An estimate of gas emissions and magmatic gas content from Kilauea volcano

Gmchimica 0 Pqunot8

n Cosmochimica Pres

.&a

Ltd. 1985. Pnnd

An esti~te

0016-7037/85/13.00

Vol. 49. &w. 125-129

+ .30

in U.S.A.

of gas missions

and rn~~tic

gas content from Kibea

volcano

L. P. GREEN~.AND, W. I. ROSE* and J. B. STOKES U.S. &c&&al Survey, Hawaiian Volcano Observatory, P.O. Box 5 I, Hawaii National Park, Hawaii 96718 (Received June 26, 1984; accepted in revisedform October 3, 1984) Abstract-Emission rates of CO2 have heen measured at Kilauea volcano, Hawaii, in the east-rift eruptive plume, and CO2 and SO2 have been measured in the plume from the noneruptive fumamles in the summit caldera. These data yield an estimate of the loading of Kilauean eruptive gases to the atmosphere and suggest that such estimates may be inferred dinctly from measured hxva volumes. These data, combined with other chemical and geologic data, suggest that magma arrives at the shallow summit reservoir containing (wt.%) 0.32% J&O, 0.32% COr, and 0.09% S. Magma is rapidly degassed of most of its CO2 in the shallow reservoir before transport to the eruption site. Because this summit degassing yields a magma saturated and in ~uili~um with volatile species and because transport of the magma to the eruption site oaxrs in a zone no shallower than the summit reservoir, we suggest that eruptive gases fmm Kilauea c~~e~~~ly shotrId be one of two types: a ‘primary’ gas from fresh magma derived directly from the mantle and a carbon-depleted gas from magma stored in the summit reservoir. INTRORUCI’ION SUMMIT area of Rilauea volcano is occupied by a caldera. The Halemaumau pit crater occurs within the caidera and, histo~~Iy, has been Kilauea’s most active vent. Most of the non-enrptive degassing of Kiiauea originates from fumaroles Iocated within and adjacent to Halemaumau. Extending from the summit are a southwest rift zone and an east rift zone; eruptions of Kilauea are confined to the summit area and the rift zones. Rilauean ma8ma originates at SO60 km depth (EATON, 1962), moves upward to a shallow summit reservoir extending from 6-7 km to 2-3 km depth (KOYANAGf et al_. 1974), and is transported down the east rift zone through a conduit at 5-10 km depth (KOYANAGI et al.. 1981). The current east-rift eruption of K&mea volcano, Hawaii, began January 3, 1983, and quickly formed a line of eruptive fissures extending about 6 km. The eruption has continued episodically to date (April 15, 1984), but has been limited to a single vent (Pu’u 0) about 19 km from the summit since May 1983. The 15th major episode of this series of eruptions began on February 14, 1984, and lasted for 19 hours; the 16th episode began March 3, 1984, and lasted for 3 I hours. During these eruptions we used an airborne infrared spectrometer technique (HARRIS et al., I98 I) to measure the emission rate of CO, in the eruptive plume. Each of these plume measurements requires several hours to complete and can be regarded as an average observed over a 3-4 hour time span; we shah assume that these averages are typical of the entire eruption. On February 13 we measured the emission rates of CO2 and SQ (both by airborne spectrometer) from the summit fumaroies of Rilauea, 19 km distant from the east-rift enrption site. Since April 1982, we have been making routine measurements of the total THE

* Michigan Tech University, Houghton, MI 49931.

SO, ffux from the summit fumaroles by a groundbased correlation spectrometer technique (CA&AD EVALL et a/., 1981). Results Of the airborne phtme measumments are @en in Table 1, and the groundbased measurements are shown in Fig I. In this paper, these data am combined with analyses of the eruptive gases (GREENLAND, 1984) and geoiogic data on magma supply rates and erupted lava volumes to provide estimates of (I) the atm~he~c loading of volcanic emissions from Rilauean eruptions, (2) the gas content of the eruptive magma, (3) the fraction of magma volatilized during storage in the summit reservoir, and (4) the initial content of gases in magma arriving from the mantle. These estimates have large associated errors (generally &30%) and the necessary assumptions are inherently imprecise: within these limi~tio~, the estimates delimit the extent and sequence of voiatilization from Kilauean magma. Throu~out the text, con~ntmtions are given in wt.% and “ton” refers to a metric ton consisting of 106 grams. ERU~IVE

GAS EMISSION RATES

The composition of the eruptive gases dire&y sampled at the vents has remained essentiahy constant over the 14month eruptive period of January 1983 to February 1984 (unpub~is~~ data), and the composition of the January 1983 samples (GREENLAND, 1984) is typical. The eruptive emission rates of all major gases have heen ealctdated in Ta~2~rn~~~rn~~n~~~~~ during the 15th and 16th eruptive episodes (Table I). The simlar emission rates shown in TabIe 2 from these two episodes probably results from a simiiarity in magma production rates rather than implying any constant temporal gas emission rate. Magma bin from these two eruptive episodes is given in Table 2, and the gas emission rats are combined with these magma v&ma to yield estimates of the fraction of magma lost by eruptive degassing (Table 2). The estimates from the two emptive episodes are not si~i~~n~y different considering the errors involved, and their average will be used henceforth. The agreement of these two estimates sUggesu both that our phnc m~lriucb ments are indeed typical of the entin etuptive span and

125

126

L. P. Greenland. W. I. Rose and J. B. Stokes Table 1. Alrborne plune measurmaentsfran Kflauea volcano,Hawaii

Emtssfons Date

(IWtviC

CO2

Dtstanceof lllo1e measurement

tOnS/ddy)

fra

vent

(km1

so2

Plwne chardcteristics av. d:;th ufdth windspeed (km/hrl (km) (km)

Eruptfve emissions,east rift: Feb.15, 1984 Mar.4, 1984

4700 3200

>lDOOO

(0.7

4-6

1.8

7.3

!5

~?OOOO

SO.5

S-7

2.3

7.4

10

(1

0.26

1.8

Non-eruptiveemfssfons,smit: Feb.13, 1984

1600

220

(1) CO2 medsured by an

11

infrared spectronrter technique (Harris etg.,

8.3

1981).

(2) SO2 measured by an ultravioletcorrelatfonspectraaetertechnique fCasadeval1etG., 1981).

that the content of volatile species in the magma is the same for different eruptive episodes. Corners the gas emission rates with ihe. magma volumes of Table 2 requires the assumptions thatonly the magma represented by lava has contributed to the gas emissions, and that there was no pie-eruptive degassing of this magma. Neither of these assumptions is strictly true. Visual o,bservation shows a continuous small plume issuing from the main Pu’u 0 vent during in&h&s between eruptive phases; this reflects pre-eruptive degassing of the magma, resulting in measured Java vommes over-estimating the volume of magma degas& during eruption. On the other hand, at the end of an eruption, degassed magma in the pond and conduit drain down the conduit, resulting in measured lava volumes under-estimating the volume of magma degassed during eruption. These are compensating errors. and. furthermore, they are small. Assuming a uniform conduit area

of 300 m* (the area observed at the surface) extending vertically to the riR transport zone, and assuming that ah magma above 4 km depth is degassed to I atm ~uiiib~um yields I X lo6 m3 magma omitted from our calculation: this amounts to only IO-20% of our Table 2 magma volumes. Degassing 0.11% SO, (Table 2) corresponds to 0.06% S. MOORE and FABBI (197 1) found 0.08% S in Hawaiian submarine (high confining pressure) basalts and 0.02% S in subaerial (I atm.) bar&s and suggested that the diffirence, 0.06% S, was lost from subaeriat basalts by degassing during eruption. FORNARIef al. (1979) reported averages of 0.11% and 0.014% S in Hawaiian submarine and subaerial basalts respectively, implying 0.09% S was lost by eruptive degas&g HARRISand ANDERSON (1983) compared the volatile components of melt inclusions in olivine with that of posteruption matrix glass in two samples of Kilauoan basalt to determine the composition of gases lost by eruption and

FIG. I. Ground-based measurements of total SO2 Bux from summit fumaroles of Kilauea volcano. Horizontal lines show change in average SO? tlux after the start of the east-rift eruption in January 1983.

Magmatic gas at Kilauea Table 2.

Data for material

balance

calculations

Data date

Phase 15 e. 984

eruption time span lava production magma production

gas eatssionrate (metric tons/day) $6

i nr3 8 ','lo" 6.4 x 106 m3

HC ? WF volatillzatlon co "2 8 sn UC ? HF

from magna

127

(wt.%)

Phase 16 3 1964

R

;;':l;$

&3

9.6 I 106 3

27000 58000 4700

1eooo 40000 3200

330 200

220 140

Notes

I:;

(2) (3)

(4) 0.022 n.27 0.12 0.0015 0.00092

0.016 0.20 0.092 n.0011 0.0006l3

11)

(Yolfe etfi.,

1984 A, 8)

(2)

Magma volumes calculated by assunlng 20% veslcularlty for the lava; any vale in the 0-5DZ range may be used without affecting the conclusions.

(3)

Calculated from maasored erupttve gas composition

(4)

Calculated from magma production rjte and gas nlsslon asswd magma density of 2.8 gmlcm

CO2 flux (Table 1) and the 19R3 east-rift given by Greenland 11984).

reportedlossesof

0.05%and 0.12% S. The correspondence of these three independent estimates of sulfur degassing suggests that the pre-eruptive sulfur content of Hawaiian magmas is constant and thus that the eruptive emission rate of S, 3 kg !80$m3 magma (from data of Table 2), also remains constant. If this So, emission rate is constant, then the atmospheric loading of pollutants from eruptions of Kilauea volcano can be estimated from the much more easily measured volume of extruded lava. The Jan. ‘83-Feb. ‘84 span of this eruption produced (WOLFEel al., 1984a) I65 X 106m3 of new lava. With a 20%vesicmaritycorrection, this corresponds to I30 X lo6 m3 magma. A. OKAMURA [pets. commun., I9841 made an independent preliminary magma volume estimate for this time period by summing the deflationary episodes of the summit magma chamber. He arrived at the same number, 130 X lo6 m3 magma, lost from the summit. Jkgassing of I30 X lo6 m3 magma at the rate of 3 kg SOx/m3 yields an estimated atmospheric loading of 4 X I@ tons SO2 due to this eruption. For comparison, emission of SOr from the summit area of Kilauea, not associated with eruptive aetivity, averaged 250 tons 802/&y over this period (Fig. I), amounting to I X Iti tons SO,. SUMMIT MAGMA CHAMBER DEGASSING HARRIS and ANDERSON (1983) found eruptive water losses of 0.17% and 0.29% in their two samples, in good agreement with the 0.24% loss given in Table 2 for this eruption. In contrast with the agreement of loss-estimates for Hz0 and S, their estimated loss of CO* (0.07%, 0.3%) is much greater than our Table 2 value of 0.02% COs. This discrepancy is consistent with the suggestion (GREENLAND et al., 1984) that the current east-rift eruption represents magma which has been largely degassed of CO2 in the shallow summit reservoir of Kilauea before transport to the east-rift eruption site. GERLACH ( 1980, 1982) has previously emphasized the importance of COxdegassing from Kilauean magmas. This degassing of CO*, amounting to 90% of the magmatic CO2 content (see below), implies that CO, became supersaturated in the magma at pressures much greater than that of the reservoir and thus that Cot had largely exsolved

.

rates with

from the magma to form a separate fluid phase before the magma arrived in the summit reservoir. If this is so, then agreement of our loss estimates for Hz0 and SO2 with previous estimates implies that summit degassing is largely limited to COr, and thus that magma arriving in the summit reservoir contains a very CO&h fluid phase. The extent of summit degassing can be estimated from measured emission rates and magma supply rates. GREENLAND et al.(1984) measured emission rates of 3600 and 300 t/d of CO, and SO,, respectively, on December 9, 1983, in Kilauea’s summit plume using the same techniques by which we measured our Table 1 data. The SO* emission rates from these two studies are well within the range of routine ground-based measurements (Fig. 1) and we shall use the average of the two estimates, 2600 and 260 t/d, for subsequent calculations. WOLFE (pers. commun.) estimated a magma supply rate to Kilauea’s summit reservoir of 12 X 10’ m3/yr for the eruptive year 1983. while the average rate for 1976-1980 has been estimated at 6 X IO'm3/yr (DZURISIN and KOYANAGI, 1981). Combining summit

the 1983 magma supply rate to the

reservoir with the summit fumarole emission rates of CO2 and SOI result in estimates of 0.29 wt.% CO1 and 0.029 wt.% SO2 (0.014% S) lost by degassing of the magma during storage in the summit reservoir. A similar estimate of Hz0 degassing cannot be made because of the obscuring effects of meteoric water contributions to the plume and condensation of water at low-temperature (5 boiling point) summit fumaroles. As noted above, agreement of our eruptive degassing estimate (Table 2) with the independent estimate of HARRIS and ANDERSON (1983) suggests that Hz0 loss during summit storage is small. If, as seems reasonable, less than 0.1% Hz0 is lost by summit degassing, then an arbitrary assumption of 0.03% Hz0 loss will be correct within a factor of

t.. P. Greenland.

128

W. 1. Rose and .I R Stokes

three, which is adequate for our purposes. The high solubility and low abundance of halogens in magma suggests that little volatilization of these from the summit reservoir should be expected. Condensate collections from summit fumaroles always have S/halogen ratios greater than 1000 (unpublished data) consistent with this expectation. We shall assume no halogens are lost by summit degassing. This estimate of S loss from the reservoir depends on the assumption that no S is lost by reaction with wailrock in its passage from the magma to the surface; this is not unreasonable in that the passageways can be expected to become quickly armored with reaction products and thus relatively inert. A more serious loss of S between magma and plume results from reaction with meteoric water and by condensation with water at the low-temperature fumaroles. The magnitude of this effect cannot be estimated. As shown below, our assumption that it is small yields results consistent with other work. SOz emission rates from the summit averaged 130 t/d in 1982 and increased in Jan. ‘83, after the eruption started, to average 250 t/d through 1983 (see Fig. I). The ratio of the rates of sulfur loss from the summit reservoir for the two time periods (I .9) is very close to the ratio (2.0) of the magma supply rate estimated for the two time periods (see above). This suggests that magma in the shallow reservoir is degassed of excess volatile components through the summit caidera fumaroles almost as fast as it is supplied to the reservoir. implying that magma subsequently erupted from the reservoir will be equilibrated with volatile species at the temperature and pressure of the reservoir. The correspondence of emission rate to supply rate suggests the possibility that at least some of the day-to-day variation in SO2 emission reflects short term variations m magma supply to the summit. Furthermore. this correspondence provides a basis for estimating the long term contribution of Kilauea volcano to atmospheric S pollution by using magma supply rate estimates (see summary of DZURISIN er al., 1984). Although there are only two summit plume measurements. the similarity of C02/S02 mole ratios ( 1I vs. 17) is consistent with the hypothesis that the SOz measurements reflect rapid degas&g of magma as it enters the summit storage reservoir. MATERIAL

BALANCE

by MOORE and FABBI ( f97 1), SWANSONand ti.4~~) (1973) and FORNARI et al. (1979) for Hawaiian subaerial basalts, and assuming 0.01% Cl and 0.04% F in accord with data from SWANSON and FABBI (1973).WRtGHT(l971), and WRIGHT and QKAMUKA (1977). it is possible to calculate a HzO-C&S-Cl-F budget and to estimate the initial content of volatile components in Kilauean magmas. Summing the gas fractions retained by the lava, lost by eruption. and lost by summit degassing (Table 3) yields 0.32% H20, 0.32% C02, and 0.18% SOz (0.09% S) for the gas content of the magma arriving in the summit reservoir. Our estimate of initial Hz0 content is within the range (0.45 I 0.15% HzO) found by MWRE ( 1965) for deepsea Hawaiian basalts, and our esttmate of SO* (0.09% Sj is in excellent agreement with the 0.08% S found by MOORE and FABBI (197 t ) and 0.11% S found by FORNARIer al. ( 1979) for deep-sea Hawaiian basal& Our estimate is compared further in terms of H:C:S ratios in Table 4 with ratios obtained from a pre-eruptive melt inclusion in trhvine and from the 1918-19 summit eruption gas (when the summit reservoir effectively rose to the surface, preventing me-eruptive degassing of the magmaj. The analysis of the summit eruption gases used in Table 4 is the “restored” composition of GERLACH (1980), who inferred a contamination of these samples by meteoric water; the original analyses, with an atomic H/S ratio of 15. result in even better agreement. The close agreement of these independent estimates makes a case for the constancy of volatile content of Kilauean magmas.

OF GASES

WOLE (pers. commun.) found typical values of 0.05% Hz0 and
CONCLUSIONS The arguments given above result in a very simple model for the fate of volatiles in the 1983-84 eruptive magma, Magma arrived in the shallow summit reservoir from the mantle with a CO*-rich fluid phase. Degassing from the reservoir was rapid and thorough. leaving the magma equilibrated with volatiles at the temperature and pressure of the shallow reservoir. Magma was transported through the east-rift conduit system at no shallower depth than the summit reservoir, and hecame degassed to 1 atm equilibrium on eruption. Because the magma is rapidly degassed to an equilibrium saturation with volatile species in the summit reservoir, and because the east-rift conduit is no shallower than the summit reservoir (KOYANAGI er al., 19~1). eruptive gas com~sitjons can differ from the Renoir-~tumtion values only if (11 reservoir degassing is coincident with eruptive degassing (e.g., the 1918-19 Halemaumau eruption). or (2)

Distributionof Kilauea gases wt. % of magma

sunmit degassing Eruotlve degassing lava retentfon Total

F

0."i? 3

0.24 0.0 co 0.0229 so Cl

0.24

0.019 0.11 0.0013 0.0008

-0.05 (0.01 ---- 0.04

0.01

0.04

0.32

0.011

0.041

0.32 0.18

Percent of Inltlal content cl F

0.0

0

0

75

6

69

12

2

16

c3

12

RR 94 -

129

Magmatic gas at Ki)auea Table 4.

This work H C s

13 2.6 1.0

H:C:S atomic ratios in un-degassed Kllauean magma Melt inclusion*

1918-19 eruption+

16 2.1 1.0

1.2 2.2 1.0

Acknowledgments-We are greatly indebtedto N. Banks,J. Moore, C. Neal, and E. Wolfe for their careful, helpful reviews of this work. We are grateful to E. Wolfe for permission to cite his unpublished data and to A. Okamura who estimated the summit magma loss at our request. Editorial handling: S. R. Taylor. RJZERENCES

*Melt inclusion

in olivine fram a subinartne east rlft eruptlon (Harris and Anderson, 1983). +Average of 7 analyses “restored” by Gerlach (1980).

there is sufficient variation in the original formation of magmas that some magmas arrive in the summit reservoir undersaturated in one or more of the volatile

species. Although this second possibility cannot be agreement of the three independent estimates of the composition of the undegassed magma in Table 4 suggests that such compositional variation is rare. The approximately constant volatile composition of the initial magma and the rapid degassing from the summit reservoir suggested by this model imply that only two volcanic gas compositions should be observed at Kilauea: an original “primary” gas (such as the 19 18- 19 eruption, Table 4) and a CO+pleted gas (such as the current east-rift eruptive gas), depending on whether the magma had been stored in the shallow summit reservoir. This is consistent with the observation that there has been virtually no change in gas composition over the ICmonth course of this eruption. In practice, the vagaries introduced by meteoric water addition, sampling of residual gases (e.g., cooling post-eruptive vents and lava lakes), degassing the magma in a highly dynamic, nonequilibrium, eruptive situation, and of reaction of the gases with wallrock and air in the vents can be expected to produce considerable variability among analyses; but these sources of variation are essentially sampling errors obscuring the actual gas composition of the eruptive magma. An important consequence of this model is that the atmospheric loading of volcanic pollutants (from Kilauea, at least) may be estimated more easily than heretofore realized. The direct relationship observed between SO2 emission from the summit and the magma supply rate suggests that noneruptive SO* emissions can be inferred from estimates of magma supply which extend back to 1952 (DZURISIN et al., 1984). Eruptive emissions of SOI can be estimated from measured lava volumes, and other pollutants can be estimated from the supposed constancy of the gas composition over time. Although such estimates are necessarily only a fnst order approximation, they do provide a basis for estimates where direct measurements are unavailable. A similar procedure would be applicable at any volcano where a basis exists for assuming that the magma gas content is limited to the saturation value at some particular depth. excluded,

CASADEVAU

T. J., JOHNSTOND. A., HARRISD. M., ROSE W. I., JR., WILLIAMSS. M., WOODRUFFL. and THOMPSON J. M. (1981) SO2 emission rates at Mount St. Helens from March 29 through December, 1980. U.S. Geol. Surv. ProjY Pap. 1250. 193-200. DZURISIND. and KOYANAGIR. Y. (1981) Changed magma budget since 1975 at Kilauea Volcano, Hawaii (al%.). EOS 62, 1071. DZURISIND., KOYANAGIR. Y. and ENGLISHT. T. (1984) Magma supply and storageat Kilauea volcano, Hawaii, 1956- 1983. J. Volcano/. Geotherm. Rex (in press). EATON J. P. (1962) Crustal structure and volcanism in Hawaii. in: The crust of the Pacific basin. Amer. Geoohvs. Union Mon. 6, 13-29. FORNARID. J., MOORE J. G. and CALK L. (1979) A large submarine sand-rubble flow on Kilauea Volcano, Hawaii. J. Voicanol. Geotherm. Res. 5, 239-256. GERLACHT. M. (1980) Evaluation of volcanic gas analyses from Kilauea volcano. J. Volcano/. Geotherm. Res. 7, 295-317. GERLACHT. M. (1982) Interpretation of volcanic gas data from tholeiitic and alkaline maiic lavas. Bull. Volcano/. 45-3, 235-244. GREENLANDL. P. (1984) Gas composition of the January 1983 eruption of Kilauea Volcano, Hawaii. Geochim. Cosmochim. Acta 48, 193-195. GREENLANDL. P., CASADEVALLT. C. and STOKESJ. B. (1984) Emission rate of CO2 and SO* from Kilauea Volcano, Hawaii. Nature (in press). HARRISD. M. and ANDERSONA. T., JR. (1983) Concentrations, sources, and losses of H20, C02, and S in Kilauean hasalt. Geochim. Cosmochim. Acta 47, 1 I39- I 150. HARRIS D. M., SATO M., CASADEVALLT. J., ROSE W. I. and BORNHORSTT. J. (198 I ) Emission rates of CO2 from plume measurements. US. Geol. Sure. Pro/: Pap. 1250, 201-207. KOYANAGIR. Y., UNGERJ. D., ENDOE. T. and OKAMURA A. T. (1974) Shallow eatthauakes associated with inflation episodes at the summit of Kilauea Volcano, Hawaii. Proc. S,vmp. ‘Andean and Antarctic Volcanology Problems, ” 621-631. KOYANAGIR. Y., NAKATA J. S. and TANIGAWAW. R. (198 1) Seismicity of the lower east rift zone of Kilauea Volcano. Hawaii, 1960 to 1980. U.S. Geol. Sum Open File Report 81-984, 15 p. MOORE J. G. (1965) Petrology of deep-sea hasah near Hawaii. Amer. J. Sri. 263, 40-52. MOORE J. G. and FABBI B. P. (1971) An estimate of the juvenile sulfur content of hasalt. Contrib. Mineraf. Petrol. 33, 118-127. SWANSOND. A. and FABBI B. P. (1973) Loss of volatiles during fountaining and flowagc of basaltic lava at Kilauea Volcano, Hawaii. J. Res.. U.S. Geol. Surv. 1, 649-658. WOLFEE., OKAMURAA. and KOYANAGIR. (1984a) Kiiauea. SEAN Bull. 9, 7-10. WOLF E., OKAMURAA. and KOYANAGIR. (1984b) Kilauea SEAN Bull. (in press). WRIGHT T. L. (1971) Chemistry of Kilauea and Mauna Loa lava in space and time. U.S. Geol. Surv. Proj: Pap. 735. 40 p. WRIGHT T. L. and OKAMURAR. T. (1977) Cooling and crystallization of tholeiitic hasah, 1965 Makopuhi lava lake, Hawaii. US. Geol. Surv. pro/: Pap. 1004, 78 p. I