway Basin; MH, Mammoth Hot Springs; MJ, Madison Junction; MR, Madison River; MV, Mud Volcano; NB, Norris Basin; RS, Rainbow Springs; SB, Shoshone Basin; SJ, Smoke Jumper;

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. SL, Shoshone Lake; SR, Snake River; TB, Tower Bridge; UB, Upper Basin; WS, Washburn Hot Springs; WT, West Thumb; YL, Yellowstone Lake; YR, Yellowstone River. Outline Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org of Yellowstone National Park shown for reference. Figure revised and redrawn after Christiansen (1984).

an attenuation problem in which very weak firstarrivals were not detected. Subsequent seismic experiments with better recording-station coverage have shown that, indeed, there was an attenuation problem with the earlier data, and that a 4.0 km s -I low-velocity zone does not exist at a shallow level beneath the western part of the Park (Brokaw 1985). 16 FOURNIER

DISTRIBUTION AND TYPES OF HYDROTHERMAL ACTIVITY

The distribution of active and recently active hydrothermal features in Yellowstone is shown in Figure 2. They are mostly within the O.6-Ma Yellowstone caldera, close to the outer edge of the main ring fracture zone or at the margins of two resurgent domes. Tn addition, pronounced hydrothermal activity occurs in a linear band along a system of faults extending northward from Norris Geyser Basin through Mammoth Hot Springs. by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org

o 10 20 3.0 40 50 km �'----�'----'�--�.�---'�--�'

Figure 2 Map of Yellowstone National Park, Wyoming, USA, showing O.6-Ma Yel­ lowstone caldera, two resurgent domes, active and recently active hydrothermal features (solid black), major faults (solid lines outside caldera region), and locations of selected named thermal areas (key to place names given in Figure 1 legend). Figure redrawn after Christiansen (1984). YNP HYDROTHERMAL SYSTEM 17 Representative analyses of thermal waters from major hydrothermal localities in Yellowstone are shown in Table 1. Waters from many of the Yellowstone hot springs and geysers were collected and analyzed in the late 1800s (Gooch & Whitfield 1888), and a later exhaustive study was carried out in 1925-30 by Allen & Day (1935). Investigators who have collected and analyzed the thermal waters and gases since 1959 include Noguchi & Nix (1963), Scott (1964), Mazor & Wasserburg (1965), Gunter & Musgrave (1966), Schoen & Rye (1970), Muffler et al (1971), Gunter (1973), Mazor & Fournier (1973), Rowe et al (1973), Thompson et al (1975), Truesdell et al (1977), Craig et al (1978), Kim & Craig (1979), Thompson & Yadav (1979), Kennedy et al (1982, 1985, 1987), Mazor & Thompson (1982), and Truesdell & Thompson (1982). Although changes in the behavior of individual hot springs and geysers are common within the Park, there has been no significant change in the range of water compositions exhibited by the hot springs and geysers for at least a century, and probably much longer. Most of the hydrothermal activity in the western half of the Park is typical of hot-water systems in which fluid pressures steadily increase downward and the maximum temperature attainable at a given depth is given by a boiling-point curve based on hydrostatic conditions. All of the major geyser basins in Yellowstone Park, except Norris and Heart Lake, lie within the portion of the 0.6-Ma Yellowstone caldera that overlaps the 2.0-Ma caldera (Figure 1, horizontal line pattern overlaying northeast line pattern). Rhyolite flows precede the ash-flow eruptions that formed the 2.0-Ma caldera, and it is likely that other rhyolite flows partly filled this caldera prior to the ash-flow eruptions that formed the 0.6-Ma caldera. Therefore, within the portion of the 0.6-Ma caldera that overlaps the 2.0- Ma caldera, the general sequence of volcanic rocks is likely to be precal­ dera rhyolite flows, relatively impermeable 2.0-Ma ash-flow tuff, per­ meable rhyolite flows, relatively impermeable 0.6-Ma ash-flow tuff, and permeable rhyolite flows. Thus, relatively permeable rhyolite flows are present at various depths that can serve as fluid reservoirs within rela­

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. tively impermeable ash-flow tuff.

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org In the eastern part of the O.6-Ma caldera, where it does not overlap the 2.0-Ma caldera, most of the caldera fill is likely to be relatively impermeable 2.0-Ma and 0.6-Ma ash-flow tuffs with little interlayered permeable rhyolite flow material, although one major rhyolite flow caps some of the ash-flow tuff. In this part of the caldera, vapor-dominated hydrothermal activity is prevalent in which fluid pressures remain nearly constant through a significant depth interval (White et al 1971). This is also the part of the Park in which magma or partly molten material may be closest to the surface (Lehman et al 1982). 00

Table 1 Chemical analyses of selected thermal waters from Yellowstone National Park "r1 0 e Number 3 5 10 ::<:I § Locality Mammoth West Shoshone Heart Lower Upper Upper Grand Norris Norris ::<:I Hot Springs Thumb Basin Lake Basin Basin Basin Basin Canyon Basin Basin Name Y-IO Lakeshore Washtub Unnamed Ojo Punch Ear Unnamed Porcelain Cinder drill hole Geyser Spring Caliente Bowl Spring Sevenmile Terrace Pool Hole Sample No. Y-IO 17484 T7214 T7348 17560 17836 17956 17615 17528 17908 Date 09/13/69 10/09/74 09/72 09/73 09/29/76 Temp. eC) 70 90 81 93.5 95 94 94 91 93.5 92 pH 7.48 7.76 9.00 9.48 1-74 8.33 8.49 8.61 3.57 SiO, (mg kg-I) 88 328 366 230 312 371 534 654 329 Al 0.01 0.14 0.06 Fe 2.4 0.05 Ca 450 2.0 0.4 0.9 0_75 0.67 0.82 0.5 2.12 6.3 Mg 80 0.51 0.05 0.01 0_01 0.02 0.01 0.Q2 0.03 0.12 Na 161 408 365 382 330 420 319 366 404 346 K 69 20 16 21 10 17 27 58 81 81 Li 1.8 3.24 1.0 6.6 3.6 3.8 4.9 3.45 5.8 3.9 NH, 1.0 0.1 HeO, 997 531 406 306 246 590 146 223 47 SO, 800 55 48 100 26 19 19 85 31 147 CI 171 261 328 365 326 289 415 427 669 569 F 4.2 14.5 25.5 36 30 28 24 12.8 5.8 6 B 3.8 3.3 2.8 2.4 3-8 3.8 3.6 16 9.9

Reference' 2 3 2 4 2 2 by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org Table 1 (Continued)

Number 11 [2 13 14 15 16 17 18 19 20

Locality Norris Crater Upper Hot Springs Josephs Coat Josephs Coat Upper Boundary Washburn Washburn Basin Hills Basin Basin Hot Springs Hot Springs Basin Creek Hot Springs Hot Springs Name Echinus Crater Hills Iron Unnamed Unnamed Unnamed Hillside Unnamed Unnamed Unnamed Geyser Geyser Spring group Sample No. 17846 17804 YF448 YF451 YF452 T9-10 17930 17304 YF429 Date 09/28/78 1920s 06/22/69 06/26/69 06/26/69 09/22/73 06/22/69 Temp. Cc) 93 88 9[ 89 86 94 83 92 91 86 pH 3.2 3.18 3.7 2.66 1.82 9.38 8.62 7.94 8.00 SiO, (mg kg-I) 278 676 365 3[7 333 238 170 21() 247 AI 0.60 0.13 0.2 Fe Ca 4.9 5.64 37.1 4.1 3.4 7.9 4.1 2 17.2 Mg 0.52 1.04 Trace 3.75 0.39 0.01 0.27 0.10 4.10 9.3 Na 166 640 77 187 109 150 18[ 9.7 27.1 i '1i K 50 133 28 16 17.1 19 7.7 1 1.2 6.5 13.7 :t: Li 0.7 6.9 0.04 0.01 0.[8 0.83 1.2 0.1 0.1 >< NH, 4 171 57.3 22.4 0.1 0.51 270 658 0 :;:J HC03 () () 0 0 0 335 251 259 107 8.2 0 >-l SO, 337 566 23[ 1530 1830 24 [4.2 10 900 1950 CI 108 890 0.1 0.1 5.4 72 107 51 :;c F 5.6 27.5 1.2 0.8 3.7 12 19.1 0.1 0.5 � B 3.0 20 2.2 5.05 0.07 0.36 1.17 1.0 6.6 7.84 :> t"' VJ Referencea 4- 4 6 VJ>< >-l rI1 • References: � (I) UnpUbliShed data, USGS (R. Barnes, analyst) (4) Unpublished data, USGS (J. M. Thompson, analyst) (2) Thompson & Yadav (1979) (5) Allen & Day (1935)

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. (3) Thompson et al (1975) (6) Thompson & Hutchinson (1980) ...... '-0 Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org 20 FOURNIER

Hot- Water Systems

The typical surface expression of a high-temperature hot-water system is boiling hot springs, possibly with geyser activity, that discharge significant quantities of neutral to slightly alkaline water. The famous geyser basins ofYellowstone National Park are probably the best examples in the world of this type of hydrothermal activity. These include Upper, Midway, Lower, West Thumb, and Shoshone Geyser Basins within the caldera and Heart Lake and Norris Geyser Basins just outside the caldera (Figure 2). Thermal waters tend to emerge where faults cut across the topographically low basins, and hydrothermal activity is most pronounced at the inter­ sections of faults. The waters are mostly chloride rich (Table I, numbers 2-9) and deposit siliceous sinter that over the years has accumulated in thick mounds and terraces. However, acid-sulfate boiling pools with little or no discharge of liquid water (Table 1, number 13) and mud pots occur in some of the geyser basins where upflowing boiling water is only a few meters below the land surface. In these instances, the residual liquid-water fraction generally flows laterally a relatively short distance to emerge as a boiling spring in a topographically low area. The steam that separates from boiling, alkaline-chloride waters generally contains H2S that oxidizes to H2S04 when it comes into contact with air in perched pools of ground­ water. Examples of steam-heated acid-sulfate features above shallow, boil­ ing, alkaline-chloride water are Iron Spring in the Upper Geyser Basin (Table 1, number 13), the Fountain Paint Pot in the Lower Geyser Basin, and the Thumb Paintpots at West Thumb. Hydrothermal fluids that issue from all the geyser basins cited above flow to the surface from reservoirs that have temperatures ranging from about 1800 to 270°C (Fournier et al 1976) at minimum depths of 100 to 550 m (estimated using theoretical boiling-point versus depth curves for hydrostatic conditions). These reservoirs are situated within thick sequences of rhyolitic lava flows and ash-flow tuff of relatively uniform chemical composition. by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Thermal waters at Mammoth are very differentfr om those in the geyser Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org basins. The Mammoth thermal waters flow to the surface through a thick sequence of sedimentary rocks that includes limestone, dolomite, and gypsum-hearing shales. Consequently, they are rich in bicarbonate and sulfate (Table 1, number 1). Relatively high concentrations of calcium and magnesium and low silica concentrations reflect a low temperature of water-rock equilibration at Mammoth: An isothermal temperature of 73°C was measured in a 112.8-m well, and 88°C was estimated by chemical geothermometry (White et al 1975). At higher temperatures magnesium and calcium are removed from solution, mostly as silicates and carbonate, YNP HYDROTHERMAL SYSTEM 21 and the fluid becomes richer in sodium and potassium. In general, thick travertine deposits, such as those found at Mammoth, are found only where limestone is in the section traversed by the thermal fluid and res­ ervoir temperatures are less than about 100--120°C.

Vapor-Dominated Systems In vapor-dominated systems, steam and other gases (mainly CO2 and some H2S) fill open fractures throughout a significant vertical extent beneath a cap that acts as a throttle, and liquid water is present mainly in pore spaces. The vapor-dominated zone essentially acts as a heat pipe, with rising steam carrying thermal energy upward and volumetrically small amounts of liquid water, formed by condensation of steam, flowingback downward. The cap at the top of the vapor-dominated zone is generally considered to be relatively impermeable rock that may support a zone of perched water that is a mixture of local meteoric water and condensed steam. Temperatures and pressures commonly change very little within the vapor-dominated zone, which (in consequence) is underpressured with respect to a normal hydrostatic system. The existence of a hydrostatically underpressured system requires either that the vapor-dominated region be perched above the surrounding lowlands in topographically high terrain or that a region ofrelatively low permeability surrounds the vapor-dominated zone, effectively preventing the free movement of cold groundwater into the system. Typical surface expressions of underlying vapor-dominated systems are fumaroles, acid-sulfate boiling pools with little or no discharge of liquid water (Table 1, numbers 14 and 15), mud pots, and absence of nearby alkaline-chloride thermal waters issuing at the surface. Throughout much of the relatively high plateau region in the east-central part of the Park, violently boiling, steam-heated pools of acid-sulfate water containing little chloride and with little or no visible discharge of liquid water dot barren landscapes composed predominantly of residual quartz, clays, sulfates, and native sulfur. Vapor-dominated conditions are thought to underlie

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. much of this region (White et aI1971), and a vapor-dominated system has

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org been confirmed by research drilling at Mud Volcano (White ct al 1975, Zohdy et al 1973). Mud Volcano is the most accessible of these localities, but other major acid-sulfate thermal manifestations are at Hot Springs Basin, Josephs Coat Hot Springs, and Washburn Hot Springs (Figure 1). Within the heat-pipe zone of large vapor-dominated systems, such as The Geysers in California and Larderello in Italy, the temperature is generally about 240°C with a corresponding vapor pressure of water about 33 bars (White et al 1971). In my experience, temperatures at the tops of small vapor-dominated zones that have been explored by drilling in various 22 FOURNIER parts of the world generally range from about 150° to 200°C, and thus the steam that heats the overlying fumaroles, mud pots, and acid-sulfate pools generally separates from the liquid-water fraction at a pressure of at least 5 to 15 bars. Therefore, upflowing steam that is rich in CO2 with minor H2S may condense underground at a temperature above 150°C and react with the surrounding rock. This reaction produces a solution rich in sodium bicarbonate, poor in chloride (only a small amount of chloride is leached from the local rock), and poor in sulfate (oxygen is not available to oxidize H2S much below the water table) (Table 1, number 16). The spring from which sample number 16 in Table 1 was collected is a very small, continuously discharging, superheated spring in a region of boiling, steam-heated pools that exhibit little or no discharge of liquid water. Chemical geothermometry (Fournier 1981) applied to this water indicates a temperature of water-rock reaction of about 180-190°C. The com­ position of a nearby nondischarging, steam-heated pool is shown by sam­ ple number 15 in Table 1. The latter water contains abundant sulfate derived from the oxidation of H2S that is streaming through the pool, a relatively large amount of the volatile constituent ammonia, very little chloride, and ratios of cations indicative of decomposition of rock with little water-mineral equilibration in the process. All the chloride-poor thermal waters in the eastern part of the Park contain relatively high concentrations of ammonia (e.g. numbers 14-16, Table 1). The highest reported ammonia concentrations have been found in feebly discharging hot springs in the Washburn Hot Springs region, about 1 km north of the northern rim of the Grand Canyon of ·the Yellowstone, where Allen & Day (1935) found 893 mg kg-1 NH4 and 2444 mg kg-1 S04 in one thermal water. Subsequent collections and analyses of thermal waters from this area by the US Geological Survey (USGS) confirm that the predominant dissolved constituent in the fluids is ammonium sulfate, and silica is the next most abundant dissolved con­ stituent (Thompson et al 1975). The amounts of dissolved ammonium sulfate in different springs are highly variable, however. Examples of

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. thermal waters from the Washburn Hot Springs area are given in Table 1

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org (numbers 19 and 20). Allen & Day (1935) noted that the gas collected from this locality contains appreciable methane (13.20%) and ethane (1.15%) and concluded that the organic gas and ammonia are probably derived from distillation of buried sediments. In general, this conclusion appears to be justified, especially in view of the occurrence of hydrocarbons at three localities in the eastern part of the Park: Tower Bridge, Calcite Springs, and Rainbow Springs (Figure 1). Love & Good (1970) studied the hydrocarbons in thermal waters from these localities and from other localities in an arcuate southeastward- to eastward-trending area about YNP HYDROTHERMAL SYSTEM 23 115 km long in northwestern Wyoming, and they concluded that the source of the hydrocarbons could be Paleozoic or Mesozoic sediments that underlie the surficialvolcanics. Evidently, hot water flushes petroleum from sedimentary rocks deep underground, and mixtures of water and petroleum rise together along fractures to the base of a vapor-dominated system situated beneath Wash­ burn Hot Springs. There, distillation at high temperature and high pressure results in steam that is rich in organic gases and for which NH3> H2S. Where this steam condenses at shallow levels, the accompanying H2S is oxidized to sulfuric acid, which is immediately converted to ammonium sulfate by excess NH3 (A. H. Truesdell, oral communication, 1988). This results in a water that is slightly alkaline, containing mostly dissolved ammonium sulfate (Table 1, numbers 19 and 20). The Mirror Plateau (that includes the eastern resurgent dome and Hot Springs Basin, Figure 2) and Mud Volcano regions may not be the only places within Yellowstone Park where vapor-dominated hydrothermal systems occur. Boiling, acid-sulfate pools and mud pots occur throughout an extensive area of acid-altered rock in the Smoke Jumper Hot Springs region (Figures 1, 2), which lies astride the Continental Divide. The outer edge of the main fracture of the 0.6-Ma Yellowstone caldera is thought to pass just to the west of this area of hydrothermal activity. It is possible that relatively dilute thermal waters with high proportions of bicarbonate to chloride fo und in the Boundary Creek region near the southwest corner of the Park (Figure 2 and Table 1, number 18), and similar waters flowing from the Hillside group of hot springs in the Upper Geyser Basin (Table 1, number 17), are in part steam condensate that flowedlaterally away from the topographically high Smoke Jumper Hot Springs system (Thompson & Hutchinson 1980).

Transition from Hot-Water- to Vapor-Dominated at End of Glaciation The surface expression of the hydrothermal system beneath the Mirror

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Plateau has not always been characterized by acid-sulfate activity. Old,

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org decaying geyser mounds and soil-covered siliceous sinter deposits are common throughout this region. Such sinter deposits form only where alkaline-chloride thermal waters flow over the ground. Therefore, the hydrothermal system has changed from hot-water- to vapor-dominated since the end of the last glaciation, about 14,000 yr B.P. During glaciation, fluid pressures within the hydrothermal system increased and decreased in proportion to thc thickncss of the ice cover (discussed in detail later). The increased hydrostatic head was probably sufficientto inhibit the development of a vapor-dominated system or even 24 FOURNIER to temporarily condense a preexisting vapor-dominated condition. After the ice disappeared, it is likely that many years were required to generate, or regenerate, vapor-dominated conditions at depth. Also during the waning stage of glacial activity, ice probably remained within the gorge of the Grand Canyon of the Yellowstone after the ice sheet had melted from the adjacent plateau. Ice filling the Grand Canyon would have kept the regional water table high, allowing alkaline-chloride hot springs to flowat relatively high elevations. As the ice in the gorge diminished in volume, the water table slowly lowered through rock that had previously been heated to very high temperatures by upflowing thermal waters. This could cause the hydrothermal activity in topographically high places to change from out­ flow of alkaline-chloride water to discharge of high-pressure steam, throt­ tled by poor permeability within highly altered and silicifiedrocks and by perched NaHCOrrich groundwater that serves to keep a pressure lid on the system. Alkaline-chloride waters still emerge at topographically low places in the eastern part of the Yellowstone caldera, such as deep within the Grand Canyon of the Yellowstone, where rapid erosion has cut a steep gorge more than 350 m deep, and at Rainbow Springs, which is about 1.8 km NW of Hot Springs Basin and only about 50 m lower in elevation. This suggests that some of the vapor-dominated hydrothermal activity in the eastern part of the Park may be relatively shallow rooted. However, in a region of generally impermeable rock underlain by a large heat source, it is possible that there may be rapid lateral transitions from shallow, alkaline-chloride hot-water to deep vapor-dominated conditions. Gravity and P-wave velocity data are compatible with the presence either of partial melt or a very deep and extensive vapor-dominated system extending from Hot Springs Basin to beneath the eastern resurgent dome (Lehman et al 1982, Smith & Braile 1984).

FACTORS AFFECTING THERMAL FLUID COMPOSITIONS

Fluid-Rock Interactions by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only.

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org The total dissolved solids in hydrothermal fluids are controlled in great part by the availability and solubilities of salts that can be leached from the wall rocks through which the fluidsflow, augmented by anions contributed fr om the influx of acid gases (mainly CO2 at Yellowstone). There is the possibility, however, that some salinity may be contributed by a very small amount of deep brine of magmatic origin that mixes with a large amount of dilute water of meteoric origin (Fournier & Pitt 1985). The maximum amount of magmatic water mixed with water of meteoric origin cannot be more than about 0.2-0.4% (discussed later). YNP HYDROTHERMAL SYSTEM 25 The effect of rock type on fluid compositions at Yellowstone was men­ tioned previously with regard to the relatively low-temperature bicar­ bonate- and sulfate-rich waters at Mammoth that issue from sedimentary rock, compared with the chloride-rich waters that issue from rhyolitic volcanic rocks. However, at high temperatures the effect of rock type is generally greatly diminished because magnesium becomes incorporated in clays and micas and the solubilities of CaC0 and CaS0 decrease as 3 4 temperature is increased, allowing waters to become relatively rich in alkali chlorides. The compositions of the alkaline-chloride waters that issue in hot springs reflect water-rock equilibration in rhyolitic rocks at estimated temperatures ranging from about 160° to 350°C (Fournier et al 1976, Truesdell & Fournier 1976, McKenzie & Truesdell 1977). However, lead isotope data for hot-spring deposits (travertine and siliceous sinter) indi­ cate that all the thermal waters in the Park have a component that has been in contact with sedimentary rock (Leeman et aI1977). Cation ratios are controlled by temperature-dependent hydrolysis and base-exchange reactions involving dissolved constituents and minerals in the wall rock. Hydrothermal alteration of rhyolitic material encountered in several relatively shallow drill holes has been described in a series of reports (Honda & Muffler 1970, Keith & Muffler 1978, Keith et al 1978, Bargar & Beeson 1981, Bargar & Muffler 1982). In general, glass-rich rhyolitic material has altered to a variety of zeolites, clays, micas, and hydrothermal K-feldspar. Minerals commonly found in veins include amorphous silica, chalcedony, quartz, calcite, fluorite, pyrite, iron oxides, and various clays and micas.

Decompressionai Boiling and Underground Mixing of Waters In addition to rock-water interactions, decompressional boiling and under­ ground mixing of different waters are important processes that affect hydrothermal fluid compositionsin the hot-water systems at Yellowstone. For example, variations in chloride and bicarbonate in Upper Basin ther­

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. mal waters, shown in Figure 3, result from varying degrees of conductive

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org cooling and decompressional boiling and from mixing of differentthermal waters that come from reservoirs that have only slightly different tem­ peratures. Waters that have cooled conductively without boiling as they flowed slowly upward to the surface fall along line AC. At the other extreme, waters that have undergone maximum decompressional boiling without significantconductive cooling fall along line BD. One end-member water has an initial composition prior to boiling shown by point A (the Geyser Hill-type water of Fournier et aI 1976), and the other has an initial composition shown by point C (the Black Sand-type water of Fournier et 26 FOURNIER

700

600

ill -0300 ·C o ::c 200 o 100

100 200 300 400 500 600 700 Bicarbonate, mg/kg

Figure 3 Chloride vs. bicarbonate fo r waters from Upper Geyser Basin. Bicarbonate concentrations are adjusted for the decrease in bicarbonate and the formation of carbonate 2 that occurs at shallow levels by the reaction 2HCOj == COj +H20+C02• The cluster of dots labeled M represents waters that are mixtures of thermal and dilute, cold groundwaters.

al 1976). Waters that plot below line AC have been diluted by shallow groundwaters. The Hillside group of springs (triangles) may have a com­ ponent of steam condensate that would account for relatively high con­ centrations of bicarbonate relative to chloride, as discussed previously in the section on vapor-dominated systems. The range in chloride concentrations shown by boiling and nonboiling hot-spring waters that are geographically close together and that have similar ratios of dissolved constituents (e.g. CI/HC03) can be used to estimate the temperature of the reservoir supplying water to the springs (Fournier 1979). This is conveniently accomplished by making use of enthalpy-chloride relations, as shown in Figure 4. All the available data for by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. thermal waters from Geyser Hill in Upper Basin that have Cl/HC03> 3 (in

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org equivalents) have been plotted in this figure. Enthalpies were obtained from steam tables (Keenan et al 1969) and are those of thc residual liquid at the measured collection temperature. In Figure 4 the initial enthalpy and chloride concentration of the reservoir fluid (point R) are given by the intersection of line SB, drawn from steam to the highest measured chloride concentraton (point B, assumed to have undergone maximum possible boiling), with the line drawn parallel to the enthalpy axis through the lowest chloride concentration (point A, assumed to have cooled entirely by conduction because of a very low rate of discharge and a temperature YNP HYDROTHERMAL SYSTEM 27

600

500 B

(1l .::£. "-.400 (1l 0 ___�II:R E �300 Q) -0 ·c 0200 -.C U

100

s 0 0 500 1000 1500 2000 2500 3000 Enthalpy, J/ 9

Figure 4 Enthalpy vs. chloride for waters from Upper Geyser Basin.

less than the boiling temperature). Thus, prior to decompressional boiling and/or conductive cooling, the initial reservoir fluid contained about 360 mg kg -l chloride at an enthalpy of about 920 J g-l (2I S°C). To the extent that the naturally discharging spring waters do not cover the entire range of "maximum possible boiling" to "entirely conductively cooled" conditions, the estimated initial enthalpy and chloride in the reservoir water will be in error. Also, in the above process care must be taken to exclude waters that have low chloride concentration because of mixing with dilute, usually cool groundwaters, such as the waters represented by the cluster of dots labeled M in Figure 3. Commonly, dilution with shallow groundwaters can be detected by increased concentrations of Mg in the

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. resulting mixed water and by flow rates that seem too large to be com­

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org patible with conductive cooling. Silica can be used even more effectively than chloride to estimate the temperature of the reservoir that is feeding a group of hot springs and geysers because of the constraint that the solubility of quartz controls the concentration of silica in the reservoir fluid. Figure S is an enthalpy-silica diagram that shows the solubility of quartz (Fournier & Potter 1982) and measured silica in the Geyser Hill waters that were plotted in Figure 4. As with the enthalpy-chloride relations shown in Figure 4, the range in silica concentration plotted in Figure S can be used to determine the enthalpy 28 FOURNIER 600

500

B �400 "" 01 0 E 300 A - 0 u := 200 (f)

100 E ., Ql en 0 0 500 1000 1500 2000 2500 3000 Enthalpy. Jig Figure 5 Enthalpy vs. silica for waters from Upper Geyser Basin.

(converted to temperature with steam tables) and silica concentration in the reservoir fluid prior to boiling (point R). Point R coincides with the solubility of quartz curve and also has the same enthalpy as point R in Figure 4. The enthalpy-silica and enthalpy-chloride data present a remarkably consistent picture. Variations in the amount of boiling and different amounts of mixing of two or more waters can be exhibited by samples collected at differenttimes from a single spring, as shown by frcqucnt sampling and analyses of waters from Cistern Spring in Norris Basin (Fournier et al 1986). Waters from this spring were collected and analyzed five times from May 1966 through June 1975. During that period the chloride concentration ranged from 455

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. to 475 mg kg-I, except for the last sample (290 mg kg-I) that was collected

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org one month before a local magnitude 6.1 earthquake. Because the 38% decrease in chloride might have been a prccursor of the earthquake, a monthly collection and analysis was made of Cistern Spring from 1976 to 1985 to check for possible changes in composition preceding subsequent earthquakes. The results, in terms of CI and S04, are shown in Figure 6, along with data for Echinus Geyser, Steamboat Geyser, and springs with low sulfate collected at Porcelain Terrace (all in Norris Basin). Since 1975, no correlation with seismic activity has been found, but large changes in the chloride/sulfate ratio and the degree of boiling have been observed • Cistern Spring • 400 • Steamboat Geyser • Echinus Geyser, pre-1979 � . to. , ' 1979-1980 • \ I. to. Acid-sulfate seeps • � t " Springs with lowest sulfate 300 I 1 0 200 400 60 t"" '" Chloride, mg/kg � Figure 6 Chloride vs, sulfate for Cistern Spring and other selected thermal waters from Norris Geyser Basin (from Fournier et aI1986),

tv by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. � Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org 30 FOURNIER that correlate with widespread seasonal boiling phenomena that occur in Norris Basin every August or September. In Figure 6 thc lcast boiled waters lie closest to the chloride-sulfate mixing line passing through point M' and the star indicating the composition of water collected from the Y- 12 research well drilled at Porcelain Terrace by the USGS (White et al 1975). More extensively boiled waters lie farther above this line. The data are interpreted as indicating mixing of three types of water: a chloride-rich and sulfate-poor component with a temperature> 260°C, a chloride-poor and sulfate-rich component with a temperature < 200°C, and a dilute, relatively cool component with little chloride or sulfate (Fournier et al 1986). Thc most pronounccd mixing (dccrease in chloride and increase in sulfate) occurs immediately after episodes of intense boiling. Variations in intensity of boiling apparently are related to changes in pressure within the hydrothermal system that are affected by several factors, including seasonal changes in water table, self-sealing by mineral deposition, clogging of channels by clay suspended in flowing waters, geyser eruptions, defor­ mation and fracturing induced by tectonic forces, and hydrofracturing.

DISSOLVED GASES RELATIVE TO AN ENTHALPY-CHLORIDE MODEL

Enthalpy-Chloride Relations Using chemical geothermometers, ranges in composition resulting from different proportions of boiling and conductive cooling (as in Figures 3- 6), and limited information from relatively shallow drill holes, Fournier et al (1976) estimated chloride concentrations, temperatures, and enthalpies of the shallowest reservoirs supplying thermal waters to the various geyser basins east of the Continental Divide and plotted the data on an enthalpy­ chloride diagram, similar to Figure 7. They concluded that all the inves­ tigated waters could be related to a deep water containing about 400 mg kg-' chloride at an enthalpy of 1570 J g-' (about 335-340°C) by boiling (lines radial to the enthalpy of steam) and by mixing with a cold dilute­ by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. water component (lines approximately radial to the origin). Truesdell Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org & Fournier (1976) performed a similar analysis, including water from Shoshone Geyser Basin west of the Continental Divide, and concluded that a deep "parent" water might have a temperature as high as 350-370°C. A major underlying assumption in the construction and interpretation of Figurc 7 is that both boiling (with separation and removal of steam from contact with the residual liquid) and mixing of fluids occur deep underground as well as near the surface, and that there may be complete or partial water-rock equilibration along the flow path or within a reservoir at an intermediate depth after boiling or mixing of fluids has occurred. YNP HYDROTHERMAL SYSTEM 31

1000

900

800 0'1 � 700 CH ...... 0'1 600 E NBO 500 Q) GB Sy "'0 ' ;:: 400 0 ..c 300 U 200 MH 0

100 HS O c.� ST 0 0 500 1000 1500 2000 2500 3000 Enthalpy, Jig

Figure 7 Enthalpy of liquid water vs. chloride for various aquifers. Letter codes as follows: 7M, Sevenmile Hole; BS, Black Sand-type water; CH, Crater Hills; CW, cold water; DUB, deep Upper Basin 270°C water; GB, Gibbon; GH, Geyser Hill-type water; HL, Heart Lake Basin; HS, Hillside Springs; LB, Lower Basin high chloride/bicarbonate water; MB, Midway Basin; MH, Mammoth Hot Springs; MJ, Madison Junction; NB, Norris Basin; P, the deep parent water of Fournier et al (1976); R, deepest and hottest reservoir fluid; SS, Shoshone Basin high-chloride water; ST, steam; SY, Sylvan Springs.

The model that is presented in Figure 7, though very speculative, provides a reasonable framework for interpreting gas and isotope data. Fournier et al (1976) suggested that 270De water might be widespread beneath Upper and Lower Geyser Basins in a permeable zone at an intermediate depth between deeper, higher temperature water and shal­ lower 20Q-2l5De local reservoirs that feed the hot springs and geysers. In

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. the construction of Figure 7 this 2700e water (point DUB) was assumed to be a mixture of the deeper � 3400e water (point P) and cold, dilute Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org water (point eW), and it was further assumed that no conductive cooling occurred before or after mixing. Therefore, the 270De water would be expected to contain all or most of the gases originally present in the deeper water (Fournier 1981), plus an atmospheric component of gas dissolved in the diluting water and a small amount of additional gas that might be leached from the local rock, such as radiogenic argon (Kennedy et al 1987). In contrast, waters that have undergone considerable boiling and loss of steam moving fr om the deeper reservoir to a shallower reservoir, such as 32 FOURNIER Norris, are expected to be somewhat depleted in deep-component gases. The distribution and variations in the compositions of the major gases (primarily CO2) and the noble gases (including their isotopes) in the various Yellowstone waters are consistent with this model.

Carbon Dioxide and Bicarbonate Waters in shallow reservoirs feeding springs in Lower Basin, Midway Basin, the western and northern parts of Upper Basin, the Sevenmile Hole region of the Grand Canyon of the Yellowstone, and Shoshone Geyser Basin all cluster close to the mixing line drawn from the 270°C water (point DUB, Figure 7) to dilute water. Departures from that line can be the result of a small amount of conductive cooling and/or a small amount of boiling of waters moving fr om the deeper to the more shallow reservoirs. Again, the thermal waters in these shallower reservoirs would be expected to contain all or most of the gaseous components originally present in the deeper reservoirs, but modified by the gases in the diluting water and by reactions with the wall rock. In these waters, little CO2 has been lost by partitioning into a separating steam phase. Therefore, CO2 is available to form carbonic acid and react with the wall rock as temperatures decrease, resulting in higher bicarbonate concentrations (e.g. Black Sand-type water) than are present in waters that have lost CO2 by boiling at high temperature before they equilibrated chemically with rock in a lower temperature reservoir (e.g. Norris and Geyser Hill-type waters) (Fournier 1981). The chloride-rich, low-sulfate, pH-neutral thermal waters at Norris have Cl/HC03 ratios > 13, the Geyser Hill-type waters (Figure 3, type A) have ratios ranging from about 3 to 8, and the Black Sand-type waters have ratios ranging fr om about 0.8 to 1.0.

Noble Gases and Their Isotopes The early studies of the relative abundances of dissolved gases, including Ne, Ar, Kr, and Xe, in the thermal waters of Yellowstone (Mazor &

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Wasserburg 1965, Gunter & Musgrave 1966, Mazor & Fournier 1973,

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org Gunter 1973) were interpreted as indicating that these gases originate largely from infiltrating meteoric water saturated with atmospheric gases. However, dissolved gas abundances are affected by processes deep within the hydrothermal system. Fluid samples collected from research wells, subject to the least contamination by atmospheric gases, were found to be markedly depleted in noble gases compared with the expected values for 20°C air-saturated water. Some samples carried a radiogenic component, as indicated by 4°Arp6Ar ratios higher than the atmospheric value and by strong enrichments in 4He relative to the other noble gases (Mazor & YNP HYDROTHERMAL SYSTEM 33 Fournier 1973). Craig et al (1978) and Welhan (1981) presented 3HerHe isotope data that indicate a subcrustal component of gas eHe) in the Yellowstone thermal waters. The largest leakage of 3He was detected at Mud Volcano, where a 3Hej4He ratio about 16 times the air ratio was detected. More extensive sampling by Kennedy et al (1985) confirmed the results of Craig et al (1978) and Welhan (1981) and provided broader coverage throughout the Park. Kennedy et al (1985) analyzed the abundances of five stable noble gases and the isotopic compositions of He, Ne, and Ar for 66 gas and water samples representing 19 different regions of hydrothermal activity. The isotopes of Kr and Xe were also determined in a fe w samples and found to be of normal atmospheric relative abundance. Kennedy et al explained their results by mixing of three source components: magmatic, crustal, and atmospheric, each having a characteristic noble gas isotopic range, with high 3He indicative of a magmatic source. In addition to the 3He anomaly at Mud Volcano, Kennedy et al (1985) found 3Hej4He 13.4 times the air ratio at Gibbon, 10.4 times the air ratio at Crater Hills, 9.5 times the air ratio at Shoshone, and generally about 7 times the air ratio elsewhere within the caldera. Except in the southwest region of the Park at Boundary Creek, the maximum 3Hej4He decreased rapidly to values < 3 times the air value immediately outside the caldera. High 3Hej4He anomalies were not found associated with either of the two zones of low seismic velocity (4.0 km S-I) reported by Lehman et al (1982) that have been widely cited as possibly indicating a body of partially molten material at a depth of 3- 4 km beneath the caldera. The lack of 3HerHe anomalies associated with the western anomaly is not surprising in view of more recent data that have shown that the western anomaly is fictive (Brokaw 1985). The anomaly at Mud Volcano and Crater Hills was attributed to a relatively shallow depth to the Curie-point isotherm and coincidence with a zone of maximum present-day uplift within the caldera (Kennedy et al 1985). The 3Hej4He anomaly at Gibbon was considered to be somewhat of an enigma, possibly reflecting the fact that the youngest rhyolite lavas in the Park (Doe et al

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. 1982) occur there.

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org Kennedy et al (1987) studied 12 different springs in Lower Geyser Basin

and found that the 3HerHe ratio ranged from � 2.7 to 7.7 times the air ratio and increased with total bicarbonate concentration. Because the total bicarbonate concentration is a function of the availability of CO2, which in turn is a function of deep boiling with phase separation prior to COT bicarbonate conversion (as previously discussed), Kennedy et al (1987) concluded that boiling and dilution of a single thermal fluidcould explain their gas data fo r Lower Basin. They also noted that a fluid that boils to a greater extent loses a larger proportion of 3He, as well as CO2, leaving 34 FOURNIER

a helium-poor residual fluid in which the isotopic composition of helium will be strongly affected by the addition of radiogenic helium. An alternative interpretation of the district-wide helium data of Ken­ nedy et al (1985), based on the Lower Basin model of Kennedy et al (1987), is that 3HetHe ratios throughout the caldera and in immediately adjacent regions are controlled mainly by boiling and mixing. Within most of the western part of the 0.6-Ma Yellowstone caldera, where locally derived meteoric waters circulate relatively freely through rhyolite lava flows, a considerable amount of mixing of the deep component with shallow water occurs, followed by adsorption of radiogenic 4He from the surrounding permeable rocks. Thus, an initially high 3HerHe ratio that is about 16 times the air ratio everywhere deep within the caldera may be generally reduced to less than 7 times the air ratio at shallow levels. Exceptions are in those few places where little shallow mixing of the ascending liquid occurs, and where steam that separates at very high temperatures makes its way to the surface with little or no dilution by other steam formed at shallow levels where local meteoric water boils. It may be more than coincidence that the highest 3HerHe ratios all have been found at the margin of or outside the boundary of the 2.0-Ma caldera shown by Chris­ tiansen (1984) and, therefore, outside the region where relatively permeable rhyolite flows are likely to be found deep within the caldera between differentash-flow tuff units, as discussed previously. At Crater Hills the one flowing hot spring (Table 1, number 12) is located within the 0.6-Ma caldera and at the projected eastern edge of the 2.0-Ma caldera. This is a likely site for prolonged and intense hydrothermal activity since the formation of the first caldera. The hot spring at Crater Hills has the highest chloride concentration of any thermal water in the Park (1044 mg kg-1 measured by Allen & Day 1935) and the highest chemical geothermometer temperatures (270-300°C). This water appears to represent the most direct upward leakage of fluid from deep in the system. Enthalpy-chloride relations (Figure 7) suggest that the parent water

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. deep in the system (point R) has a minimum enthalpy of about 1700 J

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org g-l (about 350°C) and contains about 450 mg kg-1 Cl. Water cools adiabatically by decompressional boiling in moving [rom point R to the reservoir feeding the hot spring at Crater Hills, but evidently the evolved steam and other gases remain essentially in contact with the residual liquid. This requires relatively impermeable surrounding rock. The proximity to the Mud Volcano vapor-dominated system suggests that this is likely to be the general situation in that part of the Park because low permeability is required to maintain a hydrologically underpressured zone next to a zone of hot-water upflowfr om deep in the system. Low permeability within YNP HYDROTHERMAL SYSTEM 35 the upper part of the stratigraphic section also provides a mechanism by which steam that forms deep under the Mud Volcano region rises to the surface with little contamination by gases in steam derived by boiling, shallow, local meteoric water. At Gibbon the hot-spring waters and gases issue from relatively impermeable ash-flow tuffs.There the enthalpy-chlor­ ide relations (Figure 7) show that the water in the local shallow reservoir has about the same chloride concentration as that in the deep parent water. Therefore, the Gibbon water may have cooled conductively without mixing while flowing from deep in the system to a more shallow level. In this event, the initial concentrations of noble gases in the deep water would tend to be preserved, modified mainly by leaching of radiogenic com­ ponents from the shallow reservoir rocks.

METEORIC RECHARGE OF THE HYDROTHERMAL SYSTEM

Craig et al (1956) compared stable isotopes in Yellowstone thermal waters with those in local meteoric water and concluded that any possible mag­ matic water contribution is less than a few percent. Although the interpre­ tation of the stable isotope compositions of hot-spring waters is made difficult by considerable dilution of ascending deep water by relatively locally derived meteoric water and by subsurface boiling that results in fractionation ofisotopes between steam and the residual liquid, the general conclusions of Craig et al (1956) still appear to remain valid after the collection and analyses of many more samples. Taking account of variations in chloride concentrations and of oxygen and deuterium isotopic abundances in thermal waters in the Park that result from boiling and mixing, Truesdell et al (1977) proposed two end­ member models for the behavior of steam that forms by decompressional boiling of an ascending solution. The first is a '''single-stage'' model in which all the steam remains mixed with the water and separates near a single temperature, and the second is a "continuous" model in which steam

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. is physically separated from the water as it is formed. A multiple-stage

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org model that produces results intermediate between single-stage and con­ tinuous steam separation appears to be applicable to many natural hot­ spring waters in Yellowstone. Truesdell et al (1977) assumed a temperature (360°C) and chloride cvncentration (310 mg kg-I) of a deep thermal water (taken from the enthalpy-chloride model of Truesdell & Fournier 1976) and used the measured temperature, chloride concentration, and isotopic composition of water collected from the Y-12 drill hole to calculate the relative deuterium concentration «(jd) of the deep-water component, using the (j (del) notation in which the isotopic concentration is reported as parts 36 FOURNIER per thousand difference from a standard, where

(jd= [(rsample/rsMow) -l] x 1 03, (1) rsample is the ratio of deuterium to hydrogen in the sample, and rSMOW is the ratio of deuterium to hydrogen in standard mean ocean water. They concluded that the chloride-(jd relations for Norris, Lower, Upper, and Shoshone Geyser Basins are compatible with boiling of mixtures of the deep water with cold, dilute (SOC, 2 mg kg-I Cl) meteoric water, varying in (jd according to locality in the Park. In addition, the data strongly suggest that recharge to the deep thermal water must be either precipitation from areas remote from the geyser basins or ancient precipitation from a cooler period. The major conclusions of Truesdell et al (1977) would not be changed significantly by assuming a higher temperature for the Y-12 water (it probably cooled conductively from at least 270°C rather than 238°C) and a higher chloride concentration in the deep, 360°C thermal water. Unpublished stable isotope analyses of an extensive suite of samples collected and analyzed by Robert O. Rye and additional stable isotope analyses by Tyler B. Coplen indicate that the most likely recharge area for the deep component of the geothermal water is in mountainous regions north and northwest of the caldera, where deep snow accumulates each winter.

GEOTHERMAL HEAT FLUX

The total convective heat flux in chloride-rich water and associated steam from all the Yellowstone thermal areas east of the Continental Divide was estimated in 1966-67 to be about 4.0 x 1 016 cal yr-I, or 5.3 x 109 W by the river chloride inventory method (Fournier et al 1976). Nearly all the chloride discharged by the hot-spring waters eventually enters the four major rivers that drain the Park: the Yellowstone and Madison Rivers east of the Continental Divide, and the Snake and Falls Rivers west of the Divide (Figure 1). The Firehole River (draining the Upper, Midway, and by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Lower Geyser Basins) and the Gibbon River (draining Norris Geyser Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org Basin) join to produce the Madison River. Fournier et al (1976) used enthalpy-chloride information (Figure 7) to estimate the chloride con­ centrations of the waters in the shallow reservoirs feeding individual geyser basins and the chloride concentration of the water deep in the hydro­ thermal system before decompressional boiling and before near-surface mixing (point P with 400 mg kg-I Cl). Measured rates of discharge of the rivers and of the chloride carried by those river waters were then used to calculate the rate of total discharge of the deep component of the hot­ spring water. The method automatically compensates for boiling and mixing of thermal waters as they rise. Corrections are required, however, YNP HYDROTHERMAL SYSTEM 37 for background chloride contributed to the rivers by precipitated rain and snow and by low-temperature weathering of rock. The background Cl generally amounts to about 1-2 mg kg-l in the nonthermal component of the river water, depending on the proportion of water from different nonthermal sources. (For example, snow melt that has not penetrated into the ground is lower in Cl than nonthermal groundwater that has been underground a long time in contact with various types of rock). A sum­ mation of the results for the Yellowstone and Madison Rivers, plus an additional 5% assumed for the thermal discharge into the Snake and Falls Rivers, gave a figure of 1011 kg yr- l for the deep component of the hydrothermal water and 4 x 107 kg yr-l fo r the chloride carried by that water in 1966-67 (Fournier et al 1976, Fournier & Pitt 1985). For a minimum initial water temperature of 340°C, the total amount of heat discharged by that water is 5.3 x 109 W, which translates to an average heat flux of about 2000 mW m-2 over the entire 2500 km2 of the Yel­ lowstone caldera. The above estimates are not changed significantly when the higher chloride and higher enthalpy values of point R in Figure 7 are used for the deep water instead of those of point P. Norton & Friedman (1985) reported results of monitoring the chloride fluxfr om the Yellowstone, Madison, Snake, and Falls Rivers for the 1983 water year, extending from September 1, 1982, through August 31, 1983. They calculated a total chloride flux of about 5.8 x 107 kg yr-1 in these river waters, of which about 5.5 x 107 kg yr-l were estimated to be con­ tributed by thermal water. They also found that the chloride flux from thermal waters into the drainage west of the Continental Divide is about 26% of the total for the Park, much higher than the 5% estimated by Fournier et al (1976). When 26% thermal water, rather than 5%, for the region west of the Continental Divide is applied to the 1966-67 data, a revised flux of thermal water chloride of 4.8 x 107 kg yr- 1 is obtained. This is 17% less than the thermal water Cl flux for the Park reported by Norton & Friedman (1985). The chloride flux measured in 1983 appears to be exceptionally high

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. compared not only with the 1966-67 measurements, but also with measure­

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org ments made in subsequent years (USGS, unpublished data). Figure 8 shows calculated instantaneous thermal water chloride fluxes for the Madi­ son River during the 1983 and 1985 water years and during 1966-67. A correction of 1 mg kg-1 Cl for the nonthermal contribution has been applied to the data shown in Figure 8. The wide scatter in the data after day 250 corresponds with periods of high river discharge during the spring and early summer melting of snow. During high stages of river discharge, the calculated fluxes of chloride have a greater uncertainty because small percentage errors in chloride analyses and flow measurements are mag­ nified and because discharge measurements, obtained from rating curves, 38 FOURNIER Madison River 1500

U Ql 0... I � 1000 0' e

X :::J '+-

Ql -0 500 'C 0 ..c U

O�rr.... -rrr""rrTT ..-rrr ..-rrrTT,,-r .... -r o 50 100 150 200 250 300 350 Day number (water year)

Figure 8 Calculated chloride flux vs. day of water year for the Madison River. Water year starts September 1 and ends August 31 of succeeding calendar year. Triangles indicate data for 1966, dots fo r 1967, circles for 1983, and asterisks for 1985.

are less precise as a result of changing channel profiles that require recali­ bration of rating curves. Also, during high stages of river flow, slight differences in the assumed background chloride correction make a big differencein the calculated fluxof thermal water, However, uncertainties in the background chloride correction can be minimized with the aid of a log-log plot of river discharge rate versus river chloride. This is done by extrapolating to the assumed CI concentration in the deep parent water using theoretical mixing curves that have fixed shapes, as shown in Figure 9. By this method it is easy to see at what discharge rate the correction for by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. background chloride becomes negligible. Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org Using data in Fournier et al (1976) and unpublished data of the USGS, I have calculated approximate ranges of the instantaneous flux of total thermal water discharged by all the chloride-rich hot-spring systems in the Park for the years 1966-67 and the water years 1983-86 (Figure 10). In assembling the data with which Figure 10 was constructed, no attempt was made to check the validity of the most extreme data points or to use conductivity data to estimate chloride concentrations between times when water samples were collected for analyses. The circles show where most of the data plot for low stages of river flow. The data presented in Figure 10 YNP HYDROTHERMAL SYSTEM 39

.. 400

100 Cl .., ..... Cl E

10 + 1966 -67

• 1984-85

1�������������������� 1 10 100 1000 Flow, cfs

Figure 9 Logarithmic plot of flow rate vs. dissolved chloride for the Gibbon and Firehole Rivers. Calculated mixing curves are for an assumed end-member deep thermal water containing 400 mg kg-1 chloride and non thermal waters containing I and 2 mg kg-1 chloride, respectively.

6

5

4 " QJ � co t t E 3

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. � 0 Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org u:: 2

O �--�--��----L---�--�--� 1966-67 1983 1984 1985 1986 Water year

Figure 10 Range in calculated instantaneous fluxes of thermal water vs. year. Circles show most common flux during low stages of river flow. 40 FOURNIER

clearly show about a 15-20% increase in flux of thermal water and heat during the 1983 water year compared with the other years. Interestingly, this anomalously high flux of thermal water occurred during the year preceding the 7.3 magnitude Mt. Borah earthquake on October 28, 1983. A little over three months before that earthquake the fluxof thermal water decreased to what appeared to be more normal values, and then a month before that earthquake the flux ofthermal water again increased to anom­ alously large values. Apparently, Norton & Friedman (1985) did not directly compare their data with the results of measurements made in 1966-67 at comparable stages of river flow anddid not conclude that there were significant differences between the two data sets, except in regard to the assumed flux of chloride from thermal water into the Snake and Falls Rivers. However, they did note an increase in chloride flux commencing early in October 1983 and culminating shortly after the Mt. Borah earth­ quake. Future detailed analyses of river flow, temperature, and con­ ductivity data collected hourly from 1982 through 1987 are likely to reveal significant variationsin the total flux ofthermal waters since 1983.

EFFECTS OF GLACIAL ICE COVER ON THE HYDROTHERMAL SYSTEM

A series of research wells drilled at several thermal localities in Yellowstone Park (White et al 1975) have provided information about present tem­ perature-pressure-depth relations. Near major zones of active convective upflowof thermal water, present temperatures follow boiling-point curves appropriate for water tables at the present land surface, with hydrostatic pressures equivalent to a cold rather than hot column of overlying water (Fournier 1983). Muffleret al (1971) postulated that boiling-point curves and rock temperatures were raised within the upper parts of some hot­ spring systems at Yellowstone during times of glacial cover, and that sudden decreases in the water table (as by draining of an ice-dammed lake) resulted in major hydrothermal explosive activity. Primary and secondary by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. fluid inclusions in material from several cored holes in Yellowstone

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org National Park (White et a1 1975) have been studied by Bargar & Fournier (1988) to quantify thermal effects that may have accompanied a raising of the water table and the corresponding boiling-point curve during times of thick glacial ice cover. The results from two holes (Y-6 and Y- 13) have been reported as parts of other investigations (Bargar & Beeson 1984, Bargar et al 1985). Mineral samples containing fluid inclusions suitable for study were found in drill cores from 8 of 13 wells drilled by the USGS. Primary, pscudosecondary, and secondary liquid-rich fluidinclusions in quartz and fluoritewere utilized to determine a range of homogenization temperatures YNP HYDROTHERMAL SYSTEM 41

(Th) at given depths. Within zones of major hydrothermal upflow, fluid inclusion Th values all were found to be equal to or higher (commonly 20- 50°C and up to 155°C higher) than present temperatures at the depth of collection. The greatest range in Th values was found in material collected from the Y-I3 drill hole in Lower Geyser Basin (Bargar et al I985). In the absence of glacial ice, topographic constraints place a limit of about 120- 150 m on the maximum depth of any lake occupying Lower Geyser Basin. The elevation of the water table over Lower Geyser Basin required to raise boiling-point curves sufficientlyto account for the observed fluidinclusion Th values ranges from 425 m (a conservative estimate that does not include the highest Th values because of possible necking-down phenomena after entrapment) to 750 m (a liberal estimate that includes all the data). Thus, it appears that glacial ice is required to account fo r the boiling-point curve elevation. The estimated thickness of the glacial ice over Lower Basin during the last glaciation required to explain the fluid inclusion data is within the range estimated by other criteria (K. L. Pierce, written com­ munication, 1988). Away from zones of major upflow, Th values were all found to be equal to or lower than present temperatures (commonly 10-20°C and as much as 60°C lower). The lower paleotemperatures at the margins of hydrothermal upflow maybe attributed to lower rates of discharge of warm water and/or higher rates of recharge of cold water in the past, probably during periods of glacial cover.

INTERACTION OF CIRCULATING WATER WITH THE HEAT SOURCE

In most current models of magmatic-hydrothermal systems (Lister 1974, 1980, 1983, Hardee 1982, Sleep 1983, Fournier & Pitt 1985), relatively dilute meteoric water circulates along fractures above and to the sides of a magmatic heat source. The water is separated from very hot crystalline rock or magma by an impermeable zone where quasi-plastic deformation

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. prevents brittle fracture and maintains low permeability. Heat extracted

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org by the circulating water slowly cools the system, allowing thermal cracking and faulting to progress inward so that water steadily migrates deeper into the heat source while attaining about the same maximum temperature. Although there is no direct evidence about the maximum temperature attained by the convecting hydrothermal fluid, enthalpy-chloride relations (Figure 7) provide a means of estimating a likely minimum temperature deep in the system, about 340-350°C. An approximate upper limit for the temperature likely to be attained by the water in the hydrothermal system that convects to the surface can be estimated from positions of boiling­ point and condensation-point curves (Fournier 1987) in temperature-pres- 42 FOURNIER

sure space for a dilute chloride-rich water about equivalent in salinity to a 0.7 wt% NaCI solution (Fournier & Pitt 1985). For the restricted range of compositions of the relatively dilute solutions that issue at the surface at Yellowstone, these fluids could not have attained temperatures higher than about 420°C at a depth of 4 km and 430°C at 5 km. Exceeding these temperatures would have resulted in an ascending solution intersecting the stability field of gas plus brine, resulting in the formation of a small amount of highly saline brine and a large amount of very dilute steam (Fournier & Pitt 1985, Fournier 1987). The likely depth of circulation of hydrothermal fluidsalso is a matter of conjecture. Deep drilling in other presently active hydrothermal systems has shown that permeability at temperatures above about 300°C is con­ trolled mainly by fractures. Precipitation of minerals and quasi-plastic flow are particularly efficient at decreasing permeability by closing fractures at temperatures above about 350-400°C (Fournier & Pitt 1985). Therefore, sustained circulation of hydrothermal fluids at these high temperatures is likely to take place only where the rocks are subjected to repeated brittle fracture that counteracts the processes that decrease permeability. Focal depths of well-located earthquakes are deeper than 15 km out­ side the Yellowstone caldera, but within the caldera they seldom exceed 4-5 km. Evidently the rocks below 3-4 km are too weak as a result of high temperatures (> 350°C) to sustain sufficient differential stress for seismic activity to occur (Smith & Bruhn 1984, Fournier & Pitt 1985). Therefore, the maximum depth of sustained fluid flow at hydrostatic pressure is probably about 4-5 km within the caldera and deeper at the sides. If the assumption is made that thermal energy is transferred uniformly by conduction from the heat source across a 2500 km2 surface, through hot dry rock to the base of circulation of the hydrothermal system, a thermal gradient of about 700° to lOOO°C km -I is required to sustain the average convective thermal output (Morgan et al 1977, Fournier & Pitt 1985). To account for both the large convective heat flux and the lack of

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. earthquakes below about 4-5 km, it is likely that magmatic temperatures

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org are attained beneath some portions of the caldera at depths as shallow as about 4.5-5.5 km, about the same depth range indicated by the various other geophysical methods.

CRYSTALLIZATION OF MAGMA AS A SOURCE OF HEAT AND AQUEOUS FLUIDS

If the heat carried in the advective flow of water from the system were supplied entirely by the latent heat of crystallization of rhyolite magma, YNP HYDROTHERMAL SYSTEM 43

about 5 x 10" kg yr-1 of magma (about 0.2 km3 yr-1) would be required to furnish that heat (Fournier & Pitt 1985). In contrast, cooling already crystalline rock by 300°C (to reduce the temperature from about 730° to 430°C) will furnish about the same amount of heat that can be obtained from the latent heat of crystallization of the same volume of rhyolitic magma (using a latent heat of crystallization of75 cal g-l and heat capacity of 0.25 cal g-l °C- 1). Thus, crystallizing about 0.1 km3 of rhyolitic magma and then cooling it about 300°C could furnish the heat carried to the surface by the hot-spring system. A lack of water-bearing minerals in the rhyolite erupted at Yellowstone coupled with the presence of the mineral fayalite suggests that the initial magma contained not more than 1 to 2 wt % dissolved water (Hildreth et al 1984). The chloride contained in fresh obsidian glass shows that the minimum initial chloride concentration in the magma was 1500 mg kg-1 (Wes Hildreth, oral communication, 1984). Strong partitioning of this chloride into an aqueous magmatic fluid that separates upon slow crystallization of the magma will produce a fluid containing a minimum of 12 to 25 wt% salt. The maximum quantity of water that could be liberated by crystallizing 0.2 km3 yr-I of magma initially containing 2 wt% water is 8.8 x 109 kg yr-1, and a minimum of about 7 x 108 kg yr-1 of chloride would be liberated with that water. For comparison, recall that about 10" kg yr-1 water (not including water incorporated by near-surface dilution) and 4 x 107 kg yr-1 chloride are discharged by the hot-spring system. Thus, a maximum of about 9 wt% of the deep component of hot-spring water could be magmatic, but only about 6 wt% of the available chloride can be discharged with that water. The deficiency of chloride in the discharged thermal water may be interpreted in two ways: (a) Thermal energy is supplied mainly by cooling already crystalline rock with little thermal input from the latent heat of crystallization of magma; or (b) brine evolved from presently crystallizing magma is ponding beneath the dilute hydrothermal system, and little of this magmatic water is incorporated into the convecting me­ teoric system. (A mixture of about 0.2-0.4% magmatic fluid with 99.6-

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. 99.8% meteoric water could account for all the chloride discharged by

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org the Yellowstone hot springs.) It was noted previously that stable isotope data also suggest that very little, if any, magmatic water is discharged by the hot springs (Craig et al 1956, Truesdell et al 1977). The pre­ ferred model of Fournier & Pitt (1985) is one in which the thermal energy carried to the surface by the convecting hot-spring water comes in part from crystallizing magma and in part from cooling crystalline rock, and that relatively dense brine liberated from the crystallizing magma is accumulating at depth beneath dilute, much less dense hydrothermal waters. 44 FOURNIER

LONGEVITY OF PRESENT HYDROTHERMAL ACTIVITY

Uranium series disequilibrium studies (Sturchio et al 1987) of cores from the Y-7 drill hole in Biscuit Basin, Yellowstone National Park, indicate that there was significant uranium mobility 15,000-20,000 yr ago, cor­ responding to a period of time when glaciation would have allowed tem­ peratures to be significantly higher at the depths penetrated by the drill hole. Unpublished uranium series disequilibrium studies of cores from other drill holes show that vein ages are generally much younger than the ages of the volcanic host rocks (N. C. Sturchio, written communication, 1988). Bargar & Fournier (1988) observed that Th values of primary and secondary liquid-rich fluid inclusions trapped in hydrothermal minerals sampled from within major zones of up floware all equal to or greater than presently measured rock temperatures at the point of collection. This is strong evidence that the thermal waters have not been cooler than present temperatures since formation of the host minerals. Also, because the presence of glacial ice seems to be required to explain elevated boiling­ point curves, it can be concluded that hydrothermal activity beneath the major geyser basins has been continuous since sometime within or before the last glaciation, a minimum of 14,000-45,000 yr. The fluorite and much of the hydrothermal quartz used in the Bargar & Fournier (1988) study were deposited relatively late in the sequence of hydrothermal events, after most of the surrounding rock had already been thoroughly altered by hydrothermal activity, but before the last glacial ice decreased in thickness to less than 450-750 m. The data are consistent with a model in which hydrothermal activity within the 0.6-Ma Yellowstone caldera started soon after the formation of that structure and continued unabated thereafter, with significant variations in intensity and places of discharge in response to pulses of volcanism and the waxing and waning of glacial episodes. Ovcr thc minimum time span during which hydrothermal activity

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. appears to have operated at about its present intensity (15,000 yr since the

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org end of the last glaciation), a largc amount of thermal energy has been extracted from the system, both by convection and conduction, as indi­ cated above. Ifwe use the assumed thermodynamic properties of materials given in Fournier & Pitt (1985), the calculated amount of heat extracted just by advection of chloride-rich water from the hydrothermal system in 15,000 yr could be supplied by subsolidus cooling over a 350°C range (for example, from 750° to 400°C) of a crystalline body of rock 1.2 km thick beneath the entire caldera (2500 km2), or by the latent heat associated with crystallizing a body of magma 1.2 km thick beneath the entire caldera with YNP HYDROTHERMAL SYSTEM 45 no change in temperature of the system. Crystallizing a body of silicic magma 0.6 km thick and then cooling it by 350°C would also provide the required thermal energy. Again, these are minimum requirements, as the total flux of heat from the system is undoubtedly larger than the number estimated from the chloride flux. The chloride fluxmethod does not pro­ vide information about hear' carried to the surface by steam separated from convecting liquid that cools and flows back downward into the system, or about a purely conductive component of heat flow.

VOLUMETRIC CHANGES ASSOCIATED WITH CRYSTALLIZATION OF MAGMA

Cooling of hot but already crystalline rock should lead to thermal con­ traction and a subsidence of the caldera floor. However, a significantuplif t of the caldera has been measured over the period 1923-84 (Pelton & Smith 1979, 1982, Dzurisin et a1 1986, Dzurisin & Yamashita 1987). This inflation stopped in 1985, and deflation occurred in 1986 (Dzurisin & Yamashita 1987). Studies of raised and submerged shoreline terraces show that there has been episodic uplift interspersed with local deflation of the caldera throughout the Holocene (Hamilton 1985, Meyer & Lock 1986). Meertens & Levine (1985) have suggested that the uplift is the result of horizontal compressive strain, but this does not appear to agree with fo cal mecha­ nisms of earthquakes within the caldera and the generally extensional tectonic environment of the region. Also, the average horizontal strain rates measured in the northeast part of the caldera and in the Hebgen Lakc region to the northwcst of thc caldcra from September 1984 to August 1987 appear to preclude regional tectonic movements as a primary cause of the inflation and deflation(unpublished data of J. C. Savage and D. Dzurisin). Two mechanisms that might account for both the episodic inflation and deflationof the caldera each require crystallization of rhyolitic magma in amounts that are compatible with the thermal budget fo r the magmatic­

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. hydrothermal system. Both mechanisms may operate to varying degrees

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org at the same time. The first involves episodic intrusion of basalt from the mantle into the upper crust at the base of a silicic magma chamber at a depth of about 10-15 km, and continuous crystallization of rhyolitic magma at a depth of about 4-5 km. The thermal energy discharged by the advecting thermal waters at Yellowstone could be supplied by the latent heat of crystallization of about 0.2 km3/yr of rhyolitic magma, with no change in temperature within the igneous part of the system (Fournier & Pitt 1985). If purely conductive heat loss and heat carried upward by convecting hydrothermal fluids that do not discharge at the surface also 46 FOURNIER

were taken into account, the amount of magma required to supply the thermal flux from the magmatic-hydrothermal system would be cor­ respondingly higher. Crystallization of 0.2 km3 of rhyolitic magma would lead to a volumetric contraction of about 0.014 km\ provided that the evolved, aqueous-rich magmatic fluidall escaped into the overlying hydro­ statically pressured hydrothermal system. Thus, if the latent heat of crys­ tallization of magma is an important source of thermal energy for the hydrothermal system and if the evolved magmatic fluidscontinuously and efficiently escape from the system, then the normal state of deformation of the caldera is likely to be deflation at a rate that is close to the observed rate of subsidence since 1985. Inflation will occur only during periods of episodic intrusion of basalt from the mantle into the crust at a rate fast enough to more than compensate for the decrease in volume resulting from crystallizing rhyolite magma. Stability at the caldera floorwill occur only when the volume increase resulting from basalt intrusion at the base of the relatively high-level magma chamber just balances the volume decrease resulting from crystallizing rhyolite at the top of the chamber The second mechanism that can account for rapid fluctuations from inflation to deflation involves entrapment of aqueous magmatic fluids at lithostatic pressure beneath self-sealed rock and the episodic release of these fluidsinto the hydrostatically pressured system (Fournier & Pitt 1985, Fournier 1988). Self-sealing at the base of the hydrostatically pressured hydrothermal system may occur as a result of chemical transport and precipitation of vein minerals and ductile flow of rock that becomes rela­ tively weak at temperatures exceeding 400°C (Fournier & Pitt 1985). lfthe magmatic fluid that is liberated during crystallization of 0.2 km3 yr-I of rhyolitic magma containing about 2 wt% water is trapped at litho static pressure at a depth of 4-5 km, there will be a net volume increase of about 0.026 km3 yr- I; more than enough to account fo r the observed rate of inflation of the caldera from 1923 to 1984. The density of the magmatic gas phase is not particularly sensitive to the initial ratio of water to chloride in the magma, because at the likely depth and temperature of

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. crystallization, the magmatic fluid will dissociate to a small amount of

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org very saline brine plus relatively dilute gas, as previously discussed. Episodic hydro fracturing and injection of pore fluids that have accumulated at lithostatic pressure into the hydrostatically pressured hydrothermal system would result in episodic deflationof the caldera. The pressure surge associ­ ated with this injection of magmatic fluid may be accommodated entirely or in part by an increased rate of flow of thermal water from the system, by a decrease in the rate of recharge of cold meteoric water into the system, or by condensation of a relatively small amount of steam accompanied by a slight rise in the level of the interface between liquid and steam within a YNP HYDROTHERMAL SYSTEM 47 deep vapor-dominated region that appears to be widespread in the eastern part of the Park. Also, episodic rupturing of self-sealed rock that separates regions of litho static and hydrostatic pore-fluid pressures could account for some of the seismic swarms that occur within the caldera. Although the above calculations and speculations do not demonstrate conclusively a causal relationship between the upliftand liberation of fluid from crystallizing magma, they do show that the volumetric effects of liberating a magmatic fluid could be important if the latent heat of crys­ tallization of magma is an important source of energy for the hydro­ thermal system. Future attempts will be made to determine whether a statistically significant increase in the rate of hot-water discharge has coincided with a change from inflationto deflationof the caldera. A major problem in the interpretation of the data is the large effect that the Mt. Borah earthquake appears to have had upon the hot-spring discharge rate. However, even if there is no detectable increase in the rate of discharge of thermal water coinciding with the onset of deflation, we cannot rule out the possibility that depressuring of magmatic water is a major contributing factor, because the volumetric effects may be most pronounced within the cold recharge part of the system and within the vapor-dominated zones.

MANTLE BASALT AS THE ULTIMATE SOURCE OF THERMAL ENERGY AT YELLOWSTONE

It is generally agreed that continuing injection of basaltic magma (1200- I 300°C) from the mantle into the crust over a long period of time supplies the thermal energy to melt crustal material, forming large volumes of silicic magma at 800-900°C (Christiansen 1984). Crystallizing and cooling (1200° to 860°C) about 0.08 km3 yr-I of basalt could furnish the heat for the present-day Yellowstone hydrothermal system. This yearly volume of basalt is about seven times more than the historic measured average rate of volumetric inflation of the Yellowstone caldera. If mass (either cooled basaltic magma or dense crystalline material) is not convecting or settling by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. downward into the lower crust and mantle beneath the Yellowstone cal­

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org dera to counterbalance the buoyant upflowof hotter basaltic magma, then the Yellowstone volcanic system as a whole is cooling at present and the amount of silicic melt (if any) in the shallow part of the system is steadily decreasing.

CONCLUSIONS

Meteoric water, derived in great part from precipitation on mountains to the north and northwest of the Yellowstone caldera, convects at hydro- 48 FOURNIER

static pressure to a depth of about 4--5 km within the caldera and attains a maximum temperature of about 350-430°C. This water picks up about 300-500 mg kg-I Cl from the rock through which it passes and also picks up gaseous components liberated from magma and crystalline magmatic rocks. Cation ratios are fixed by temperature-dependent water-rock reac­ tions. There is a possibility that some or even most of the chloride in the relatively dilute hydrothermal system is derived by interaction with an underlying brine of magmatic origin that is postulated to be present at depth. Even though some variation in chloride is likely to occur in the deepest parts of the convecting system, these variations appear to be small, probably because rocks with similar compositions occur at depth throughout the caldera, and because there has been a lot of flushing of dissolved constituents by hydrothermal advective flow throughout a 2- Myr interval of volcanic activity. Thus, the assumption in the application of boiling and mixing models that a deep "parent" water has a uniform chloride concentration is probably a good first approximation for use in estimating conditions deep in the system. Heated fluidconvects upward and to various degrees in different places boils and/or mixes with shallow fluids ofmore local meteoric origin. Water­ rock chemical reequilibration occurs in relatively shallow local reservoirs that directly feed individual geyser basins. Variations in dissolved gases in general, and in 3HerHe ratios in particular, can be explained entirely in terms of variation in underground boiling and mixing of deep and shallow waters. During periods of glaciation, temperatures in zones of major hydrothermal upflow increased at given depths as a result of elevation of boiling-point curves by increased hydrostatic head. Vapor-dominated systems in the eastern part of the Park formed at the close of the last glaciation partly as a result of a lowering of the water table by flow toward the Grand Canyon of the Yellowstone and partly in response to a large input of thermal energy beneath a thick sequence of relatively impermeable rock. The convective heat discharged by the hydrothermal system is gen­ erally about 4 x 1016 cal yr-1, but an approximate 20% increase was by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. observed in 1983, the year preceding the Mt. Borah earthquake. Hydro­

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org thermal activity has been continuous at about its present level for at least 15,000 yr and probably much longer. It is not known whether the magmatic-hydrothermal system, as a whole, is presently cooling down or heating up. The thermal energy that drives the hydrothermal convection can come entirely from the cooling of already crystalline rocks, entirely from the latent heat of crystallization of silicic magma at a depth of about 4--5 km, or from a combination of the two. The recent uplift of the Yellowstone caldera suggests that the injection of magma into the system and/or the liberation of fluid from a crystallizing magma are processes YNP HYDROTHERMAL SYSTEM 49 that are presently active. If magmatic fluids are presently being liberated from a crystallizing siliceous magma at a depth of 4-5 km, they should dissociate to highly saline brine plus relatively dilute gas. Very little (if any) of that brine is becoming incorporated into the portion of the hydrothennal system that convects to the surface. The magmatic fluidis ponding beneath the dilute fluid and/or is being trapped in a region where pore-fluidpres­ sures are lithostatic. Accumulation of magmatic fluids at litho static pres­ sure and episodic injection of portions of this fluidinto the hydrostatically pressured system may account for the observed inflationand deflation of the caldera. However, the observed uplift and subsidence also may be accounted for by episodic intrusion of basalt from the mantle into the upper crust at a depth of about 10-15 km and continuous crystallization of silicic magma at about 4-6 km without accumulation of magmatic gases at lithostatic pressure. Uplift would occur only when the rate of basalt intrusion more than compensated for the normal subsidence resulting from crystallization of rhyolitic magma. At this time the most plausible model for the magmatic-hydrothermal system is one in which the thermal energy carried to the surface by the advecting hot-spring water comes in part from crystallizing silicic magma and in part from cooling crystalline rock, and in which brines are accumulating deep in the system. It is likely that crystallization and cooling are occurring at some levels while new magma accumulates at others.

ACKNOWLEDGMENTS

Since I first started working on the Yellowstone hydrothermal system in 1969, I have benefited greatly from collaboration with many colleagues in the USGS, with special thanks to A. H. Truesdell, L. J. P. Muffier, J. M. Thompson, G. W. Morey, J. J. Rowe, and particularly Donald E. White, who firstintroduced me to the joys of studying the hydrogeochemistry of Ycllowstone's thermal waters and who has tried to keep me on the right

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. track ever since. Over the years the National Park Service (NPS) has been

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org and continues to be very supportive of our research in the Park. R. A. Hutchinson of the NPS has been especially helpful in collecting hot-spring water samples for analyses from remote regions and in monitoring the chemical behavior of various thermal waters over many years. D. R. Norton and Irving Friedman of the USGS have been deeply involved with a cooperative USGS-NPS program to monitor the chloride flux in the major rivers in Yellowstone since 1982, and they have provided me with much of the data for calculating the recent convective heat flux from the system. 50 FOURNIER

Literature Cited Craig, H., Lupton, J. E., Wei han, J. A., Poreda, R. 1978. Helium isotope ratios Allen, E. T., Day, A. L. 1935. Hot springs of in Yellowstone and Lassen Park volcanic the Yellowstone National Park. Carnegie gases. Geophys. Res. Lett. 5: 897-900 Inst. Wash. Publ. 466. 525 pp. Daniel, R. G., Boore, D. M. 1982. Anom­ Bargar, K. E., Beeson, M. H. 1981. Hydro­ alous shear wave delays and surface wave . thermal alteration in research drill hole velocities at Yellowstone caldera, Wyom­ Y-2, Lower Geyser Basin, Yellowstone ing. 1. Geophys. Res. 87: 273 1--44 National Park, Wyoming. Am. Mineral. Doe, B. D., Leeman, W. P., Christiansen, 66: 473-90 R. L., Hedge, C. E. 1982. Lead and stron­ Bargar, K. E., Beeson, M. H. 1984. Hydro­ tium isotopes and related trace elements thermal alteration in research drill hole as genetic tracers in the upper Cenozoic Y-6, Upper Firehole River, Yellowstone rhyolite-basalt association of the Yellow­ National Park, Wyoming. US Geol. Surv. stonc Plateau voleanic field, U.S.A. J. Prof Pap. 1054-B. 24 pp. Geophys. Res. 87: 4785-4806 Bargar, K. E., Fournier, R. O. 1988. Effects Dzurisin, D., Yamashita, K. M. 1987. of glacial ice on subsurface temperatures Vertical surface displacements at Yellow­ of hydrothermal systems in Yellowstone stone caldera, Wyoming, 1976-1985. J. National Park, Wyoming: fiuid-inclusion Geophys. Res. 92: 13,753-66 evidence. Geology 16: 1077-80 Dzurisin, D., Yamashita, K. M., Johnson, Bargar, K. E., Muffler, L. J. P. 1982. Hydro­ D. J. 1986. Preliminary results of precise thermal alteration in research drill hole Y­ leveling and trilateration surveys in Yel­ II from a vapor-dominated geothermal lows tone National Park, Wyoming, 1985. system at Mud Volcano, Yellowstone US Geot. Surv. Open-File Rep. 86-265-B. National Park, Wyoming. Guideb. Wy o. 28 pp. Geol. Assoc. Ann. Field Conf, 33rd, pp. Eaton, G. P., Christiansen, R. L., Iyer, 2 1 39-5 H. M., Pitt, A. M., Mabey, D. R., et a!. Bargar, K. E., Fournier, R. 0., Theodore, 1975. Magma beneath Yellowstone Nat­ T. G. 1985. Particles in fluid inclusions ional Park. Science 188: 787-96 from Yellowstone National Park-bac­ Fournier, R. O. 1979. Geochemical and teria? Geology 13: 483-86 hydrological considerations and the use of Benz, H. M., Smith, R. B. 1984. Sim­ enthalpy-chloride diagrams in the pre­ ultaneous inversion for lateral velocity diction of underground conditions in hot­ variations and hypocenters in the Yel- spring systems. 1. Volcanol. Geotherm. 10wstone region using earthquake and Res. 5: 1-16 controlled source data. J. Geophys. Res. Fournier, R. O. 1981. Application of water 89: 1208-20 geochemistry to geothermal exploration Bhattacharyya, B. K., Leu, L. K. 1975. and reservoir engineering. In Geothermal Analysis of magnetic anomalies over Systems: Principles and Case Histories, ed. Yellowstone National Park: mapping of L. Rybach, L. J. P. Muffler, pp. 109 43. Curie point isothermal surface for geo­ New York: Wiley thermal reconnaissance. J. Geophys. Res. Fournier, R. O. 1983. Active hydrothermal 80: 4461-65 systems as analogucs of fo ssil systems. In Blank, H. R. Jr., Gettings, M. E. 1974. Com­ The Role of Heat in the Development plete Bouguer gravity map: Yellowstonc­ of Energy and Mineral Resources in the Island Park region. US Geol. Surv. Open­ Northern Basin and Range Province. Geo­ File Map 74-22, scale 1/125,000 thermo Res. Counc. Sp ec. Rep . No. 13, pp. Brokaw, M. A. 1985. Upper crustal interpre­ 263-284

by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. tation of Yellowstone de lerminedfrom ray Fournier, R. O. 1987. Conceptual models of trace modeling of seismic refraction da ta. brine evolution in magmatic-hydro­ Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org MS thesis. Univ. Utah, Salt Lake City. thermal systems. In Hawaiian Volcanism. 180 pp. US Geol. Surv. Prof Pap. 1350, ed. R. W. Christiansen, R. L. 1984. Yellowstone mag­ Decker, T. L. Wright, P. H. Stauffer, pp. matic evolution: its bearing on under­ 1487-1 506 standing large-volume explosive volcan­ Fournier, R. 0. 1988. Hydrothermal activity ism. In Studies in Geophysics: Explosive in relation to late Holocene dynamics in Volcanism: In ception, Evolution, and Haz­ Yellowstone caldera, Wyoming. Proc. ards, pp. 84-95. Washington, DC: Natl. Conf Sci. in Nat!. Parks-Toward the Acad. Press Year 2000, Fo rt Collins, Colo., 1986, Vol. Craig, H., Boat, G., White, D. E. 1956. Iso­ 3, pp. 92-107 tope geochemistry of thermal waters. Na tl. Fournier, R. 0., Pitt, A. M. 1985. The Yel­ Acad. Sci. Nat. Res. Counc. Publ. 400: 29- lowstone magmatic-hydrothermal system. 39 Tr ans. Geotherm. Counc. In t. Symp. Geo- YNP HYDROTHERMAL SYSTEM 51

thermo Energy, ed. C. Stone, lnt. Vol., pp. 1984. Recent uplift of the Yellowstone cal­ 319-27 dera fr om precision gravity measure­ Fournier, R. 0., Potter, R. W. II. 1982. An ments. Eos, Trans. Am. Geophys. Un ion equation correlating the solubility of 65(45): 1118 (Abstr.) quartz in water from 25° to 900°C at pres­ Keenan, J. H., Keys, F. G., Hill, P. G., sures up to 10,000 bars. Geochim. Cosmo­ Moore, J. G. 1969. Steam Tables. New chim. Acta 46: 1969-73 York: Wiley. 162 pp. (Int. Ed .-Metric Fournier, R. 0., White, D. E., Truesdell, Units) A. H. 1976. Convective heat flow in Yel­ Keith, T. E. c., Muffler, L. J. P. 1978. Min­ lowstone National Park. Proc. UN Symp. erals produced during cooling and hydro­ Dev. and Use of Geotherm. Resour., San thermal alteration of ash flow tuff from Francisco, 1975, I: 73 1-39. Washington, Yellowstone drill hole Y-5. J. Vo lcanol. DC: US Govt. Print. Off. Geotherm. Res. 3: 373-402 Fournier, R. 0., Thompson, J. M., Hutch­ Keith, T. E. c., White, D. E., Beeson, M. H. inson, R. A. 1986. Fluctuations in com­ 1978. Hydrothermal alteration and self­ position of Cistern Spring, Norris Geyser sealing in Y-7 and Y-8 drill holes in north­ Basin, Yellowstone National Park, WY - ern part of Upper Geyser Basin, Yel­ variable boiling and mixing 1962- 1985. lowstone National Park, Wyoming. US Extended Ahstr., In t. Symp . Wa ter-Rock Geol. Surv. Prof Pap. I054-A. 26 pp. In teraction, 5th, Reykjavik, Iceland, pp. Kennedy, B. M., Smith, S. P., Reynolds, 206-9 J. H. 1982. Noble gases in Yellowstone Gooch, F. A., Whitfield, J. E. 1888. Analyses Park. Eos, Trans. Am. Geophys. Union 63: of waters of the Yellowstone National 1145 (Abstr.) Park, with an account of the methods of Kennedy, B. M., Lynch, M. A., Reynolds, analysis employed. US Geol. Surv. Bull. J. H., Smith, S. P. 1985. Intensive sam­ 47. 84 pp. pling of noblc gases in fluids at Y cJlow­ Gunter, B. D. 1973. Aqueous phase-gaseous stone: I. Early overview of the data; re­ phase material balance studies of argon gional patterns. Geochim. Cosmochim. and nitrogen in hydrothermal fe atures at Acta 49: 125 1-61 Yellowstone National Park. Geochim. Kennedy, B. M., Reynolds, J. H., Smith, Cosmochim. Acta 37(3): 495-5 13 S. P., Truesdell, A. H. 1987. Helium iso­ Gunter, B. D., Musgrave, B. C. 1966. Gas topes: Lower Geyser Basin, YeJlow­ chromatographic measurements of hydro­ stone National Park. J. Geophys. Res. 92: thermal emanations at Yellowstone Na­ 12,477-89 tional Park. Geochim. Cosmochim. Acta Kim, K., Craig, H. 1979. Radioactivities of 30: 1175-89 hot springs in Yellowstone Park. Eos, Hamilton, W. L. 1985. Deformation models Tr ans. Am. Geophys. Union 60(46): 944- for interpreting warped Holocene terraces 45 (Abstr.) of Yellowstone Lake and the Yellowstone Leeman, W. P., Doe, B. R., Welhan. J. 1977. River at the Yellowstone caldera, Wyom­ Radiogenic and stable isotope studies of ing. E05, Trans. Am. Geophys. Un ion 66 hot-spring deposits in Yellowstone Na­ (46): 853 (Abstr.) tional Park and their genetic implica­ Hardee, H. C. 1982. Permeable convection tions. Geochem. J. 11: 65-74 above magma bodies. Te ctonophysics 84: Lehman, J. A., Smith, R. B., Schilly, M. Moo 179-95 Braile, L. W. 1982. Uppcr crustal structure Hildreth, W., Christiansen, R. L., O'Neil, of the Yellowstone caldera from delay J. R. 1984. Catastrophic isotopic modi­ time analyses and gravity correlations. J. fication of rhyolitic magma at times of Geophys. Res. 87: 2713-30 caldera subsidence, Yellowstone Plateau Lister, C. R. B. 1974. On the penetration of by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. volcanic field. J. Geophys. Res. 89: 8339- water into hot rock. Geophys. J. R. As/ron.

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org 69 Soc. 39: 465-509 Honda, S., Muffler, L. J. P. 1970. Hydro­ Lister, C. R. B. 1980. Heat flow and hydro­ thermal alteration in core from research thermal circulation. Ann. Rev. Earth drill hole Y-I, Uppcr Geyscr Basin, Yel­ Planet. Sci. 8: 95-117 lowstone National Park, Wyoming. Am. Lister, C. R. B. 1983. The basic physics of Mineral. 55: 1714-37 water penetration into hot rock. In Hydro­ Iyer, H. M., Evans, J. R., Zandt, G., Stewart, thermal Processes at Seafloor Sp reading R. M., Cookley, J. M., Roloff, J. N. 1981. Centers, ed. P. A. Rona, K. Bostrom, L. A deep low-velocity body under the Yel- Laubier, K. L. Smith, Jr., pp. 141-68. New 10wstone caldera, Wyoming: delineation York: Plenum using teleseismic P-wave residuals and tec­ Love, J. D., Good, J. M. 1970. Hydro­ tonic interpretations. Geol. Soc. Am. Bull. carbons in thermal areas, northwestern 92: 792-98 Wyoming. US Geol. Surv. Prof Pap . 644- Jackson, S. M., Smith, R. B., Olson, T. L. B. 23 pp. 52 FOURNIER

Mazor, E., Fournier, R. O. 1973. More on Sum. Prof Pap . 435, pp. 179-84 noble gases in Yellowstone National Park Sleep, N. H. 1983. Hydrothermal convection hot waters. Geochim. Cosmochim. Acta at ridge axes. In Hydrothermal Processes 37(3): 515-25 at Seafloor Sp reading Centers, ed. P. A. Mazor, E., Thompson, J. M. 1982-. Evo­ Rona, K. Bostrom, L. Laubier, K. L. lution of geothermal fluids deduced from Smith, Jr., pp. 71-82. New York: Plenum chemistry plots: Yellowstone National Smith, R. B., Braile, L. W. 1984. Crustal Park (U.S.A.). J. Volcanol. Geotherm. Res. structure and evolution of an explosive 12: 351--60 silicic volcanic system at Yellowstone Mazor, E., Wasserburg, G. J. 1965. Helium, National Park. In Studies in Geophysics: neon, argon, krypton, and xenon in gas Explosive Vo lcanism: In ception, Evolution, emanations from Yellowstone and Lassen and Ha zards, pp. 96--109. Washington, Volcanic National Parks. Geochim. DC: Nat!. Acad. Press Cosmochim. Acta 29: 443-54 Smith, R. B., Bruhn, R. L. 1984. Intraplate McKenzie, W. F., Truesdell, A. H. 1977. expansional tectonics of the Eastern Geothermal reservoir temperatures esti­ Basin-Range: inferences on structural mated from the oxygen isotope com­ style from seismic reflectiondata, regional positions of dissolved sulfate and water tectonics, and thermal-mechanical models from hot springs and shallow drill holes. of brittlc-ductile deformation. J. Geophys. Ceothermics 5: 51-62 Res. 89: 5733 62 Meertens, c., Levine, J. 1985. Compressive Smith, R. B., Christiansen, R. L. 1980. Yel­ tectonic strain as a possible mechanism lowstone Park as a window on the earth's for long-term vertical deformation of the interior. Sci. Am. 242: 104-17 Yellowstone caldera. Eos, Trans. Am. Smith, R. B., Shuey, R. T., Freid1ine, R., Ceophys. Un ion 66(46): 853 (Abstr.) Otis, R., Alley, L. 1974. Yellowstone hot Meyer, G. A., Lock, W. K. 1986. Origin spot: new magnetic and seismic evidence. and deformation of Holocene shoreline Geology 2: 45 1-55 terraces, Yellowstone Lake, Wyoming. Stanley, W. D., Boehl, J. E., Bostick, F. X., Geology 14: 699-702 Smith. H. W. 1977. Geothermal signif­ Morgan, P., Blackwell, D. D., Spafford, icance of magnetotelluric sounding in R. E. 1977. Heat flow measurements in the eastern Snake River Plain-Yellow­ Yellowstone Lake and the thermal struc­ stone region. J. Geophys. Res. 82: 2501- ture of the Yellowstone caldera. J. Geo­ 14 phys. Res. 82: 3719-32 Sturchio, N. c., Binz, C. M., Lewis, C. H. Muffler, L. J. P., White, D. E., Truesdell, III. 1 987. Thorium-uranium disequilib­ A. H. 1971. Hydrothermal explosion cra­ rium in a geothermal discharge zone ters in Yellowstone National Park. Geol. at Yellowstone. Geochim. Cosmochim. Soc. Am. Bull. 82: 723--40 Acta 51: 2025-34 Noguchi, K., Nix, J. 1963. Geochemical Thompson, J. M., Hutchinson, R. A. 1980. studies of some geysers in Yellowstone Boundary Creek thermal area of Yel- National Park. Proc. Jp n. Acad. 39: 370- 10wstone National Park: II. Thermal 75 water analyses. Trans. Geotherm. Resour. Norton, D. R., Friedman, I. 1985. Chloride Counc. 4: 189-92 flux out of Yellowstone National Park. J. Thompson, J. M., Yadav, S. 1979. Chemical Volcanol. Geotherm. Res. 26: 231-50 analyses of waters from geysers, hot Pelton, J. R., Smith, R. B. 1979. Recent crust­ springs and pools in Yellowstone National al uplift in Yellowstone National Park. Park, Wyoming, from 1974-1978. US Science 206: 1179-82 Geol. Surv. Open-File Rep. 79-704.49 pp. Thompson, J. M., Presser, T. S., Barnes, by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Peiton, J. R., Smith, R. B. 1982. Con­ temporary vertical surface displacements R. B., Bird, D. B. 1975. Chemical analy­

Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org in Yellowstone National Park. J. Geophys. sis of the waters of Yellowstone National Res. 87: 2745-5 1 Park, Wyoming, from 1965-1973. US Rowe, J. J., Fournier, R. 0., Morey, G. W. Ceol. Surv. Open-File Rep. 75-25. 59 pp. 1973. Chemical analysis of thermal waters Truesdell, A. H., Fournier, R. O. 1976. Con­ in Yellowstone National Park, Wyoming, ditions in the deeper parts of hot spring 1960--1965. US Ceol. Surv. Bull. 1303. 31 systems of Y cllowstonc National Park, pp. Wyoming. US Geol. Surv. Open-File Rep. Schoen, R., Rye, R. O. 1970. Sulfur isotope 76-428. 29 pp. distribution in solfataras, Yellowstone Truesdell, A. H., Thompson, J. M. 1982. The National Park. Science 170: 1082-84 geochemistry of Shoshone Geyser Basin, Scott, R. C. 1964. Records of post­ Yellowstone National Park. Guideb. Wy o. earthquake chemical quality of ground Geol. Assoc. Ann. Field Conf, 33rd, pp. water. In The Hebgen Lake Montana 153-59 Earthquake of August 17, 1959. US Ceol. Truesdell, A. H., Nathenson, M., Rye, R. O. YNP HYDROTHERMAL SYSTEM 53

1977. The effects of subsurface boiling and White, D. E., Fournier, R. 0., Muffler, dilution on the isotopic compositions of L. J. P., Truesdell, A. H. 1975. Physical Yellowstone thermal waters. J. Geophys. results of research drilling in thermal areas Res. 82: 3694-3704 of Yellowstone National Park, Wyo­ Welhan, J. A. 1981. Carbon and hydrogen ming. US Geol. Surv. Prof Pap. 892. gases in hydrothermal systems; the search 70 pp. fo r a mantle source. PhD thesis. Univ. Yellowstone National Park Task Group. Calif., San Diego, La Jolla. 216 pp. 1987. The Objectives fo r Deep Scientific White, D. E. 1978. Conductive heat flows Drilling in Ye llowstone Na tional Park. in research drill holes in thermal areas of Washington, DC: Natl. Acad. Sci. 69 pp. Yellowstone National Park, Wyoming. Zohdy, A. A. R., Anderson, L. A., Muffler, US Geol. Surv . J. Res. 6: 765-74 L. J. P. 1973. Resistivity, self-potential, White, D. E., Muffler, L. J. P., Truesdell, induced-polarization and surveys of a A. H. 1971. Vapor-dominated hydrother­ vapor-dominated geothermal system. mal systems compared with hot-water sys­ Geophysics 38(6): 1130-44 tems. Econ. Geol. 66: 75-97 by RICE UNIVERSITY - FONDREN LIBRARY on 02/27/10. For personal use only. Annu. Rev. Earth Planet. Sci. 1989.17:13-53. Downloaded from arjournals.annualreviews.org