Address reprint requests to T.M. Gerlach (tgerlach@usgs.gov)
Herein, we report the results and interpretation of SO2 and CO2 emission measurements obtained at Crater Peak from late 1991 through early 1993, a period that included three subplinian eruptions. (For the eruption chronology, see Keith and others, this volume) (-- Web note: not available) Sulfur dioxide measurements were taken at Mount Spurr volcano in July of 1991 as part of the Alaska Volcano Observatory (AVO) gas-monitoring program for the volcanoes of the Cook Inlet. However, few SO2 flux measurements were made until after the third eruption on September 16-17, 1992, because of budget constraints, the surprisingly low SO2 flux observed at Crater Peak following eruptions on June 27 and August 18, 1992, and the perception that the activity would cease soon after the initial eruptions, as it did during the 1953 eruption at Crater Peak. Shortly after the September 16-17 eruption, AVO intensified SO2 monitoring and added CO2 flux determinations in the gas-monitoring program at Mount Spurr volcano. The CO2 measurements were motivated by the working hypotheses that (1) both SO2 and CO2 were degassed noneruptively from magma and that (2) subsequent dissolution in liquid water within the volcano resulted in scrubbing of SO2 by hydrolysis reactions while CO2 remained available for emission when the liquid boiled.
Measured SO2 emission rates can vary owing to natural causes such as variation in emission rate, changes in sun angle, and amount of ultraviolet light and plume opacity due to the presence of suspended ash. Uncertainty of COSPEC measurements arises from several operational factors including operator variance (pilot, instrument manipulation, data reduction) and measurement of plume velocity (Casadevall and others, 1994).
At Crater Peak, the accurate measurement of plume velocity was usually the largest source of uncertainty. We assumed that the plume was carried along at the same speed as the ambient winds. Wind direction and speed near the vent are controlled by adjacent mountainous topography. Winds passing over the location of Crater Peak in this part of the Alaska Ranges can be turned, shifted, and back-eddied. Local wind directions were occasionally 180o from the regional trend reported by the National Weather Service (NWS) in Anchorage, Alaska. Wind speed and direction were determined in several ways. One method relied on wind and speed determinations made twice daily at Anchorage International Airport by the NOAA radiosonde. One of these measurements is made at 2:00 p.m. local time, which coincides with local solar noon (highest ultraviolet levels) and the normal times of the SO2 flights. When forecast winds were light and variable, we assigned 5 knots as the operative value. Another method was to compare onboard navigational instruments or global positioning satellite (GPS) receivers with NWS forecasted winds aloft.
Aircraft speed was determined in several ways. During close orbits of the vent, geographical landmarks were identified and noted on the SO2 chart recorder. The chart recorder was then used as a clock to time one orbit. Using the distance from the vent as a radius, the aircraft ground speed was calculated. At other times, radar tracking from Anchorage International Airport was used to calibrate onboard instruments for reduction of aircraft airspeed to ground speed. Ideally, a moderately high emission rate is desirable (greater than 500 t/d of SO2) along with wind speeds of nearly 10 nautical miles per hour. During times of low SO2 flux or low wind speeds, orbits within a kilometer of the vent were required. Plume positions during these times, or when wind directions drove the plume into mountainous terrain, would not allow the aircraft to fly safely beneath the entire plume; these alternate routes resulted in minimum values being measured for SO2.
The SO2 and CO2 emission rate data obtained for 1991 through 1993 are presented in table 1. The data as a time series that includes the three 1992 eruptions of June 27, August 18, and September 16-17 are shown in figure 3; periods of low UV radiation are also shown in this figure. Data obtained after the September 16-17 eruption when COSPEC flights were more frequent and the MIRAN instrument was used for CO2 measurements are shown in figure 4. All emission rate data are reported in metric tons/day (t/d) above the background level upwind of the volcano.
[Table,38K,PS]
Table 1.—Sulfur dioxide and carbon dioxide fluxes measured at Crater
Peak, Mount Spurr volcano, Alaska, 1991-1993
COSPEC measurements did not detect significant increases above background in SO2 emissions prior to the June 27, 1992, eruption. Some COSPEC flights made during times of increased seismicity recorded SO2 emissions slightly elevated above background levels. For example, two measurements in May and June were 88 and 21 t/d, respectively, during a period of increased seismic activity. COSPEC measurements on June 29, just 2 days after the first eruption, showed SO2 emission rates of only 5 t/d. Measurements on September 10, a week prior to the September 17 eruption, were at background levels, and measurements 4 and 6 days after the eruption were only slightly above background (23 and 24 t/d). Strong H2S odors were frequently noted during airborne monitoring flights throughout the course of the 1992 eruptions. The periods of strongest H2S odor correlated with times of minor to background SO2 emission rates.
On September 24, SO2 emission rates reached 79 t/d. By September 25, they rose to 300 t/d, just as the first CO2 measurements showed 11,000 t/d. Measurements on September 28 showed CO2 at 12,000 t/d. Sulfur dioxide values peaked on September 29 at 750 t/d, coincident with a CO2 flux of 8,700 t/d.
In the week following the October 2-5 tremor episode, SO2 remained above background levels, although it declined somewhat from 450 to 240 t/d, as CO2 declined from 4,800 to 2,900 t/d. During this period, aerosol plumes of sulfate fume derived from volcanic SO2 became visible for the first time downwind from the crater. These plumes extended beyond the water vapor condensation cloud near the vent and spread out along an inversion layer in the atmosphere (fig. 5).
On October 3, 4, and 5, SO2 fluxes were measured before and after the onset of volcanic tremor. After about 2 hours into each tremor episode, SO2 emission rates decreased to approximately half their pre-tremor values (fig. 6). Sulfur dioxide emission rates, but not CO2 rates, were also down sharply after the tremor episode on October 12. Measurable SO2 levels gradually declined over the next 2 weeks and returned to near background levels by the end of October 1992. Three of five CO2 measurements over the same period gave emission rates over 2,000 t/d; the other two measurements showed background levels.
Both MIRAN and COSPEC surveys were made of the atmosphere around the volcano immediately after the November 9 earthquake swarms. Sulfur dioxide emission rates never rose above background levels following the crisis, but CO2 emissions rose above background to a level of about 1,000 t/d by the third day after the crisis and stayed at about this level through April 1993. Occasional later measurements showed lower CO2 levels of only a few hundred tons per day, which may represent the normal baseline CO2 fluxes from Crater Peak at times of inactivity.
Integrating the noneruptive CO2 emission rate data for the period September 25, 1992, to April 24, 1993 (fig. 4) gives a total CO2 emission of approximately 110 kt. A similar treatment of the COSPEC SO2 flux data results in a total SO2 emission of only about 6 kt.
The great difference between noneruptive and eruptive SO2 degassing may reflect the role of liquid water at Crater Peak in hydrolyzing most of the noneruptively emitted SO2 gas; an abundance of water may also have played a role in effectively quenching shallow intrusions during their ascent to the surface. After obtaining an average SO2 emission rate of only 5 t/d (table 1) at Crater Peak just 2 days after the June 27 eruption, we speculated that the extremely low, noneruptive SO2 emissions resulted from the hydrolysis of SO2 to aqueous H2S and sulfate by liquid water present in the volcano. One likely source of water is a liquid-dominated hydrothermal system postulated to exist between 1.5 and 2 km beneath Crater Peak (Motyka and Nye, 1993). The presence of a hydrothermal system does not, however, guarantee that SO2 emissions from shallow magma will be effectively absorbed. For example, Mount Pinatubo contained a hydrothermal system sufficiently impressive to attract exploration drilling for geothermal energy in 1988-89 (Delfin and others, 1992). Nevertheless, SO2 emission rates at Mount Pinatubo reached values as high as 13 kt/d in the weeks before the climactic June 15, 1991, eruption (Daag and others, 1994). The Mount Pinatubo hydrothermal system had a low permeability--too low for geothermal-energy production (Delfin and others, 1992)--which may have inhibited its capacity to rapidly recharge water lost from boiling and to prevent (or greatly diminish) pre-eruption SO2 emissions. We suggest that, in addition to the presence of a hydrothermal system, important factors contributing to SO2 scrubbing at Crater Peak may have included a highly permeable volcanic edifice and large sources of liquid water from snow melt and melt from glaciers directly above and surrounding Crater Peak, thus permitting rapid and abundant recharge of water to the volcano. A km-long zone of warm springs (40 oC) discharges from the base of the volcano about 4 km directly downslope from Crater Peak. Total warm-water flow from this system is estimated to be approximately 1,000 l/min, and stable isotope data indicate that the meteoric recharge for the groundwater system supplying the springs is derived from local snow melt and precipitation falling at middle to upper elevations (Motyka and Nye, 1993).
The main gases, in addition to water vapor, released from shallow degassing magma of convergent-plate volcanic systems are CO2, SO2, H2S, and HCl. All these gases are soluble in liquid water that may be present in volcanic edifices. When the water boils, the dissolved gases will partition preferentially into the vapor phase. The dissolution of gases in water and their partitioning between vapor and liquid during boiling are topics beyond the intended scope of this report. We refer the reader to a discussion of these and related topics in Henley and others (1984); much of the interpretation that follows is based on data and concepts they present. We focus entirely on liquid-vapor interactions. In most cases, water-rock interactions are likely to proceed at a slow rate relative to the time scales of volcanic degassing and liquid-vapor reactions during boiling. We assume, therefore, that water-rock interactions introduce only secondary effects that can be ignored here.
Because aqueous HCl is a strong acid, it dissociates in water to hydrogen and chloride ions (HCl(aq) ===> H+(aq) + Cl-(aq)). The concentration of HCl(aq) will therefore be negligible, and its exsolution as HCl gas in the vapor phase during boiling will be negligible, except for unusually acidic solutions at temperatures much greater than 250 oC. Thus, water in a volcano will dissolve HCl released from degassing magma, but because of its strong dissociation, HCl will not normally exsolve to a vapor phase when the resulting solution is boiled.
Aqueous SO2 is thermodynamically unstable in water relative to sulfate species and aqueous H2S, except at temperatures greater than 300 oC and pH values less than or equal to 2. The hydrolysis of SO2 by the disproportionation reaction:
(1) 4H2O(l) + 4SO2(aq) ===> H2S(aq) + 3H+(aq) + 3HSO4-(aq)proceeds strongly in the direction indicated. Consequently, the concentration of SO2(aq) is very low and the contribution of SO2 to the vapor phase during boiling is negligible compared to the bulk amount of SO2 gas that may have been injected initially into the water during magma degassing.
The gases CO2 and H2S dissolve in water as CO2(aq) and H2S(aq). Their dissociation (H2O(l) + CO2(aq) = H+(aq) + HCO3-(aq); H2S(aq) = H+(aq) + HS-(aq)) is not appreciable in ground water and hydrothermal waters provided pH is below 9. Unlike SO2, they are relatively stable with respect to water, although some H2S(aq) may react with dissolved Fe+2 to form pyrite (FeS2). The concentrations of CO2(aq) and H2S(aq) in waters that have absorbed magmatic gases are, therefore, significant compared to those for aqueous HCl and SO2. Carbon dioxide and H2S are thus distributed between vapor and liquid phases during boiling. The distribution of either gas between vapor and liquid can be represented by a distribution coefficient, Bi, defined as follows:
(2) Bi = [ni/nH2O]vap/[ni/nH2O]liqwhere [ni/nH2O]vap is the molar ratio of the gas i to H2O in the vapor, and [ni/nH2O]liq is the ratio of the gas to H2O in the liquid. Giggenbach (1980) derived the following regression equations, valid for 100 to 340 oC, from experimental gas-solubility data for CO2 and H2S in water:
(3) log BCO2 = 4.7593 - 0.01092T (4) log BH2S = 4.0547 - 0.00981Twhere T is in oC. Over temperatures from 100 to 340 oC, the respective values for BCO2 range from 4,650 to 11; those for BH2S range from 1,185 to 5. Thus, both CO2 and H2S will move preferentially into the vapor phase during boiling, and this preference is stronger the lower the temperature of boiling.
With equations (3) and (4) and other equations provided in Henley and others (1984), it is possible to calculate the distribution of CO2 and H2S between vapor and liquid as boiling progresses. The results depend on the temperature of boiling, the extent of boiling (that is, the fraction of the total water in the system transferred to the vapor phase as steam), the isothermal or isoenthalpic nature of the boiling process, and the open or closed character of the system. The percent of CO2 and H2S remaining dissolved in the liquid, relative to the initial amounts in solution, as a function of the percent isoenthalpic boiling for a closed system initially at a temperature of 260 oC are shown in figure 7. For these conditions, only 25 percent of the initial CO2 and 43 percent of the initial H2S remain dissolved in the liquid after just 3 percent boiling. The temperature after 3 percent isoenthalpic boiling is 250 oC. The curves produced by closed-system, isothermal boiling at 260 oC are similar but somewhat above those shown in figure 7 for the same range of percent boiling. For example, 29 percent of the initial CO2 (instead of 25 percent) remains in the liquid after 3 percent boiling. Because the values for BCO2 and BH2S increase as the system cools, CO2 and H2S will partition into the vapor at a faster rate from isoenthalpic boiling than from isothermal boiling. In open-system boiling, gases are removed from solution at exponential rates and cause marked changes in gas content of the liquid after only a few percent boiling. If open-system rather than closed-system conditions are assumed, less than 5 and 30 percent of the initial CO2 and H2S, respectively, remain dissolved in the liquid after just 3 percent boiling. At lower temperatures (higher Bi values), less boiling is required to produce the same changes in the gas content of the liquid for both open and closed systems. In general, boiling is more effective in stripping CO2 and H2S from aqueous solutions the lower the temperature of boiling and the more open and isoenthalpic the behavor of the system during boiling.
Our observations on SO2, CO2, and H2S emissions during the 1992 eruptions of Crater Peak are consistent with the foregoing discussion of SO2 hydrolysis and the partitioning of CO2 and H2S during boiling. We suggest that scrubbing of SO2 by liquid water in the volcano effectively masked noneruptive magmatic emissions of SO2 from the time of precursory seismic activity in August of 1991 to September 24, 1992 (table 1). We particularly stress this interpretation for the COSPEC measurements from June 8 to September 25, 1992, which show SO2 emissions of background to only 88 t/d at times before, after, and between the three eruptions. At these times, SO2 emissions of at least several hundred t/d would normally be expected on the basis of past experience (Casadevall and others, 1981, 1983, 1994; McGee, 1992). In addition, the hydrolysis of SO2 accounts for the increase in the sulfate content of the Crater Peak lake water prior to the first eruption. The persistently strong H2S odor during this period, and during later periods when SO2 emissions were at or near background levels, is also consistent with this interpretation. In addition to direct dissolution of H2S degassed from shallow magma, the hydrolysis of aqueous sulfur dioxide by reaction (1) converts 25 percent of the degassed SO2 into aqueous H2S. Because magmas generally emit much more SO2 than H2S, the production of H2S by hydrolysis probably was significantly greater than the H2S from magmatic degassing. Aqueous H2S derived both from direct magma degassing and from hydrolysis of degassed SO2 would be preferentially partitioned into the vapor phase and released in emissions during boiling of water. Hydrogen sulfide has a lifetime in the atmosphere of only about 1 day, because of oxidation reactions (Graedel, 1977). Thus, while it is possible that the minor (5-88 t/d) SO2 fluxes during the June 8 through September 25 period represent magmatic degassing of SO2 through rare dry pathways to the surface at this time, we cannot rule out the atmospheric oxidation of H2S emissions from boiling water as the source of the detected SO2.
Large SO2 emissions were possible during the three explosive eruptions because magma ascended rapidly to the surface and could degas directly to the atmosphere. This effectively prevented contact with liquid water and loss by scrubbing of most of the SO2 released during the eruptions. The TOMS data indicate that 15 to 25 percent of the sulfur released in the three eruptions was emitted as H2S (Bluth and others, this volume) ( -- Web note: not available). This suggests the possibility that H2S emissions also occurred from the boiling of water during the explosive events. Hence some of the sulfur released in the explosive eruptions may have been degassed noneruptively as SO2 at an earlier time, converted to aqueous H2S by hydrolysis, and then released during boiling from heating associated with the ascent of the erupting magma.
On September 25, about a week after the September 16-17 eruption, the system finally dried out enough to permit SO2 fluxes in excess of 100 t/d (table 1). The week of time may have been necessary to transfer the heat required to dry out a sufficient volume of the rock surrounding the shallow magma so that significant amounts of SO2 could degas and escape to the atmosphere without encountering liquid water. It is possible that some additional shallow magma was supplied during this week; this magma augmented the supply of magmatic gas and heat needed to create and maintain a dry escape route to the surface. We suggest that during this period the liquid-vapor boiling surface migrated outward away from the shallow magma, because the rate at which heat was supplied for boiling outpaced the rate at which recharging water removed heat.
The CO2 emission rate data (table 1) indicate that magma was degassing, and these data support the hypothesis of SO2 loss by scrubbing. Together, the CO2 and SO2 fluxes from September 25 to about October 10 indicate that magma was degassing through boiling water as well as through a zone of dry rock. The values of CO2/SO2 weight ratios for magmatic gases from convergent-plate volcanic systems are <15, except for a few samples collected from degassed domes (e.g., Showa-Shinzan) several years after dome emplacement (Symonds and others, 1994). Most of the CO2/SO2 values at Crater Peak are in the 10-100 range (table 1). A higher CO2/SO2 is expected if a significant fraction of the degassing takes place through boiling water. Magma degassing through boiling water would cause relatively minor fractionation of CO2, because of its tendency to partition preferentially into a vapor phase, compared to its effect on SO2, which is susceptible to hydrolysis. Consequently, the CO2/SO2 values of the gases emitted to the atmosphere will be higher than the magmatic gas values, and they will increase as the fraction of degassing through boiling water increases relative to degassing through a zone of dry rock. The high CO2 emission rates of 11 and 12 kt/d at times of relatively modest (190 to 300 t/d) SO2 emissions and the high CO2/SO2 values of 53 to 90 on September 25 and 28 suggest that significant degassing of magma through water was also occurring in the week immediately after the September 16-17 eruption, even though SO2 fluxes were then very low (<100 t/d), presumably because of SO2 scrubbing by boiling water.
Sulfur dioxide emission rates peaked near the end of September, only about a week after the dry-out zone was established. By early October, removal of heat by recharge of meteoric water apparently began to outpace the supply of magmatic heat again, the dry escape route through the wet edifice began to close, and SO2 emissions started to drop. The exhaustion of gas, as well as heat, from the shallow magma probably contributed to the drop in SO2 emission rate. The coincident decline in CO2 also suggests the exhaustion of volatiles from the shallow magma.
We attribute the drop in SO2 emission rates during tremor episodes in early October (fig. 6) to the recharge of liquid water back into the dry-out zone around and above the shallow magma. As water invaded fractures within this zone, vigorous boiling ensued, presumably giving rise to tremor, and the adjacent hot rock and melt cooled sufficiently to permit the presence of an increasing amount of liquid water. As the process proceeded, SO2 was increasingly lost by scrubbing, SO2 emissions dropped, and CO2/SO2 and H2S/SO2 values rose. After the tremor crisis of October 12, the system remained effectively waterlogged without a dry zone for SO2 escape. Sulfur dioxide emissions declined to background levels where they have remained.
Carbon dioxide emissions of greater than 2 kt/d suggest that episodes of boiling continued to release CO2 and H2S after the October 12 tremor crisis. Carbon dioxide emissions and strong H2S odors were evident again 3 days after the November 9 earthquake swarms. We speculate that these swarms were related to one or more intrusions too small to penetrate through the water-saturated zone but sufficiently large to produce thermal perturbations that led to boiling and release of CO2 and H2S. Carbon dioxide fluxes of 1 to 3 kt/d continued for over 2 months. A measurement on April 24, 1993, still showed a flux of about 1 kt/d. We suspect these emission levels signify continuing restlessness, because measurements on July 7, 1994, did not detect CO2 above the background level (Doukas, 1995).
We strongly recommend early monitoring of CO2 when Cook Inlet volcanoes become restless. Because of its strong preference for the vapor phase during boiling, CO2 emissions from degassing magma are less likely to be masked by the presence of water, whereas SO2 emissions may be lost totally from interactions with water and thus render misleading COSPEC results. We also recommend that CO2 emission rate data sets be obtained for inactive Cook Inlet volcanoes so that baseline fluxes will be available for comparison when these volcanoes become restless.
REFERENCES CITED Casadevall, T.J., Johnston, D.A., Harris, D.M., Rose, W.I., Malinconico, L.L., Stoiber, R.E., Bornhorst, T.J., Williams, S.N., Woodruff, L. and Thompson, J.M., 1981, SO2 emission rates at Mount St. Helens from March 29 through December, 1980, in Lipman, P.W., and Mullineaux, D.R., eds., The 1980 eruptions of Mount St. Helens, Washington: U.S. Geological Survey Professional Paper 1250, p. 193-200. Casadevall, T. J., Rose, W. I., Gerlach, T. M., Greenland, L. P., Ewert, J., Wunderman, R., and Symonds, R., 1983, Gas emissions and the eruptions of Mount St. Helens through 1982: Science, v. 221, p. 1383-1385. Casadevall, T.J., Doukas, M.P., Neal, C.A., McGimsey, R.G., and Gardner, C.A., 1994, Emissions rates of sulfur dioxide and carbon dioxide from Redoubt Volcano, Alaska during the 1989- 1990 eruptions: Journal of Volcanology and Geothermal Research, v. 62, nos. 1-4, p. 519-530. Daag, A., Tubianosa, B., Newhall, C., Tungol, N., Javier, D., Dolan, M., de los Reyes, P.J., Arboleda, R., Martinez, M., and Regalado, M.T.M., 1995, Monitoring sulfur dioxide emissions at Mount Pinatubo volcano, in Newhall, C. G., and Punonbayan, R. S., eds. 1995, Fire and mud: Eruptions and lahars of Mount Pinatubo, Philippines: PHIVOLCS and University of Washington Press, in press. Delfin, F.G., Sussman, D., Ruaya, J.R., and Reyes, A.G., 1992, Hazard assessment of the Pinatubo volcanic-geothermal system: Clues prior to the June 15, 1991 eruption: Transactions Geothermal Resources Council, v. 16, p. 519-528. Doukas, M.P., 1995, A compilation of sulfur dioxide and carbon dioxide emission-rate data from Cook Inlet volcanoes (Redoubt, Spurr, Iliamna, and Augustine), Alaska during the period from 1990 to 1994: U.S. Geological Survey Open-File Report 95-55, 15 p. Giggenbach, W.F., 1980, Geothermal gas equilibria: Geochimica et Cosmochimica Acta, v.44, p. 2021-2032. Graedel, T.E., 1977, The homogeneous chemistry of atmospheric sulfur: Reviews of Geophysics and Space Physics, v. 15, p. 421- 428. Harris, D.M., Sato, M., Casadevall, T.J., Rose, W.I., JR., and Bornhorst, T.J., 1981, Emission rates of CO2 from plume measurements, in Lipman, P. W., and Mullineaux, D. R., eds., The 1980 eruptions of Mount St. Helens, Washington: U.S. Geological Survey Professional Paper 1250, p. 201-207. Henley, R.W., Truesdell, A.H., Barton, P.B., Jr., and Whitney, J.A., 1984, Fluid-mineral equilibria in hydrothermal systems, in Robertson, J.M., ed., Reviews in Economic Geology, v. 1, El Paso, Tx., Society of Economic Geologists, 267 p. McGee, K. A., 1992, The structure, dynamics, and chemical composition of noneruptive plumes from Mount St. Helens, 1980- 88: Journal of Volcanology and Geothermal Research, v. 51, p. 269-282. Motyka, R.J., and Nye, C.J., 1993, Fumarolic gas chemistry (1982) and thermal spring water chemistry, Crater Peak (1985), Mount Spurr, Alaska, in Solie, D.N., and Tannian, F., eds., Short notes on Alaskan geology, 1993: Alaska Division of Geological and Geophysical Surveys Professional Report 113, p. 31-40. Schnetzler, C.G., Doiron, S.D., Walter, L.S., and Krueger, A.J., 1994, Satellite measurements of sulfur dioxide from Redoubt eruptions of 1989-1990: Journal of Volcanology and Geothermal Research, v. 62, nos. 1-4, p. 353-358. Symonds, R. B., Rose, W. I., Bluth, G. J. S., Gerlach, T. M., 1994, Volcanic-Gas studies: methods, results, and applications, in Carroll, M. R., and Holloway, J. R., eds., Volatiles in Magmas: Mineralogical Society of America Reviews in Mineralogy, v. 30, p. 1-66.
