The October 1974 basaltic tephra from Fuego volcano: Description and history of the magma body

The October 1974 basaltic tephra from Fuego volcano: Description and history of the magma body

Journal of Volcanology and Geothermal Research, 4 (1978) 3--53 © Elsevier Scientific Publishing Company, Amsterdam - - Printed in The Netherlands 3 ...

4MB Sizes 13 Downloads 26 Views

Journal of Volcanology and Geothermal Research, 4 (1978) 3--53 © Elsevier Scientific Publishing Company, Amsterdam - - Printed in The Netherlands

3

THE OCTOBER 1974 BASALTIC TEPHRA FROM FUEGO VOLCANO: DESCRIPTION AND HISTORY OF THE MAGMA BODY

WILLIAM I. ROSE, Jr.~, *, A L F R E D T. ANDERSON, Jr. 2, L A U R E L G. WOODRUFF l ,** and SAMUEL B. BONIS 3

~Michigan Technological University, Houghton, Mich. 49931 (U.S.A.) 2 University of Chicago, Chicago, Ill. 60637 (U.S.A.) 3Instituto Geogrdfico Nacional, Guatemala City (Guatemala) (Revised and accepted February 23, 1978)

ABSTRACT Rose, W.I., Jr., Anderson, A.T., Jr., Woodruff, L.G. and Bonis, S.B., 1978. The October 1974 basaltic tephra from Fuego volcano: description and history of the magma body. J. Volcanol. Geotherm. Res., 4: 3--53. The largest of a series of 20 vulcanian eruptions of Fuego volcano between 1944 and 1976 occurred in October 1974 in four distinct 4--17-hour pulses over a 10-day period. The eruption produced more than 0.2 km 3 of pyroclastic high-A1203 basalt (equivalent to >0.1 km 3 of void-free magma), quenched at a temperature of about 1050°C. Early erupted magma was rich in plagioclase and poor in mafic phenocrysts, magmas erupted next in sequence were poor in all phenocryst phases, while the final magmas were rich in all phenocryst phases. These changes, and variations in 8 major and 21 minor elements can be explained by crystal/liquid fractionation of high-A1203 basalt. Crystals in Fuego's magma contain inclusions of glass with an average H20 content of about 3%, an amount that suggests that crystallization began at a depth of ~ 5 kin. Petrographic data, observed extrusion rates, glass inclusion analyses, regional geological features, and various geophysical considerations collectively suggest that crystallization and differentiation occurred within a vertical, dike-like conduit during a period of a few days to several months. The Fuego parental basalt contained 5 % MgO, 1 0 % FeO* and about 0.8% K 2 0 , it began crystallizationat 1030 -+ 50°C with plagioclase (Angs.89),olivine (Fo76.66)and magnetite (4.3--15% TiO:) the principal phases. Plagioclase was apparently rafted upward by bubbles within the m a g m a body. Tidal triggeringof crystal growth, m a g m a ascent and eruption are suggested.

* Present address: National Center for Atmospheric Research, Boulder, Colo. 80307, U.S.A. The National Center for Atmospheric Research is sponsored by the National Science Foundation. **Present address: University of Chicago, Chicago, Ill. 60637, U.S.A.

INTRODUCTION

Regional setting of Fuego volcano Fuego volcano (14°29'N, 90°53'W) is situated in the WNW-trending Quaternary volcanic chain. The Middle America trench parallelsthe volcanic chain about 100 km to the south. A belt of shallow and intermediate-depth earthquake foci occurs between the trench and the volcanic chain. The tectonic setting has been interpreted as a converging plate boundary (Molnar and Sykes, 1969) at which the Cocos plate is underthrusting the Caribbean plate beneath the volcanic chain. The volcanic chain and subduction zone have been divided into segments based on volcanic lineaments and seismicity (Stoiber and Cart, 1973). Fuego lies within the central Guatemalan segment of the arc, which extends from Santa Maria volcano to Pacaya volcano (Fig. 1). Fuego is the southernmost of four vents, which form a 5 km long north-southtrending line. Each of the vents forms a summit along the north-south ridge; from north to south the summits are: Yepocapa (3880 m), Acatenango (3976 m), Meseta ( ~ 3 6 0 0 m) and Fuego (3763 m). Material erupted from the four vents forms a single coalesced stratovolcano complex elongated north-south. Nearly all historic activity of this volcano group has occurred at Fuego (Mooser et al., 1958).

Historical pattern of activity at Fuego The historic activity at Fuego includes more than 60 eruptions since 1524, these have been short-lived (duration of a few hours to several days), violent QUATERNARY COMPOSITE CONES ~° TACANA 1 4 0 9 ~

6~'c;

OF

UEX/CO

I

a

N

a

A R E A OF INSET

' ~ ~

"~

I ~ rsc

t

~''~

PACAyA LZ552m)

Fig. 1. Location maps of Quaternary volcanic chain of Guatemala Highlands (after Rose et al., 1977).

(kinetic energy/thermal energy = 0.01--0.05), vulcanian eruptions, commonly with pyroclastic flows. Four 20--50-year clusters of eruptions account for more than 75% of the activity (Fig. 2). Twenty eruptions make up the recent cluster, which began in 1944 and continues (Table 1). Following the method of Wickman (1966), Fuego's probability of eruption in any month (~b in Fig. 3) is 0.02 in each of the first 36 months of repose and decreases noticeably thereafter (Fig. 3). 2000 M

M 1900M

5]800

>

M

j

M 1700

1600-

----

;4

1500 Reported E r u p t i o n s Major E r u p t i o n s ~ M

0

~ Running Average No Eruptions / 40 yrs

I0 Eruption Clusters

Fig. 2. Historic activity of Fuego volcano. Data from Mooser et al. (1958), and Table 2.

Events and aftermath of the Fuego eruption, 1974 The 1974 eruption was Fuego's most voluminous since 1932 (Deger, 1932) and began at 4:00 a.m. (local time) on October 10 (or 1000 GMT), with small ash eruptions from the summit crater (Bonis, 1974). Four distinct episodes of basaltic airfaU and ashflow activity followed (Table 2). After October 23 the eruption diminished greatly in intensity and by December 4 the plume from the central crater was less than 1.5 km in height and contained almost no ash (Crafford, 1977). A photographic record of the dramatic pyroclastic activity was pub~shed by Buell and Stoiber (1976). The 1974 eruption produced a substantial increase in stratospheric particulate matter in the northern hemisphere (Meinel and Meinel, 1975; Volz, 1975) and probably in the southern hemisphere as well (Hoffman and Rosen, 1976). The commencement of 1974 activity at Fuego coincided with minima in the luni-solar tidal acceleration (i.e., at low tides, Fig. 4). The first major

TABLE 1 F u e g o volcanic activity 1 9 5 7 - - 1 9 7 7 (November) Date

Character

Rating*

Source(s)**

19 Feb. 1957

Ash, ash flow

2--3

4 - - 2 0 Aug. 1962 Oct. 1962

Ash, ash flows a n d lahar 2--3 Central crater explosive 4 activity and lahar Strong explosive activity, 2 central crater. Ash cloud e s t i m a t e d to 12,000 m Explosive activity 2--3 central crater, m u d flows, ash flow Explosive activity 4 central crater Ash eruption, ash flows 2 Ash, cloud e s t i m a t e d to 3 12,000 m Ash eruption 3

Pough a n d Mulford {1957); BVE no. 9-1 BVE no. 3; BVE no. 9-1 BVE no. 3

9 Nov. 1962

29--30 Sep. 1963

7 Feb. 1966 20 Apr.--1 May 1966 12--13 Aug. 1966 22--24 Apr. 1967 14--15 Sep. 1971

Ash e r u p t i o n , ash flows volume ~ 6 × 107 m 3

2

22 Feb.--3 Mar. 1973

Small ash e r u p t i o n , ash flows

3

22--23 Mar. 1973

Very weak ash e r u p t i o n s

4

10 Oct.--4 Dec. 1974

Largest (?) e r u p t i o n since 1932; v o l u m e > 0.2 k m 3

1

28 May 1975

Small ash eruptions, possible ash flow Sporadic very small ash eruptions, possible ash flows 29 July Increasing vapor emission followed by small airfall ash e r u p t i o n s Larger airfall e r u p t i o n s ash falling at Antigua

5

23 Jul.--4 Aug. 1975

18 Sep. 1975 and continuing 11--21 Oct. 1975

3 Mar. 1977 19 Apr. 1977 11 Sep. 1977 a n d continuing

Small ash e r u p t i o n Small ash e r u p t i o n Weak airfall e r u p t i o n s

4

BVE no. 3 ; BVE no. 9-1

BVE no. 4; BVE no. 9-1

BVE no. 7 D. Eberl, R. Dorion

Rose (1977) New York Times, April 24, 1967 CSLP event 85-71, cards 1290, 1295, 1296; Rose et al. (1973) CSLP event 24-73, cards 1573, 1583; Bonis and Salazar (1974); Rose (unpublished) Bonis and Salazar (1974); Rose ( u n p u b l i s h e d ) CSLP event 134-74, cards 1955, 1967, 1971; m u c h u n p u b l i s h e d material from m a n y sources CSLP event 134-74, card 2263; Rose (unpublished) CSLP event 134-74, card 2263

5

CSLP event 134-74, card 2307

4

CSLP event 134-74, card 2307; SEAN, v. 1, no. 1, p. 2. SEAN, v. 2, no. 3, p. 2. SEAN, v. 2, no. 4, p. 2. SEAN, v. 2, nos. 9, 10, 11.

5 5 4

For pre-1957 activity see Mooser et al. (1958, pp. 58--59). * R a t i n g s 1 = largest event in t h e series; 2 = " a m o n g t h e l a r g e s t " ; 3 = " i m p o r t a n t " ; 4 = "weak"; 5 = "very weak".

** BVE = Bulletin of Volcanic Eruptions, Volcanological Society of Japan, International Association o f Volcanology, I U G G ; CSLP ffi S m i t h s o n i a n Center Short-Lived P h e n o m e n a ; SEAN = Scientific Event Alert Network ( S m i t h s o n i a n I n s t i t u t i o n ) Natural Science Event Bulletin.

5

4

5

6

7

i

i

I

8

I

"

12

OCTOBER

1 15 I t4 1 15 i 16 1 Jr

i;!'ii~!ll~Nl'Lllil' 18

~

~ F9 [20

' [ 21

'~ ' 22

P'

fig. 4. Calculated vertical component of tidal acceleration in microgals, for the latitude of .~uego during the period of October 2--November 4, 1974 (all time is GMT). The shaded ;imes are periods of significant activity (from Table 1 ). Tidal data obtained through a :omputer program written by H. Pollack, based on formulae described in Pollack (1973) tnd Longman (1959).

Gols

-IO0

I i

25 I 24

° 25

, 26

~ °~ 27

~ 28

' s [ 291

I

v

I

50

i I",/~', 5!

~,

'i i

!

, $

i= 4

NOVEMBER

] 2

i

I

ID

TABLE 2 Chronology of the 1974 Eruption of Fuego (Bonis, 1974; R. Dawson, personal communication; R.E. Stoiber, personal communication; T. Crafford, personal communication) Events* 10 October

4 a.m. eruption begins. Small ash eruptions from summit (< 1000 m), very small ash flows (< 3 kin)

14 October

2 a.m.--7 p.m. intense activity, vulcanian explosions, ash flows. Stratosphere venting observed

16 October

4--5 p.m. intensity of activity increasing after relatively quiet period. Ash and mud fall in rain at Escuintla

17--18 October

9: 45 p.m.--12: 30 a.m. Very large vulcanian eruption. Cloud height to stratosphere (>7 km above cone). Ash flows and extensive airfall. Largest of four main phases of eruption. Pulsation of intense explosions with 1-minute periodicity

19--20 October

8 p.m.--5:30 a.m. Smallest of four intense vulcanian activity periods. 3 km high expanding ash column. Activity consisted of milder airfall ash emission with a column by morning of 20 October

23 October

1 p.m.--7 p.m. Last intense activity of 1974 eruption. Vulcanian column to stratosphere level. Ash flows and extensive airfall. Lithic fragment and bomb shower to southwest and west of the summit. Following this date, activity was very mild

23 October--4 December

Waning airfall ash activity, column 1.5 km high, generally decreasing in intensity. Little, if any, ash in plume after 4 December

* All times local time = G M T - 6 hours.

straightforward as it is in many examples (Rose et al., 1973), and a four-step integration is needed (Fig. 6). Uncertainty in the volume of fine ash scattered to great distances inhibits a maximum volume estimate, b u t at least one line of evidence suggests that the maximum bulk volume is unlikely to be more than 0.4 km 3. For the total volume of ash to exceed 0.4 km 3, more than 50% by weight of the ash would have to have fallen at distances exceeding 100 km and this ash would be very fine-grained with a mean grain size of less than 20 ~m (Davies et al., 1978). Since the ash is thought to originate primarily by vesiculation, the size of particles produced is influenced strongly by the size of vesicles. As shown in Table 3, the vesicle sizes in Fuego ash average a b o u t 75 ~m, ranging from a minimum of a b o u t 1 ~m to a maximum of 250 pm. Ash samples taken within a few tens of kilometers of Fuego consist of a mix-

10

J

/

/

/ I Q~ezolten~n~o @ ~0 ' °

.

.

\ • " Solola ~ •

.



".3

_1--.

".

:__

.'

.....

%,

O%o~

......

•....... --

.

"

io'~"



11o-

_ ~_~-~

~--,.,, " f



CY.':b.W#/2

. . . . .

"'~"

/.

.--

~

Oc)ober !4

dL,~y~hj"~ Lake

Isopochs

in centimeters Sample

[?3

t

Ahtten I

1ocotions

0

0 L: :

Anolysed sGmples

20 :=Lz=l:

: :_~

kilometers

40 = :3

FUEG° t

.

October 14 ONLY

Fig. 5. Map of ash thicknesses measured in the field followed the 1974 Fuego eruption. Map below was measured on October 15 and 16 following the initial phase of the eruption, but prior to later activity. Map based on data collected by the Instituto Geogr~fico Nacional, Guatemala.

ture of multi-vesicled ash, crystal fragments and shards, whereas distal samples are greatly enriched in shard particles. Sparks and Wilson (1976) have pointed out that the minimum vesicle size is likely to be very similar to the minimum shard size in subaerial eruptions where vesiculation is the dominant fragmentation mechanism. Further, they show that submicron size ash is not likely to be a major portion of the total size range in a Plinian column, based on bubble growth dynamics. It is interesting that stratospheric observations on the particulate materials (much of it submicron size: Hoffman and Rosen, 1976) haveshown that the dominant particle types are volatile condensates, probably sulfuric acid and not silicate. Thus evidence of any substantial volume c o m p o n e n t of small grain size ash is absent, and the total volume is probably less than 0.3 km 3. The volume could be greater than 0.3 km 3 if (1) better data on the far-flung ash showed that larger amounts of fine material were erupted and (2) if vulcanian eruptions can be shown to be subject to fragmentation mechanisms different from Plinian eruptions (Self, 1977). As the ashes were collected in the field, several density measurements of freshly fallen ash were made. The average density of ash was about 1.1 g/cm 3. This is

11 I 0,000- .. uncertain

Volume of ash blanket

B\ extrapolation

V = volume A : intercept t = thickness C = slope

" %

cl'~:~ "~tl , ~ \ tl ~L

F v:j,(r, dt

\ C2~

rq rt~ V(tht"tz'):A° / to'at÷A° | tC2dt*

I"J~3

\\\\

I ~000"

Area,. krn 2

Oct. 14 . . .{only) ... ~"

"\,,~ . "-.,.

'?~_

= 0.21 xlogm z'

" C~

"x°

IOO-

Min. volume (Oct 14)

2dtl C.

""~Z,~totalash blanket

\ \ \ \ ejz ~" 3

V(t,,te) > [t2tcdt '~t I ) D3 x 109m3

c4E~

\

IO o.ooz

I

I

OOl

0.1

I

1.0

Thickness, m. Fig. 6. Integration of volume of Fuego ash blankets of 1974. See Rose et al. (1973) for further explanation of volume determination by this method.

a b o u t 40% of the density of dense basalt, thus the minimum volume of ash erupted must be reduced to 0.1 km 3 of dense rock. Davies et al. (1978} have estimated that ash flows associated with all four phases of the eruption produced between 0.005 km 3 and 0.01 km 3 of porous ash, or a b o u t 2--5% of the airfall volume. Data on the ash blanket following the October 14 period b u t preceding the October 17 activity (Figs. 5 and 6) establish that the first intense phase certainly produced more than 10% of the total volume of ash. Based on the combined observations of many people, we estimate that it represented a b o u t 30% of the total. Because later phases produced ash blankets which overlapped the October 14 blanket, it was more difficult to be sure a b o u t the relative sizes of later phases. Observers generally agree that the second phase of the eruption (October 17) was the most violent and produced the most ash of the four. We estimate that a b o u t 40% of the total ash produced was erupted on October 17. Later eruptions were smaller, the October 18--19 phase probably produced a b o u t 10% of the total volume, the October 23 phase a b o u t 20%.

12

TABLE3 Petrographic characteristics of bombs and lap(Ill which fell in each of the four phases of the 1974 Fuego eruption. Values given outside of parentheses are means, those within are ranges. Vesicle percentages are given as volume percentage of rock, other percentages as volume percent of solid material (vesicles excluded) October 14

October 17

October 19

October 23

Groundmass (vol.%)

62(40--67 )

68( 58--8] )

66(60--75 )

55(51--69)

Plagioclase (vol.%)

Size (mm) Olivine (vol.%) Size (ram) Pyroxene (vol.%) Size (ram) Opaques (vol.%) Size (mm) Vesicles {vol .%) Size (ram)

31{18--51) 0.39(0.14--0.81 ) 3.6 0.47{0.15 0.86) 0.8 0.35(0.14--0.76) 2.6 0.23(0.12--0.52) 38( 32--47 ) 0.33(0.05-1.3}

22(13--34} 0.36(0.12--0.73) 6.0 0.53(0.23--0.82) ] .6 0.47(0.31--0.59) 2.6 0.24(0.14--0.54) 40( 34- 45 ) 0.072(0.01--0.26)

21(14--25) 0.46(0.18--0.95 ) 9.4 0.54(0.21--0.7) 0.9 0.20(0.09--0.54) 2.6 0.26(0.07--0.49) 33{ 28--38 ) 0.065(0.001--0.18

27(15--32) 0.41(0.18--0.57 / 12.6 0.59(0.16--2.0) 1.5 0.53(0.13--1.82) 3.9 0.27(0.11--0.66} 26( 20--31 ) 0.049(0.001--0.18)

Number of

~

,

e;

determinations

Sampling Locations where ash was sampled are shown in Fig. 5. From approximately 350 samples of airfall material, about 80 were selected for detailed study. We chose samples representing each of the four pulses of the eruption. Samples which had not been affected by any precipitation were particularly selected, so that soluble chemicals from ash surfaces could be extracted. Most selected samples were ash which had lapilli sizes much larger than phenocryst sizes, this selection was made to avoid atmospheric fractionation of the ash. Evidence to show that little atmospheric fractionation did occur in samples we examined is given in Fig. 7. Besides the ashes, 25 bombs were studied petrographicaUy. Most of the samples studied in detail are from localities between 10 and 40 km of the summit crater.

Petrography of the ash The Fuego tephra of 1974 are, like other historic tephra, olivine-bearing high-alumina basalts, made up of phenocrysts (generally > 0.1 mm), microphenocrysts (10--60 Urn), glass (or devitrified glass) and round, subspherical.vesicles (1--400 ~m). Phenocrysts are dominantly plagioclase (An9s-s0) and olivine (Fo76-66). Magnetite (4.3--15.0% TiO2), augite (En4sFs13Wo42) and oxyhornblende are minor phenocrysts. In general the percentage of phenocrysts and their average size increases with time, while the percentage of vesicles and their size decreases (Table 3, Fig. 8). Clots of crystals (up to several millimeters in diameter) are more com-

13

2,

AI20 s

0

0 0

AI203 O

C

O

0

0

0 o

% 2or

O

Fl [3

E3

MgO

4[]

E3

c~[]:s

[]

c]

Fin

13

E]

3~

Weight % 2

®

e

0 0

K2O

~®e e e

e

I'0 i , , , 2O 30 40 5O Downwind Distance - Kilometers



@

6i0

Fig. 7. Variation of chemical composition of October 14 ash samples from Fuego as a function of distance downwind from the vent. Lack of systematic changes is cited as demonstrating that atmospheric crystal fractionation is not very important in explaining chemical differences in ash samples.

mon in tephra of the intermediate stages than at either the beginning or end. The clots contain euhedral to subhedral plagioclase, olivine, magnetite and c o m m o n l y ophitic augite. They contain less than 10% of intersertal glass and voids.

Plagioclase phenocrysts The plagioclase phenocrysts vary in size, shape, aggregation, zoning pattern, growth symmetry, composition, density and type and c o n t e n t of inclusions. These properties (except for density) are described in detail in another report (Anderson, in preparation) and are summarized below. Most of the plagioclase phenocrysts are platy, rhomb-shaped crystals, but subequant prisms are also abundant. A b o u t half of the crystals are euhedral, most of the remainder are subhedral. Most phenocryst grains consist of single (but twinned) crystals with only one major center of growth. A b o u t 20% of the phenocrysts have two or more centers of growth.

14

40 J x--

',,vesicles

5O

X

!

',o 1!o , 201 modal %

.

/ "~

/

i0

oltvLn%~ °"

• o-"

~i

magnetite ~-o

I

,7

j-o

.... o / "

,;

Y

2s

OCTOBER

Fig. 8. Graphical summary of modal changes in Fuego magmas during 1974 eruption (from Table 3).

The phenocrysts have three kinds of regions: (a) weakly normally zoned regions; (b) oscillatory-zoned regions, and (c) patchy-zoned regions rich in inclusions of gas and glass. Three arrangements of the above regions are common: (1) type 1 crystals (Fig. 9) have cores of (a) surrounded by rims of (b) in which may occur one or more regions of (c); (2) most crystals are of type 2 and have a core of (c) with rims of (b) in which may occur one or more regions of (c) (Fig. 10). (3) A subordinate group of crystals (type 3) is characterized by oscillatory zoning (b) from core to rim, interrupted by regions of (c) (Fig. 11). All plagioclase phenocrysts have a border zone up to 30 um thick which commonly contains inclusions of microphenocrysts (Fig. 12) and which is more sodic than the rest of the grain. The border zone is absent on anhedral (fractured) margins of the phenocrysts, although most of the fractured margins are coated with glass or devitrified matrix materials (Fig. 13). The several patterns of zoning described above identify three types of plagioclase phenocrysts. All three types are present in tephra from each major pulse of the 1974 eruption of Fuego volcano. Any two types occur together in a single lapillus.

15

Some of the t y p e 1 and 2 plagioclase phenocrysts have thicker oscillatory zoned regions on one crystal face than on the alternate face of the same crystal form (Fig. 14). Intervening patchy-zoned regions are reversely arranged, being thicker on the face which has the thinnest oscillatory zoned region.

Composition and density of plagioclase phenocrysts The normally-zoned cores of crystals have compositions between An9s and Ans9. The oscillatory-zoned portions of crystals range in composition from An9s to Ans3, except for the outer rim about 30 p m wide which is about AnTs. Some of the outer parts of the largest oscillatory-zoned crystals are about Ansg. The inclusion-rich patchy-zoned regions range in composition between about Anss and An83. The latter regions are c o m m o n l y but not invariably more sodic than adjacent oscillatory-zoned regions. The densities of euhedral plagioclase phenocrysts range from 2.66 to 2.76 g/cm 3 for type 1 crystals and from 2.61 to 2.75 g/cm 3 for types 2 and 3 crystals (Fig. 15). The range in densities observed reflects the presence on or within the crystals of groundmass, vapor, and/or olivine or magnetite, all of which are observed inclusions in sectioned crystals. Crystals denser than 2.74 g/cm 3 occur in both ashes 83 (October 14) and 178 (October 23). Most of the crystals have densities less than the ideal value of 2.73 g/cm 3 for their composition of Ans5 and this is because of glass and gas inclusions within the crystal. If the lower density is due entirely to inclusions of gas, 1--5 vol.% of gas is present in the crystals in such a way that vacuum impregnation cannot readily displace it. Allowing for thermal expansion, about 2.59 g/cm 3 is the lowest density which the plagioclase phenocrysts might attain in the body of magma beneath Fuego.

Inclusions in phenocrysts Phenocrysts contain inclusions of other minerals and of glass and gas. A crystal of pargasitic hornblende occurs in the weakly zoned calcic core of one type one plagioclase phenocryst (Fig. 16). Subhedral crystals of calcic plagioclase (An90-92) are c o m m o n in olivine phenocrysts. Subhedral magnetite crystals also occur in olivine phenocrysts but are less c o m m o n than plagloclase. Some plagioclase phenocrysts cohtain inclusions of olivine and rarely magnetite. Inclusions of devitrified glass are more c o m m o n than inclusions of glass in plagioclase phenocrysts. Small, isolated, spherical inclusions of glass are characteristic of the weakly zoned cores of type 1 plagioclase phenocrysts (Fig. 9). Large, irregular and interconnected inclusions of glass are characteristic of the patchy-zoned regions of the plagioclase phenocrysts (Fig. 10). Many inclusions, particularly the largest ones, are largely occupied by vapor, are connected to cracks and have bordering rims of sodic plagioclase {Fig. 17). Less c o m m o n l y inclusions of glass in plagioclase have smaller vapor bubbles, no sodic rim and no crack (Fig. 18).

U

17

Fig. 9. Type I plagioclase phenocryst, ash flow sample 83 (October 14), under crossed polarizers. Small divisions of the scale are 10 ~m apart. The calcic core (C) is light gray and contains a group of isolated inclusions of glass, the largest of which are now filled with gas and cut by fractures which terminate within the surrounding patchy-zoned region. The corners of the core are round. The rim consists of alternating groups of oscillatory zones and patchy-zoned regions crowded with inclusions of gas and glass. The outermost sodic rim is about 10 ~m wide and shows as a light-gray border adjacent to the dark adhering groundmass. The fractured portions of the crystal margin are natural. The grain is mounted in epoxy which shows strain birefringence. Fig. 10. Type 2 plagioclase phenocryst, ash fall sample 53 (October 14), under crossed polarizers. Small divisions of the scale are 10 ~m apart. The subhedral core region contains numerous, large, irregular inclusions of glass and gas. Although the core is patchily zoned, the patchy zoning is not evident in transmitted light. An oscillatory-zoned region surrounds the core and contains a thin patchy-zoned region near the margin of the grain. Fractures penetrate the entire grain. Fig. 11. Type 3 plagioclase phenocryst, ash fall sample 53 (October 14) in reflected light (Nomarski interference contrast). The smallest scale divisions are 10 ~m apart. The central core is oscillatory-zoned and lacks either an unzoned or patchy-zoned core. Possibly this and other type 3 grains lack such cores because of a sectioning artifact. The illustrated crystal has a thick patchy-zoned rim which typically terminates inward on a euhedral core. The outward termination of the patchy-zoned region is irregular and overlain by sinuous oscillatory zones which conform to the underlying irregularities. Large (black) inclusions of vapor (V) are typically concentrated near the inner boundary of the patchy-zoned region. The outermost sodic rim is present as a 10 ~m thick, inclusion-free border around the periphery of the crystal. Fig. 12. High-magnification view (reflected light, Nomarski interference contrast) of the outer sodic rim (R) on a plagioclase phenocryst from ash fall sample 53 (October 14) showing the typical partial inclusion of microphenocrysts of pyroxene (P) in the outer part of the outermost rim. The scale bars are 10 ~m apart. The thickness of the rim typically diminishes toward the apex of the reentrant near the top of the picture. Microlites are crowded together in the groundmass (GM).

Fig. 13. High-magnification view (reflected light, Nomarski interference contrast) of the margin of a plagioclase phenocryst from ash fall sample 53 (October 14). The scale bars are 10 ~m apart. The sodic rim (R) and oscillatory-zoned interior (I) are truncated by an irregular fracture (F). Although the groundmass (G) adheres to the fracture, no sodic rim is developed between the groundmass and the interior of the grain. Fig. 14. Asymmetrically zoned type 1 plagioclase phenocryst from ash flow sample 83 (October 14) in crossed polarized light. The crystal is about 2 m m across (the smallest scale bars are 10 ~m apart). On the right-hand side the outer patchy-zoned region is thinner than on the left-hand side, whereas the oscillatory-zoned region is thicker on the right than on the left. The depicted pattern of asymmetrical zoning occurs on a small proportion of the plagioclase phenocrysts; most of the plagioclases are symmetrically zoned. Within the core (C) there is one large (but thin) inclusion of glass (G) as well as a cloud of tiny inclusions of glass and a surface f'flled with tiny bubbles of vapor. The core is subhedral with some round corners designated (R).

18

J /f

14 CUM 12 NO. 10

6 4

2 2.6

/

/

/ 2.7 DENSITY gm./cm;

2.8

Fig. 15. Graph of the distribution of plagioc!ase phenocrysts according to density. The ordinate indicates the total number of crystals less dense than the particular density value shown on the abscissa. The densities of the plagioclase phenocrysts were determined in different ways for the separate samples. Sample 83 plagioclase were density titrated by adding measured volumes of a dense liquid to a less dense liquid. S o m e evaporation of the low-density diluent (dimethylformamide) occurred during titration. Consequently the densities of the sample 83 plagioclases m a y be too low. The sample 178 plagioclases were measured by i~sink-float rate technique described in the text. Both methods have an estimated accuracy of -+0.02 g/cm 3. The densities of the plagioclases in the two samples are the same within the limits of error. Most crystals are significantly less dense than the value for Anss (2.73).

Augite crystals commonly contain subhedral inclusions of plagioclase (Fig. 19), and such augites are generally anhedral to subhedral, and weakly oscillatory zoned. The included grains of plagioclase commonly lack 1 pm scale oscillatory zoning. Although some augite crystals are subhedral and contain abundant inclusions of glass, most of the augites have ophitic textures similar to the one depicted in Fig. 19. Inclusions of quenched melt in olivine are commonly glassier, larger and less abundant than similar inclusions of glass in plagioclase. Olivine crystals which occur in aggregates (Fig. 20) contain abundant inclusions of glass. Some inclusions of glass in olivine are transected by cracks and have microlites (daughter crystals) of plagioclase. Rarely blebs of sulfide are present together with glass and crystals of magnetite in olivine (Fig. 21), permitting an estimate of the oxygen fugacity of the intratelluric melt. Inclusions of glass are present in crystals of magnetite and augite too, and these inclusions are subspherical in shape and intermediate in size between those in plagioclase and those in olivine. The presence of inclusions of glass in all of the phenocrysts helps to establish the composition of parental liquid and to outline the crystallization sequence for the magma.

Size distribution o f olivine phenocrysts Maximum grain diameters of olivine are the same for the three early ashes and larger for the latest ash (Table 3).

19

CHEMICAL COMPOSITION OF THE ASH Table 4 shows that the chemical composition of the 1974 Fuego ash is that of high-alumina basalt. There are significant compositional variations in ashes produced in the four pulses of the eruption; some of these are shown graphically in Fig. 22. MgO, FeO*, MnO, Cr, Co, and Ni show consistent increases. K20, Rb, Ba and the rare earth elements (REE) show an initial increase and subsequent decrease. MgO shows the largest changes of the major element oxides, and though its average increases steadily in each pulse, its range within the latter pulses also increases markedly. We calculated a weighted average composition (WAC) of 1974 ash for Fuego (Table 4) based on the compositions and relative volumes of each of the major periods of activity. The WAC is within the range of compositions of the high-alumina basalt of Kuno (1960) but it has a higher FeO*/MgO ratio than most calc-alkalic basalts from the Cascades. CHEMICAL COMPOSITIONOF INCLUSIONSOF GLASSIN PHENOCRYSTS

Major element data Inclusions of glass and devitrified glass were analyzed in the phenocrysts of plagioclase, olivine and magnetite from several tephra samples. A broad range of glasses is present in each tephra sample. The glasses range in composition from high-alumina basalt with 0.5 wt.% K20 to basaltic andesite with 1.58 wt.% K20 (Table 5). Glasses with less than 0.8 wt.% K20 are mostly restricted to olivine. A selection of analyses of inclusions of glasses low in K20 is presented in Table 6, because of their bearing on the probable composition of the parental melt of the Fuego magma. The analyzed inclusions of glass in plagioclase are located in patchy-zoned regions. The spherical inclusions of glass which occur within the weakly zoned cores of type 1 plagioclase phenocrysts are mostly smaller than 15 pm in diameter, and no reliable analysis of them exists. Crystals of plagioclase included within olivine are mostly more calcic than An90 and similar to the cores of type 1 plagioclase, whereas the plagioclase containing the analyzed inclusions of glass is less calcic. It is likely that the liquid from which the cores of the type 1 plagioclase grew was more similar to the inclusions of K20-poor glass in olivine than to comparable inclusions in patchy zoned plagioclase. Residual glass in the lapilli (Table 7, column 4) has a composition close to the K20-rich, MgO, A1203-poor inclusions whereas the K20-poor, MgO, A1203rich inclusions are close to that of the weighted average composition of the eruption. Thus, the inclusions of glass in phenocrysts have compositions which range from that of the bulk magma to that of the residual glass.

Fig. 16. Crystal of pargasitic hornblende (Hb) included in the calcic core (C) of a type 1 plagioclase phenocryst from ash fallsample 178 (October 23) as seen in plane light on the universal stage. The scale bars are 10 # m apart. Small inclusions of glass are present in the core together with the hornblende. Microprobe analysis of the hornblende gave: 42.4% SiO2, 13.5 A120~, 10.4 "FeO", 15.1 MgO, 12.1 CaO, 2.3 Na:O, 0.4 K~O, 2.7 TiO 2 and 0.05 Cl.

Fig. 17. High-magnification view (reflected light, Nomarski interference contrast) of the border of sodic plagioclase (S) around the inside of an inclusion of glass (G) in a plagioclase phenocryst from ash fall sample 53 (October 14). The scale bars are 10 u m apart. The inclusion is intersected by a fracture (F) which connects to a nearby inclusion of vapor (V). The inner border of the sodic plagioclase against glass is highlighted with black ink. The border is about 5 um thick. The included glass contains 0.02 wt.% S. Fig. 18. High-magnification view (reflected light, Nomarski interference contrast) of an unbordered inclusion of glass (G) in the same plagioclase phenocryst as shown in Fig. 17. The inclusion of glass appears not to be intersected by a fracture and contains 0.12 wt.% S. Fig. 19. Augite crystal (A) which contains inclusions and attached grains of plagioclase (P) from ash fall sample 53 (October 14) seen in plane polarized light. The augite is anhedral and contains faint oscillatory zones (not visible in this view). A few inclusions of glass (G) and magnetite (M) are present.

21

Volatile element data

Chlorine shows fair coherence with K20 up to about 1 wt.% K20 (Fig. 23). The C1/K20 ratio ranges from 0.18 to 0.09 with most of the high values found in K:O-poor glasses in olivines and most of the low values found in K:O-rich glasses in plagioclases. The residual glasses outside the phenocrysts have the lowest C1/K20 ratios (0.07--0.08). Some inclusions in the patchy-zoned regions of some plagioclase phenocrysts have CI/K20 ratios as high and higher than do inclusions in the patchy-zoned cores. Sulfur decreases with increasing K20 for inclusions in olivines and plagioclases (Fig. 24). Included among the S-poor inclusions is one glass trapped between two crystals and one inclusion with a large vapor bubble. Water as estimated by difference shows weak positive correlation with S and ranges from 0 to 8 wt.% (Fig. 25). Individual analyses carry a statistical uncertainty (two standard deviations) of 3 wt.%. In general, the estimated concentrations of H:O are higher in the basaltic glasses poor in K20 and rich in A1203 than in the more potassic andesitic glasses poor in Al:O3. GROUNDMASS COMPOSITION OF FUEGO ASHES

The groundmass component of Fuego ash was analyzed after mineral separation of phenocrysts from several of the 1974 samples. Inspection of the separates shows that a component of small microlites of plagioclase and mafic minerals remains in the material. The groundmass analyses, summarized in Table 7, are less differentiated than the compositions of residual glass and are similar in composition to the average glass inclusion. SOLUBLE ELEMENTS ON ASH

Soluble S, CI and other elements are found in significant concentrations on the Fuego ash samples. A summary of results of analyses of leachates is given in Table 8. The soluble material deposits on the ash surfaces during Fig. 20. Aggregate of crystals of olivine (O), plagioclase (P) and magnetite (M) from ash flow 83 (October 14) seen in plane polarized light. The smallest scale bar divisions are 10 ~m apart. The olivine crystals are euhedral to subhedral and separated from each other by thin films of glass. In the olivine crystals are inclusions of plagioclase, glass and magnetite. The section is doubly polished, consequently the olivine does not display the usual relief. Fig. 21. High-magnification view of a bleb o f sulfide (S) and crystal of magnetite (M) in an inclusion of glass (G) in a phenocryst of olivine from ash fall sample 178 (October 23). The scale divisions are 10 um apart. The olivine (O) partially surrounds the glass which abuts the e p o x y mounting medium (E) at the bottom of the picture. Tiny bubbles of vapor and microlites are peripheral to the magnetite grain. Another larger sulfide bleb in a magnetite phenocryst was found by microprobe analysis to contain 52.6 wt.% Fe, 1 2 . 0 wt.% Cu, 35.0 wt.% S, and 0.4 wt.% Ni. Minute grains of.magnetite (?) comprise less than 1% of the sulfide blebs.

98.96

Total

1.04

0.45 -0.32 0.61 91.3 1.93 109

Eu

0.48 10.7 425 559 7.76 16.24 2.84

Tb Yb LU Th Zr Hf Cu

Cs Rb Ba Sr La Ce Sm

48.4--52.3 19.6--21.2 8.5--10.1 3.31--4.43 8.7--10.1 3.2--3.8 0.62--0.95 0.88--0.95

0.43--0.52 8.8--15.0 350--546 540--581 6.88--8.63 14.3--18.18 2.5--3.18 0.98--1.1 0.41---0.48 -0.28--0.41 0.58--0.63 79.1--110 1.71--2.4 80--140

Trace elements ( p p m )

50.6 20.5 9.2 3.92 9.6 3.5 0.74 0.90

SiO~ Al203 Fe203 MgO CaO Na=O K20 TiO2

Major element oxides (wt.% )

range

Oct. 14 (14 samples)

98.98

51.8 19.4 9.4 4.36 8.8 3.6 0.86 0.82

0.39 11.7 509 542 8.42 17.92 2.91 1.05 0.49 2.13 0.34 1.11 88.9 2.1 123

u

* 9.7--13.9 435--598 530--571 * * * * * * * * 80.6--99.3 * 115--130

50.8--52.6 18.9--20.2 8.7--10.4 3.75--5.19 8.1--9.5 3.5--4.1 0.81--1.08 0.78-0.98

range

Oct. 17 (5 s a m p l e s ) .

0.55 9.3 413 511 7.19 15.29 3.28 1.04 0.4 2.4 0.33 0.60 77.9 2.03 106

99.46

50.3 18.6 10.8 5.53 8.7 3.4 0.75 0.98

8.3 300 524 5.49 12.85 2.17 0.80 0.58 0.22 0.70 69.7 1.37 99

* * 66.3--89.3 * 80--130

m

99.4

48.3 18.4 11.8 7.7 8.8 2.9 0.61 0.89

* * 64.8--75.1 * 90--120

5.4--11.4 260--345 483--557 * * * * *

47.0--49.9 17.6--19.5 10.6--13.7 5.79--9.67 8.5--9.2 2.6--3.3 0.5--0.71 0.75---0.98

range

Oct. 23 (4 s a m p l e s )

6.5--12.2 317--494 496--525 * * * * *

47.8--51.7 17.7--18.8 9.7--12.3 4.8--7.93 8.4--9.4 2.8--3.7 0.6--0.83 0.95--1.03

range

Oct. 19 (8 s a m p l e s )

Chemical c o m p o s i t i o n o f ash f r o m t h e f o u r pulses o f t h e 1974 e r u p t i o n o f F u e g o v o l c a n o

TABLE 4

0.31 0.82 84.7 1.89 112.3

10.48 432 540 7.44 16.14 2.77 0.99 0.49

99.36

50.49 19.43 10.35 5.01 9.03 3.41 0.76 0.88

Whole eruption* * (weighted average)

b~ b~

84.3 977 20.2 28.18 17.28 24.71

60--105 930--1000 13.3--30.4 23.6--32.75 * 23.3--26.12

85.5 988 26.4 30.07 11.46 25.79

~

57.5--102.5 950--1060 21.5--32.5 * * *

range

Oct. 17 (5 s a m p l e s )

86.6 1124 34.5 34.3 15.25 28.9

65--95 1075--1250 21.3--49.7 * * *

90 1300 42.6 63.13 32.66 24.3

62.5--105 * 35.1--60.9 * * *

range

~

u

range

Oct. 23 (4 samples

Oct. 19 (8 samples

* Only one value in group measured for these elements. **Weighted by volume (30% Oct. 14: 4 0 % Oct. 17, 1 0 % Oct. 19, 2 0 % Oct. 23.; see text).

Zn Mn Ni Co Cr Sc

Trace elements ( c o n t i n u e d )

range

Oct. 14 (14 s a m p l e s )

T A B L E 4 (continued)

86.2 1060 28.7 36.5 17.8 25.5

Whole eruption * * (weighted average)

tO c.O

24

TABLE 5 Microprobe analyses of glass inclusions within crystals of ash from the 1974 eruption of Fuego volcano Host

Plagioclase (26 analyses)

Olivine (18 analyses)

Magnetite (3 analyses)

Overall average (47 analyses)

SiO:

51.9 + 2.7 (48.1--56.2) 15.8 -+ 1.8 (13.1--19.4) 10.1 _+ 2.2 (7.0--12.7) 4.3± 1.2 (3.3--5.1) 7.6 + 1.3 (6.4--11.1) 3.7 + 1.0

51.5 -+ 1.9 (48.1--56.0) 16.9 +- 0.9 (15.6--18.2) 9.5 -+ 0.8 (8.3--11.4) 4.0 +- 0.6 (3.0--4.8) 8.0 ± 0.7 (7.1--9.4) 3.8 ± 0.6

50.8 (50.1--51.6) 16.3 (16.1--16.6) 9.6 (9.3--10.1) 4.1 (3.9--4.4) 8.0 (7.9--8.1) 4.3

51.7

Al~O3 FeO* MgO CaO Na20 K:O TiO:

P:Os Cl S H~O

(2.2--5.8) 1.13 ± 0.31 (0.5--1.58) 1.31 _+ 0.41

(2.7--4.6) 0.85 +- 0.18 (0.57--1.32) 1.09 ± 0.19

(4.2--4.4) 0.92 (0.84--1.01) 1.02

(0.7--1.9) 0.29 -+ 0.06 (0.2--0.4) 0.12 +_ 0.03 (0.07--0.19) 0.13 + 0.12 (O--O.5) 3.2 -+ 2.7 (0--8.7)

(0.67--1.45) 0.28 -+ 0.05 (0.2--0.3) 0.11 -+ 0.02 (0.07--0.16) 0.20 +0.16 (0.02--0.50) 2.6 + 2.1 (0--6.9)

(0.8--1.15) n.d.

16.3 9.8 4.4 7.8 3.8 1.0 1.2

0.29

0.1

0.11

0.08

0.15

2.8 (1.7--3.9)

2.9

n.d. = not determined. e r u p t i o n a n d f a l l o u t a n d can be e x t r a c t e d b y H 2 0 - l e a c h i n g ( T a y l o r a n d S t o i b e r , 1 9 7 3 ) . O f course, s a m p l e s w h i c h h a v e b e e n a f f e c t e d b y rain or o t h e r surface m o i s t u r e have lost these chemicals. F u e g o ' s ashes s h o w e d significant c h a n g e s in t h e a m o u n t s a n d ratios o f soluble c o n s t i t u e n t s as t h e 1 9 7 4 e r u p t i o n p r o c e e d e d (Rose, 1977). Fig. 26 s h o w s t h e p a t t e r n s o f c h a n g e o b s e r v e d . O n e set o f e l e m e n t s (S, Ca, Na, Mg, Zn) s h o w an a p p a r e n t c o r r e l a t i o n w i t h t h e i n t e n s i t y o f a c t i v i t y , a n d clearly decrease in t h e final phases o f activity. O t h e r e l e m e n t s (C1, K, Ba, Cu) s h o w n o t e n d e n c y t o decrease in t h e l a t t e r phases o f activity. A large range o f conc e n t r a t i o n s o f t h e s e soluble e l e m e n t s is s h o w n f o r e a c h s u b g r o u p . T o g e t h e r w i t h t h e analyses Of t r a p p e d glass inclusions, t h e soluble m a t e r i a l a d s o r b e d o n t h e fresh F u e g o ashes gives i n f o r m a t i o n o n t h e c o m p o s i t i o n o f t h e e r u p t i v e gas. R o s e ( 1 9 7 7 ) has d e s c r i b e d h o w t h e s e d a t a c a n be i n t e g r a t e d i n t o a " b u d g e t " o f fugitive e l e m e n t s f o r t h e e r u p t i o n . C o n c l u s i o n s are: (1) Soluble S a n d Cl o n ash p r o b a b l y result f r o m scavenging o f t h e gas p l u m e b y t h e ash.

25

(2) Soluble S and C1 on ash can be used as a remote sensor to detect relative changes in the eruptive gases during an ash eruption. (3) A significant portion (10--40%) of the total S and C1 erupted is quickly returned to earth by the ash scavenging.

TABLE 6 Compositions* of inclusions of low-K20 glasses in phenocrysts, Fuego volcano, October 1974 eruption 1

2

3

4

5

6

7

SiO~ Al203 FeO** MgO CaO Na20 K~O TiO 2 P20~ S C1 H20***

54.0 16.8 10.4 5.0 8.0 3.0 0.9 1.3 0.1 0.14 0.13 4

53.5 17.8 9.7 4.8 8.1 4.3 0.8 1.0 n.d. 0.08 0.07 4

54.0 1~1.8 11.8 5.2 7.2 3.8 0.9 1.7 n.d. n.d. 0.16 2

51.0 19.1 10.0 5.4 9.3 3.0 0.6 1.1 n.d. n.d. 0.09 5

51.9 18.1 10.4 5.1 9.7 3.2 0.6 0.7 n.d. 0.13 0.10 3

54.5 18.1 9.4 3.5 8.3 3.7 1.0 1.2 n.d. 0.10 0.12 6

50.5 19.4 9.3 5.0 9.0 3.4 0.8 0.9 n.d. n.d. n.d. n.d.

Sum*

95.2

93.1

96.9

94.3

97.1

93.8

98.4

Explanation of columns: 1 = inclusion of glass in olivine (Fo~3), sample VF74°50, October 1 4 . 2 = inclusion of glass in olivine (FoTs), sample VF74-50, October 14. 3 = inclusion of glass in plagioclase (Ang0), sample VF74-131, October 17.4 = inclusion of glass olivine (Fo79), sample VF74-131, October 1 7 . 5 = inclusion of glass in olivine (Fo~3), sample VF74-178, October 23. This inclusion also contains a crystal of magnetite (4.4% TiO~, 0.2% Cr~O3). 6 = inclusion of glass in plagioclase (Ans,), sample VF74-178, October 23. This inclusion is only 15/zm in diameter and is in a crystal which also contains an inclusion of pargasitic hornblende. 7 = weighted average composition of October 1974 eruption, from Table 4, with Fe as FeO. * Compositions determined by electron microprobe analysis using an energy-dispersive (solid-state) detector for the first 9 elements listed with an uncertainty of about 2% of the amount reported for the major oxides and up to 10% of the amount reported for the minor elements. Sulfur and CI analyses are based on focusing crystal spectrometers and have an uncertainty of about 0.01 wt.% absolute. The tabulated oxides are normalized to 100%. * * A l l Fe is reported as FeO. *** H20 is estimated by difference between the initial sum of the analysis as oxides and t h e sum of a concurrent analysis of an anhydrous glass (see Anderson, 1974). The two standard deviation error is about 3 wt.% absolute for an individual analysis. *The sum found in the original analysis of the oxides (excluding H20 ) before normalization to 100%. The sums of the normalized values may he slightly different from 100 due to rounding errors. n.d. = not determined.

26

TABLE 7 Major element compositions of g~oundmass separates of 1974 Fuego ash Groundmass separates (5 analyses) SiO2

Recalculated to 100% (H20-free)

52.34 (52.0--52.6) 18.74 (18.0--20.2) 9.06 (8.71--9.23) 3.40. (3.21--3.42) 9.36 (8.5--10.3) 3.87 (3.84--3.99) 0.77 (0.69--0.86) 1.16 (0.95--1.38)

Ai20 ~ FeO* MgO CaO Na20 K:O TiO2 Total

98.70

II-

Glass inclusions average (recalc.; Table 5)

Residual glass in lapilli

53.03

53.9

62.0

18.99

17.0

14.7

9.18

10.2

8.1

3.44

4.6

2.2

9.48

8.1

4.5

3.92

4.0

4.4

0.78

1.0

1.9

1.18

1.3

1.6

100.00

100.1

99.0

s~, ppm Rb

i (~

'\

\

c',-% Fe 0 ~

9

x I0 %MnO

Q, ppm ' L.Q 6

I

/

~02 "

O/oBo o \ * I00

d//

ppmNi

J

I4

;J

x 50

'

%MgO

4

\ 117 119 OCTOBER

215

14

I

I• 119 OCTOBER

215

Fig. 22. Graphical summary of changes with time of selected chemical constituents of Fuego ash during the 1974 eruption (data from Table 5).

188 569 174.5 27.5 435 34.9 0.42 0.34 2.24 0.071 0.36

98--250 113.5--794 140--230 14--44.5 250---975 23--47.5 0.22-0.74 0.002--1.48 1.6--2.8 0.012-0.33 0.22--0.48

range

87.8 625 183.8 19.9 531 33.5 0.27 0.64 0.77 0.11 0.27

u

38--130 485--846 150--215 16.5--23 140--880 27.5--40.5 0.18--0.38 0.11--2.25 0.05--1.6 0.08--0.15 0.16--0.40

range

October 17

210 569 177.1 20.3 556 23.5 0.48 0.38 1.98 0.112 0.55

~

October 19

*E is enrichment factor for that element (relative to Mg) over the ash.

CIS Na K Ca Mg Ba Zn Mn Cu Se

October 14

108--343 227--972 141--220 12.5--28 210---915 15--36 0.19-0.84 0.001--0.84 1.25--3.3 0.082-0.144 0.32-0.64

range

134 271 110.4 38.3 145 10.5 0.61 0.04 0.54 0.185 0.20

October 23

63--250 109--571 61--273 8--113 75--335 7--14.4 0.4--0.79 0.01-0.113 0.15-0.9 0.18-0.19 0.06-0.48

range

139 532 166 25.9 428 28.3 0.40 0.40 1.21 0.099 0.31

Whole eruption (weighted average)

Soluble chemicals on Fuego 1974 ash from each pulse of the eruption (values expressed in ppm relative to weight of ash)

TABLE 8

670 7500 5.9 4.4 15.3 -0.1 5.1 1.2 1.0 ?

E*

t,O

28

(4) S/C1 ratio in the eruptive gas was apparently highest in the most intense phase (October 17) of the Fuego eruption and decreased thereafter. (5) Most of the Na ÷, K ÷, Ca 2÷, Mg ~÷ and other cations detected in the ash leachates probably was leached from the silicate ash by the scavenged acids. (6) The dominant salts which coat the ash are probably CaSO+ (hydrated) and NaC1.

0.20

~.gERII/R/

WT.% CI

I X

0.15"

4-

1 4 17 23 OL 0 • • PL ~" @ F7

MT * RG

X +

o.os o!+

,!o

,!s

2:0

NT.% K20 Fig. 23. C1-K20 variation plot for inclusions of glass in phenocrysts of olivine (OL), plagioclase (PL) and magnetite (MT) and for residual glass (RG) from three major pulses of the eruption (14 = October 14 ash samples 50 and 53; 17 = October 17 ash sample 131, and 23 = October 23 ash sample 178). The line drawn through the low-K20/high-C1 inclusions in olivines is a line of constant CI/K20 ratio (0.15) and indicates the expected behavior of CI and K20 in a gas-free solidifying magma. Most inclusions in plagioclase and some inclusions in olivine have ratios of CI/K20 less than 0.15 consistent with preferential loss of C1 probably caused by effervescence.

• in ol

wt%S

o in pl

.2

• in mag

[]

ergo,

O []



[]

.15 [] 0[] E3

[3

[] •

[]

~,& O D OeO [3

[]

.05

[]

% 0!5

i

1.0

I"1

wt% K~O Fig. 24. S-K20 variation plot for trapped glass inclusion analyses from phenocrysts in three samples of 1974 Fuego ash.

29

@

~rror

v@

I

@

,t

%4

[] V

V

#

H20

[]

@

V

v

pl Vol 0 mag

oV~

t• v

I

!

I

.1

wt %S

.2

I

I

.3

Fig. 25. H20-S variation plot for trapped glass inclusions in phenocrysts of Fuego 1974 ashes. I000

x

CI

I00

~o

I0

ppm

1.0-

c° 0.1

ll4

1'7

1'9

23

OCTOBER

Fig. 26. Graphical summary of changes of leachate chemistry of volcanic ashes erupted successively from Fuego, October 1974 ( data from Table 9).

30

DISCUSSION AND INTERPRETATION

Composition of the parental liquid of the Fuego magma Two lines of evidence support the hypothesis that the weighted average composition (WAC, Table 4) can be considered as a parent magma for the 1974 eruption. First, its composition is similar to commonly observed ashes from the 1974 eruption and other Fuego activity (Fig. 27). The scattering of points with MgO > 6% in Fig. 27 are interpreted as samples which have accumulated olivine and pyroxene. The greatest density of points in Fig. 27 is in the region of 4--6% MgO. Secondly, the analyses of glass inclusions in Fuego phenocrysts support a parental magma similar to the WAC. The array of compositions of inclusions plotted in Fig. 28 defines an area outlined by lines connecting the extreme compositions with the respective MgO and A1203 concentrations in their host minerals. The concentrations of MgO and A1203 in the initial liquids should lie within the area of intersection and between the respective analyses of inclusions in olivine and plagioclase. From Fig. 28, these initial liquids fell along a path extending from about 4.8% MgO and 18% A1203 to about 4.0% MgO and 14% A1203. The initial liquids richest in MgO are also poorest in K20 (Table 6). Similar compositions of initial liquids are suggested by analyses of inclusions of glass in ash samples from October 17 (sample 131) and October 23 (sample 178). The compositions agree closely with the WAC and support our interpretation that it does represent the composition of the parental liquid. The parental liquid may differ from the parental magma because of the presence of crystals carried in the magma. In particular, the cores of type 1 plagioclases may have been carried with the parental magma from a reservoir of magma into a body where most of the crystallization and differentiation took place. However, we note that analyses 4 and 5 of Table 6, if adjusted for about 1% reabsorption of olivine have compositions which are within wt% Si02 [] 140ct 017 019

-55

• .-c~

.

• other

-50 O

I

4

6

8

,I

I

I

I

wt.°/oMgO

Fig. 27. Graph of MgO vs. SiO 2 for whole-rock (or bulk ash) samples from Fuego. Points called " o t h e r " represent analyses of materials erupted prior to 1974. The remaining points represent samples erupted in the four pulses of the 1974 eruption (as labeled).

31

w t % MgO

\o,\

°::::

15

10 !

I

I

I

I

10

15

20

25

I

w t Yo AI203

Fig. 28. MgO-A1203 plot of trapped glass inclusion data from phenocrysts in a sample of October 14 ash of Fuego's 1974 eruption designed to help show how to estimate the composition of the parental liquid of the Fuego magma. Inclusions of glass in plagioclase may react with the plagioclase host and lose or gain Al~O3. Inclusions of glass in olivine may lose or gain MgO. If other daughter minerals are present within the inclusion, then reaction relations are more complex. Generally no other minerals are present in the inclusions of glass in the Fuego phenocrysts. The composition of the initial liquid of each inclusion should therefore fall on a line connecting the present composition of the glass and its host mineral (see text for further explanation).

analytical error of the weighted average composition of the eruption except for CaO and K20. The differences for CaO and K20 are small and not meaningful to us. In sum, we conclude that the WAC (Table 4) adequately represents the parental liquid-rich magma which crystallized and differentiated to yield the texturally and compositionally varied eruptive pulses. If any significant amount of solid or liquid material related to the October 1974 eruption remains below Fuego, it probably has a composition close to this parental magma. Temperature and water pressure o f the Fuego magma Temperatures of quenching have been inferred from the distribution of MgO between glass and olivine (Roeder, 1974; Anderson and Sans, 1975). The temperatures assume the melt is anhydrous and range between 1010 and 1130°C with all but the lowest temperature falling in the interval between 1050 and 1130°C. The temperatures carry an uncertainty of +30°C due to calibration and analytical errors. Temperature correlates weakly with the apparent concentration of H20 (Fig. 29). We believe that this correlation is due to the enhanced solubility of olivine in an H20-rich melt at fixed temperature (Yoder and Tilley, 1962). Anomalous glass inclusions in plagioclase with > 17% A1203 (therefore requiring addition of host to the initial liquids) also have high concentrations of S and H20 and lack

32

T°C

/

~

'

'

'

'

r

"

o

x

,

I

L

I

2

4 H20

l150r

i

1100 r 1050 r 1000"

' 0

t

WT. %

i 6

Fig. 29. Graph of temperature of quenching and concentrations of H~O inferred for inclusions of glass in olivine phenocrysts from ash samples 50 (October 14, circles), 131 (October 17, crosses), and 178 (October 23, asterisks). The temperature of quenching is inferred from the distribution of MgO between glass and olivine (Anderson and Sans, 1974). The concentration of H20 is estimated by difference according to the procedure outlined by Anderson (1974). The precision error in temperature is -+30°C. For H20 the precision error is ± 3 wt.%, absolute. A suggestion of positive correlation between temperature and H20 is displayed.

borders of sodic plagioclase. Inclusions with low volatile content apparently crystallize sodic plagioclase borders after entrapment. We interpret that volatile-rich glass inclusions resorb their hosts. The resorption is best explained by an increase in temperature of the magma after entrapment, a result of loss of H20 and c o n c o m m i t t a n t liberation of heat of crystallization. In glass inclusions within olivine hosts we have estimated that resorption of the host equivalent to a b o u t 1% of the weight of the inclusion is required to explain the increase in MgO in the glass inclusions. This implies a temperature increase of 15--20°C, if one assumes that the change in solution of olivine in melt with temperature is the same for the Fuego melt with H20 as it is for dry tholeiitic basalt with olivine on the liquidus (Yoder and Tilley, 1962). We thus estimate that the temperature of beginning of intratelluric crystallization of the Fuego parent liquid is 20°C less than the temperature of quenching or about 1030 +50°C. The liquidus temperature of the high-alumina basalt investigated by Yoder and Tilley (1962) is 1060°C at 10 kbar of water pressure. The Fuego magma has a larger ratio of Fe to Mg than the Warner Basalt investigated by Yoder and TiUey. Since iron will lower the olivine liquidus of tholeiitic basalts (Tilley et al., 1967) we expect that the olivine liquidus of the Fuego magma would parallel that of the Warner Basalt b u t lie a b o u t 80°C below it. The olivine liquidus of the Warner Basalt passes through 1110°C (= 1030 + 80°C) at pressures of H:O between 2 and 3 kbar according to Yoder and Tilley. The concentration of H:O in the Fuego magma suggested by the above line of reasoning is a b o u t 5 wt.% and consistent with the average of 4 wt.% H20 in the inclusions of K20-poor glass (Table 6). Because the Fuego magma erupted traces of pargasitic hornblende and because pargasitic hornblende occurs as an inclusion in a core of a t y p e 1 plagioclase phenocryst which contains inclusions of basaltic glass in surrounding patchy-zoned regions (Fig. 16), our inferred temperature and pressure of H:O for the onset of crystallization should be consistent with the stability of

33

pargasitic amphibole. According to Holloway (1973) pargasite is stable to 1030°C at a water saturation pressure of 2 kbar, to 1050°C at a water saturation pressure of 4 kbar and falls below 1000°C at a water saturation pressure below 1 kbar. Our estimated conditions of beginning of crystallization (1030 + 50°C and PH~O = 2--3 kbar) are consistent with the stability of pargasite. In sum, we think our estimated conditions are sound.

Oxygen fugacity of the Fuego magma The oxygen fugacity of the Fuego magma is estimated to lie between 10 -9"s and 10 -1°"7 at 1050°C. This estimate is based on the relation between the concentration of sulfur (Sin) in the melt in equilibrium with a sulfide melt and magnetite. The results correspond to extremes of composition of coexisting sulfide blebs, magnetite phenocrysts and inclusions of basaltic glass in the Fuego samples. The oxygen fugacity is calculated according to: 2880 log fo2 = 6 log S m - 29.84 - - 6 log Cs - 6 log aFe S + 2 log T

aFe30 '

where T is in °K, Cs is the su!fide-carrying capacity of the melt at 1200°C (calculated according to Haughton et al., 1974) and aFeS and aF%O, are the activities of the respective compounds. Ideality was assumed. The equation above is derived from the work of Haughton et al., using the thermochemical data of Robie and Waldbaum (1968) for AG~ of FeS and Fe304. The equation refers to S 2- in the melt, thus it is necessary to assume that all of the analyzed S in the melt is S 2-. This seems plausible in view of the presence of blebs of FeS-rich sulfide. By comparison, estimates of the oxygen fugacity of Kilauea tholeiite axe found to be between 10 -9.' and 10 -'0.6 at l l l 0 ° C compared to 10 -8"s inferred from Fe-Ti oxides. The important point is that sulfur concentration in melt is a sensitive indicator of fo~ if it is buffered by FeS and FeaO4 and the concentrations of S in Fuego and Kilauean melts are comparable at similar temperatures. Consequently, the oxygen fugacity of the parent melt of Fuego is comparable to that of ferrobasalts from Kilauea, Hawaii.

Density and viscosity of the Fuego magma We have calculated viscosities and densities for the Fuego magma (Table 9) according to the methods of Bottinga and WeiU (1970) and Shaw (1972). All iron has been taken as FeO. The estimated densities are sensitive to H20, but n o t to temperature. The estimated viscosities depend heavily u p o n temperature as well as H20 content. The most appropriate values of viscosity and density for the purposes of crystal settling are those intermediate between the bulk rock composition and the groundmass composition at 3% H20 and 1040°C. We consider 3% H20 an appropriate average for a melt whose H20 concentration dropped from a b o u t 5% to 0% with crystallization. The values

34

TABLE 9 Density and viscosity of Fuego magma 1

2

3

4

5

6

Composition

H~O (%)

P,0~0

P~,40

~,o:o

~,4o

G1 GM WR G1 GM WR Gl GM WR G1 GM WR Gl GM WR

0 0 0 1 1 1 2 2 2 3 3 3 5 5 5

2.60 2.59 2.61 2.52 2.52 2.54 2.46 2.45 2.47 2.39 2.39 2.40 2.28 2.27 2.28

2.60 2.59 2.61 2.51 2.51 2.53 2.44 2.43 2.45 2.37 2.36 2.38 2.24 2.23 2.25

13,000 17,000 7,600 4500 5600 2800 1700 2100 1100 710 860 490 160 190 120

2500 3000 1500 940 1100 610 400 470 270 180 220 130 50 50 40

Explanation of columns: 1 = compositions used are: GI = average glass inclusion of Table 6, GM = average groundrnass of Table 8, and WR = whole-rock weighted average of the eruption of Table 5.2 = Weight percent of H20. 3, 4, 5, and 6 = densities and viscosities at 1020 and 1140°C respectively.

are: p = 2.39 g / c m 3 and ,? = 530 poise. The a p p r o p r i a t e viscosity f o r the extrusion o f t h e m a g m a c o r r e s p o n d s t o t h e u p p e r , w a t e r - p o o r p a r t o f t h e m a g m a b o d y w h i c h we estimate b e l o w t o o c c u r at 5 0 0 - - 1 0 0 0 a t m pressure c o r r e s p o n d ing t o 1--3 wt.% HzO and a t e m p e r a t u r e o f 1050°C. The range o f viscosity i n d i c a t e d f o r t h e g r o u n d m a s s u n d e r these c o n d i t i o n s is 3 7 0 0 t o 6 0 0 poise. I n t e r a c t i o n s b e t w e e n crystals will stiffen the F u e g o m a g m a b y a f a c t o r o f f o u r or m o r e , d e p e n d i n g u p o n the reliability o f t h e R o s c o e c o r r e c t i o n (see discussion in Shaw, 1972). C o n s e q u e n t l y , the a b o v e viscosities are minimal.

Extrusion and mixing of the Fuego magma T h e m a x i m u m v o l u m e t r i c rate o f e x t r u s i o n can place a helpful limit o n the d i a m e t e r o f t h e vent. A c c o r d i n g t o pipe f l o w the v o l u m e t r i c flow rate, Q, equals (~r4/~t) × ( A P / h ) w h e r e r is the radius o f t h e pipe, ~ is t h e viscosity, and AP/h is t h e pressure gradient in excess o f the h y d r o s t a t i c gradient. F o r b u o y a n t ascent, we take AP/h = Apg w h e r e Ap is t h e d i f f e r e n c e in d e n s i t y b e t w e e n t h e wall r o c k and m a g m a ( a b o u t 0.3 g / c m 3) a n d g is 9 8 0 c m / s 2. With - 3 7 0 0 t o 6 0 0 poise a n d Q = 4 × 109 cm3/s, r = 2 . 3 - - 3 . 6 m. Because the h y d r o s t a t i c pressure gradient will diminish as s o o n as flow c o m m e n c e s , the value o f r inferred f r o m t h e rate o f e x t r u s i o n is a m i n i m u m . Because the pres-

35

sure gradient affects r only as the fourth root, a tenfold decrease in pressure gradient will increase r by a factor of only 1.8. The observable vent of Fuego is a b o u t 20 m in radius and probably is enlarged by explosive effervescence and degassing of the magma near the summit. In subsequent portions of this paper we rely upon the successive episodes of the eruption to give us a spatial view of the b o d y of magma beneath Fuego. The evidence which we discuss in the following sections suggests that vertical position is one, if not the only spatial parameter reflected in the makeup of the successive eruptions. Other possible parameters include horizontal position and mixing ratios of separate but co-extruding bodies of magma and rock. The purpose of this section is to put together and discuss evidence which bears on the above possibilities. It is important to note that individual samples of ash include material extruded over spans of time as long as t w o days. Therefore differences between lapilli in any ash sample may arise both because of time-related changes in extruded material and because of gradients present in magma which extruded at the same time. Within individual lapilli and b o m b s there occurs a wide assortment of crystals of plagioclase. Therefore the various types of plagioclase phenocrysts existed close together in the magma before fragmentation and extrusion. Some individual phenocrysts have several inclusions of glass with varied chemical compositions. Many of the differences can be explained by fracturing of selected inclusions followed by gas loss and crystallization. Some variations suggest successive entrapment from variably differentiated melts, however. Differences between inclusions in different crystals are larger. In particular it is n o t e w o r t h y that the concentration of K20 may vary by a factor of 3 in inclusions of glass in phenocrysts from a single ash sample. If K20 is completely stored in the liquid, then a three-fold enrichment in K20 corresponds to 67% of crystallization or a b o u t twice the observed percentage of phenocrysts. Part of the range in phenocryst modes of individual lapilli listed in Table 3 may reflect variable crystallization of co-extruded magma, but much of the range in phenocryst modes is statistical in origin. Each extrusive pulse contains phenocrysts which grew in part during the first few percent as well as after 50% of solidification. Although the above facts indicate that some portions of the magma which extruded during individual eruptive pulses were more crystallized than others, the facts neither require nor exclude different bulk chemical compositions of the variously crystallized portions. The decreases with time of vesicle size (Table 3) and of surface adsorbed sulfur (Fig. 26) suggest that the later extruded ashes were derived on the average from greater depths. At greater pressures vesicles would be smaller. The fraction of volatile ingredients present in the gaseous state would be less at greater pressure also. If the principal spatial parameter reflected by the successive ashes were horizontal distance from cold walls, then first extruded material might be expected to come from the hot, fluid center of the b o d y with a low c o n t e n t of crystals and vesicles followed by wall material enriched in crystals and vesicles, opposite to the observed trends. It is conceivable that hotter magma would vesiculate more readily and yield the observed sequence

36

as a result of increasing efficiency of vesiculation with increasing temperature~ However, the variable grain size of microphenocrysts in separate lapilli is suggestive to us of co-extrusion of bits of magma which were variously cooled before eruption. In sum, we infer that variously crystallized and cooled portions of the magma body mixed together during the individual eruptive episodes, and the average trend from beginning to end of the eruption is suggestive of extrusion of generally deeper levels of the magma body with time.

Differentiation o f the parent magma Because most of the petrographic and chemical variations observed suggest crystal fractionation prior to eruption, ashes which fell during each of the four periods of intense activity possibly represent progressively deeper levels of a body of differentiating magma beneath Fuego. To characterize the differentiation in the magma body, we first made some simplifying assumptions: (1) That~he composition of the parental magma is given by the weighted average composition of the magma extruded during the October 1974 eruption. (2) That the differentiation was by crystal fractionation and that bulk compositional differences are explained only by different proportions of phenocrysts and groundmass components. Assumption 1 is justified in light of the preceding discussion of the composition of parent magma. The petrographic and compositional evidence presented above generally supports assumption 2. Phenocrysts, groundmass and glasses have measurable ranges in composition for each of the eruptive pulses. The ranges in composition for plagioclase (An9s-80) and olivine phenocrysts (Fo76-66) and inclusions of glass are comparable for each pulse. It is likely that the average compositions of the phenocrysts from the different pulses are within the observed ranges. Variations of compositions within the ranges observed would not significantly affect our results or conclusions. Following the above assumptions, least squares calculations (Wright and Doherty, 1970) were made to show what mixes of crystals would have to be added to or subtracted from the weighted average (parent magma) composition to derive the magmas of each of the four pulses of the eruption (Table 10, Fig. 30). Trace element calculations were also used to test the consistency of the model (Table 10) and employed a Rayleigh and mixing calculation. Rayleigh calculations used the distribution coefficients given in Table 11 and partitioned the trace elements between glass and the crystals for the weighted average composition and weighted average mode. Mixing calculations were then made to model the whole-rock trace element concentration changes which would result from the addition and subtraction of crystals suggested from major element mixing for each eruptive pulse. In general, reasonable agreement of observed and predicted trace element values is found and we conclude that such a differentiation model is consistent with the data we collected, within limits of uncertainty in both the analytical data and the distribution coefficients. For the first pulse the differentiation involved some

38

Froctionol model

Observed mode

14-

/

(

~_17ol 0 cD

Calculoted mode

<

!

019-

2 25-5

0 +5

5

0 +5

PLAG,

5

0 t5

%

}4-

~17z~ o

\

019-

250 +5

-5 0 -5 O L + C P X + MAG, °/o

0 +5

Fig. 30. Comparison of modes of Fuego magmas erupted during each pulse of the 1974 eruption. Each m o d e is expressed relativeto the weighted average m o d e of the entire eruption. Negative values indicate loss of plagioclase or mafic minerals, positive values indicate additions. Center column is observed mode. Column on left is m o d e obtained by fractionation from parent based on whole rock compositions of the weighted average for the eruption and each of the four principal pulses (Table 12). Column on right is m o d e obtained by independent mixing calculations using groundmass/glass analyses (Table 8) and whole-rock data for each pulse. Both the mixing calculations performed used microprobe determined compositions for plagioclase,olivine,clinopyroxene and magnetite T A B L E 11

Distribution coefficients used in trace element Rayleigh calculations*

Ba Ce Cr Eu La Lu

Labradorite

Clinopyroxene

0.01 0.12 -0.34 0.I 0.06

0.026 0.15 10 0.51 0.I 0.56

0.01 0.007 5 0.007 0.01 0.016

--27 ----

--

Rb

0.07

0.03

0.01

--

Sm Sr

0.067 2

0.5 0.12

0.007 0.014

--

--

2

13

Magnetite

Ni

Co

5

Olivine

(5)

15

-8

* S o u r c e s : Gill ( 1 9 7 4 ) , A r t h ( 1 9 7 6 ) , L e e m a n ( 1 9 7 6 ) , E w a r t et al. ( 1 9 7 3 ) .

39

addition of plagioclase, and removal of other minerals, while the subsequent pulse involved removal of all minerals. The ashes erupted on October 23 (final pulse) may be accumulative, perhaps representing the base of the magma body. Fig. 30 shows results testing the consistency of the fractionation model. In general, the changes in proportion of minerals and groundmass predicted in the model are observed in the petrography, and reiterated in mixing calculations using groundmass and mineral compositions. Such tests support the model (see also following section on groundmass composition). Mixing calculations (right-hand column of Fig. 30) performed using wholerock data (Table 4) together with data from either groundmass separates (Table 7) or from glass inclusions in phenocrysts (Table 5) give results which generally agree well with each other and with the observed petrography of the ash. More plagioclase (about 5%) must be mixed with mafic components and glass from inclusions to model bulk composition than is the case if groundmass separate compositions are mixed. The difference arises from two factors: (1) groundmass compositions are contaminated by small amounts of microphenocrysts of plagioclase and, (2) glass inclusions within phenocrysts (many of which are plagioclase) may be subject to post-entrapment crystallization of plagioclase, which would change their composition. Thus, an estimate of the composition of the uncrystallized portion of the Fuego magma can be obtained from the groundmass and glass inclusion data (Table 7). Use of a single constant estimate of the composition of uncrystallized magma in mixing calculations to show model changes associated with differentiation reproduces the modal changes obtained by direct petrographic observation and by calculations shown in Table 10 (see Fig. 30). Thus, a simple mechanical differentiation of the Fuego magma explains most of the chemical and mineralogical diversity observed.

Crystallization of the Fuego magma The range in compositions of inclusions of glass in phenocrysts and the zoning of the plagloclase phenocrysts demonstrate that plagioclase (An9s-90), olivine (Fo67-73), magnetite (4% TiO2) and probably pargasitic hornblende crystallized from a basaltic liquid close in composition to the parental magma. Most of the olivine crystallized from a basaltic melt with less than 1% K20. Most plagioclase nucleated homogeneously and primarily before or during episodes of cellular growth characterized by patchy zoning and high concentrations of inclusions of glass and gas. Uncommon augite crystals probably grew mostly within the rigid walls of the body of magma because they are usually intergrown with plagioclase and rarely occur as isolated euhedral crystals. With enrichment of the liquid in SiO2 and K20 the olivine crystals are about constant in composition but the plagioclases become more sodic and the magnetities richer in TiO2. The evidence for the crystallization of magnetite from a basaltic melt deserves emphasis, because of the controversy regarding the role of magnetite

40

in orogenic magmas. Eggler and Burnham (1973) recently found that magnetite does not form near the liquidus of a Mr. Hood andesite at reasonable conditions. The magnetite phenocrysts from Fuego contain inclusions of basaltic glass and occur as inclusions in olivine and plagioclase phenocrysts which also contain inclusions of basaltic glass. We conclude that magnetite did crystallize from the basaltic magma of Fuego. The only alternative explanation that occurs to us is that the magnetite (and other minerals) and its included glass might be decomposition products of amphibole. In the latter case the compositional similarity between the glass in magnetite with that in the other phenocrysts and with the bulk rock would be coincidence. Such a coincidence is unlikely because the amphibole is nepheline normative and would be expected to yield a nepheline-normative melt together with plagioclase, pyroxene, olivine and magnetite. Additional inferences regarding the history of crystallization of the body of magma are possible based on interpretation of the patterns of growth of the zoned plagioclase phenocrysts. Such interpretations are discussed in detail by Anderson (in preparation) and summarized below. The subhedral cores of large type 1 plagioclase phenocrysts are inferred to have formed in a comparatively large and compositionally uniform body of magma because such cores are subequant, weakly zoned and lack oscillatory zones. The comparatively large size and ratio of volume to surface area of the cores suggest comparatively slow crystallization rates which would permit equant growth with few inclusions of melt. Slow crystallization implies either relatively slow cooling expected for a thick body of magma or slow decompression if the magma is saturated with vapor rich in H20, or both. Large size and/or stagnation are expected attributes of a reservoir of magma, compared to a conduit or feeder. The subround shape of the cores (Fig. 9) suggests resolution possibly caused by either heating or decompression in the absence of an aqueous gas phase. Sinking from cool top towards hotter interior of a magma body could explain heating and resorption. Oscillatory-zoned and patchy-zoned regions which mantle the cores (Figs. 9 and 14), and which constitute crystals lacking weakly zoned cores (Fig. 10) are interpreted to reflect more rapid growth than the cores. Subregular episodes of crystallization of sets of sodic oscillations and subequally spaced intervals of crystallization of patchy-zoned regions rich in inclusions of glass and gas are interpreted by Anderson (in preparation) to result from supercooling caused by vapor-saturated decompression. In sum, the several types of plagioclase phenocrysts possibly reflect initiation of growth at,different stages of ascent (decompression). The cores of type 1 crystals grew slowly, probably in a thick body of magma and suffered resorption probably during the initial stages of ascent of the magma when it probably was unsaturated with vapor rich in H20. Type 2 crystals began to grow during periods of accelerated decompression. Type 3 crystals and oscillatory-zoned regions of other crystals grew during ascent which probably was periodic.

41

Models of crystal settling and flotation Table 12 lists estimated rates at which the phenocrysts and vapor bubbles might move through the Fuego magma. The estimates assume Stokes behavior and a constant density and Newtonian viscosity of the melt. They are maxima because we used minimum values for viscosity obtained from Table 9. The observed changes in the sizes and model proportions of the phenocrysts, if interpreted in terms of models of crystal settling, may help constrain the vertical dimension of the b o d y of magma and/or the time available for settling. Future investigations are planned to examine the size distributions of Fuego olivines in more detail.

Movement of crystals and the history of the Fuego magma To get an idea of the time scales involved in crystal movement, consider the larger olivine phenocrysts in the last pulse (Table 3). We assume as a starting point of reference a spherical b o d y of magma. If it initially contains all of the erupted magma (0.1 kin3), it is 288 m in radius. The upper eighttenths of the total volume lies above 60 m from the base. The 2-mm-diameter crystals would have settled a b o u t 230 m into the sphere. For an anhydrous magma (Table 12) the time is a b o u t 2 years. For the condition with 3% H20 the time is a b o u t 6 weeks. The time required increases as the b o d y of magma TABLE 12

Settling rates of Fuego phenocrysts and rise rates of vapor bubbles

Plagioclase (p = 2.70) Olivine (p = 3.53) Pyroxene (p = 3.39) Magnetite (p = 5.19) Bubbles (p < 0.01) Bubbles (1000 arm, p = 0.08)

V ~ (3% H20)

V*

V* (dry)

V*

16 100 50 60 14 0.6

60 1500 1100 400 5000 200

0.3 5 2 3 0.9 0.04

1 70 50 20 300 14

We assume the following: (1) The viscosity ~ = 530 poise for composition with 3% water and ~ = 9040 poises for composition which is anhydrous at 1040°C. (2) There is negligible yield strength o f melt (see text). (3) The density (p) of melt = 2.39 for composition with 3% water and p = 2.60 g/cm 3 for composition which is anhydrous and at 1040°C. (4) The petrographic data (Table 3 ) indicate the sizes of particles. (5) The drag coefficient of individual crystals is the same as for spherical particles of the same volume. (6) Vesicles observed in tephra were quenched at 10 atm. * V = [2g(Ap )r~/9~ ]. Stoke's Law. All velocities are given in meters per year and are calculated for the average size o f the phenocryst. Rates with asterisks are calculated for the largest phenocryst.

42

becomes vertically elongated (dike-like} and decreases if the b o d y of magma is sill-like. These estimates take no account of gradual degassing of ascending magma, of mixing between various levels of the magma b o d y during extrusion, of growth of the crystals during settling. Since the analyses of glasses included in olivines suggest that most olivines grew at an early stage, the shorter settling times may be more likely. This is so because (1) olivines may have grown to their petrographically observed sizes early, and (2) crystal fractionation could have occurred when the magma was H20-rich. We therefore consider it possible that the observed gradient in size and abundance arose in a few months if the magma b o d y is elongate vertically or in a few days if it is shortened vertically. In the case of the plagioclase phenocrysts there is a minimum in modal abundance for the intermediate pulses of the eruption. Our data are consistent with the following model of plagioclase accumulation. Plagioclases became enricl~ed upward in the top part of the magma body by flotation from the middle part of the magma body. The plagioclase phenocrysts are t o o dense (Fig. 14) to allow simple archimedian buoyancy. Flotation by bubbles is supported by the presence of zones of inclusions of gas and glass in most of the phenocrysts and is consistent with the sympathetic asymmetric development of inclusion-rich and inclusion-poor zones on the phenocrysts (Fig. 13 and Anderson, in preparation). Flotation of plagioclase above a certain level in the magma b o d y can be explained by the effect of pressure on effervescence and bubble size. The level of minimum concentration of plagioclase then corresponds approximately to a critical level above which the average vapor bubbles are large enough to cause average plagioclase crystals to float upward. Below the critical level, the plagioclase phenocrysts are neither depleted nor enriched by flotation, consistent with the intermediate modal abundance of plagioclase in the last (deepest) pulse of magma (Fig. 30).

Alternatives to crystal settling and flotation We have shown h o w the textural and compositional attributes of the Fuego ashes can be explained according to some simple mechanical models of crystal settling and flotation. Other explanations are possible, but require more unsupported assumptions. For example the observed and calculated modal variations could be explained by magma mixing: the earliest pulse of magma presumably from the upper part of the b o d y of magma could have incorporated and reacted with a residual magma drained from the wall rock area. However, we would, have to assume that the residual magma was appropriately rich in plagioclase constituents and poor in olivine constituents, because we have no evidence regarding the nature of such a magma. It is difficult to explain the variations by flowage differentiation alone: flowage differentiation will concentrate crystals away from walls, b u t it is n o t evident that it can concentrate one mineral and dilute another. However, Wright et al. (1976) have reported variable ratios of plagioclase and pyroxene in melt which flowed into drill casings in Hawaiian lava lakes. Finally, it is conceivable that the Fuego magma

43

preserved internal heterogeneities from its site of origin. In the last explanation we are swayed by the indication that the compositional variations are best described b y assemblages of minerals which include plagioclase. Plagioclase is n o t stable with olivine in the subcontinental mantle and presumably is not involved at the site of inception of the magma. Thus, we would have to assume that crystal/melt fractionations in the absence of plagioclase coincidentally are well approximated by plagioclase. In sum, our data are consistent with a variety of alternative explanations; however, we prefer the models of crystal settling and flotation because they are supported by textural observations and other physical data.

Depth range of crystallization of the Fuego magma We can estimate minimum depths for the crystallization of the Fuego magma in three ways. First the concentrations of H20 (maximum a b o u t 5%) in inclusions of glass in phenocrysts imply a pressure (about 2 kbar) adequate to dissolve the observed H20. Second, the a m o u n t of undercooling needed to account for the sodic rims on the plagioclase phenocrysts can be related to a given decline in pressure of H20. Finally, we may expect the pressure in the magma to exceed the hydrostatic pressure by a value approximately equal to the short-term strength of rock. The first case was discussed above. The other estimates are summarized below. Because most crystals have 4 to 12 zones in the outer sodic rim, and because a significant emission of vapor and ash occurred four days (8 tides) prior to the first blast on October 14, we infer that fractures which opened to the surface on or a b o u t October 10 resulted in a significant loss of gas from the Fuego magma in its conduit (see following discussion on possible tidal triggering of oscillatory zoning). Possibly an overpressure on the magma was relieved at that time. The short-term strength of crustal rock (0.2--1 kbar) probably was exceeded by the magma on October 10: Evidence for larger vent overpressures has been presented by Melson and Saenz (1973) for the 1968 explosion of Arenal. It seems possible that the top of the zone of phenocryst growth is controlled by the above pressure and lies a few kilometers beneath the summit of Fuego, near the base of the volcanic edifice. The marked increas~ in sodium in the outer rims of the phenocrysts corresponds to an undercooling of a b o u t 150°C in view of the experimental data of Lofgren (1974). According to the results of Yoder and Tilley (1962) and of Eggler (1972) a 150°C undercooling of plagioclase crystallization temperature could be achieved b y a decompression of 1--2 kbar. Fracturing of the grains subsequent to strong undercooling can be explained by stiffening of the magma with consolidation induced by the undercooling. Alternatively a difference in pressure between inclusions of glass and gas included in the crystals and the environment outside the crystals could fracture the grains if undercooling was caused by degassing. Undercooling caused by loss of gas is consistent with the absence of sodic rims on some inclusions of

44

glass in phenocrysts and with microprobe estimates of several percent of H20 in some of the inclusions. The various lines of evidence discussed above are consistent with a beginning of crystallization of the Fuego magma at a pressure of about 2 kbar and a quenching event at a pressure of about 1 kbar. The zone of phenocryst crystallization probably begins 5--10 km below the volcanic edifice and may terminate near the base of the volcano. Thus the range of vertical distance over which phenocryst crystallization occurred probably is a b o u t 5 km. It is not necessary that any extruded portion of the b o d y of magma occupied the entire 5-km range in depth at any one time. Possibly the magma b o d y moved as a unit from a depth of 8 km to a depth of 3 km without ever having a total vertical dimension greater than a few hundred meters.

Thickness and shape of the crystallizing portion of the body of magma It is possible to estimate the thickness of the solidifying part of the magma b o d y if a rate of crystallization can be estimated, and if the enthalpy budget of the magma is known well enough. Anderson (in preparation) has interpreted the oscillatory zoning of the plagioclase phenocrysts in terms of a tidal triggering model which yields an estimated rate of crystallization. The 1-t~m scale compositional zones of the oscillatory-zoned portions of the crystals are inferred to have formed in response to periodic vertical movements of magma in a conduit. Although some crystals differ in zoning patterns within individual lapilli, correlations of zoning patterns between some crystals suggest a general rather than local cause of the 1-t~m scale zones. Because there are a b o u t 28 1-pm scale zones between patchy-zoned regions on the plagioclase phenocrysts, Anderson (in preparation} inferred that the twice-daily tides triggered crystallization of the zones. The hypothesis of tidal triggering of crystal growth implies rates of growth of the oscillatoryzoned portions of the phenocrysts of a few microns per day. The patchy-zoned, inclusion-rich zones preferentially occur at periodic intervals and possibly reflect periods of growth influenced by vapor bubbles attached to the surfaces of the plagioclase crystals. Comparatively rapid decompression, possibly triggered by tidal minima during fortnightly maximum amplitude fluctuations (see Fig. 4) can account for effervescence. Rates of crystal growth implied for the patchy-zoned regions are about 100 pm/day. We approximate the average rate of crystallization of the first pulse of magma b y dividing the total crystal fraction (0.3) by a time of three months represented by approximately 200 compositional zones found on several large crystals. The actual rates of crystallization probably varied considerably. Possibly the average actual rates of crystallization were considerably greater than 10% per m o n t h because most of the mass of the crystals comprises individuals with fewer than 50 bands. These growth rates for Fuego plagioclase phenocrysts, estimated based on the tidal hypothesis, span the same ranges that were determined by different methods (Kirkpatrick, 1976; Lofgren, 1974) for other undercooled plagioclases. Although such data are very sparse, they

45

show that the " t i d a l " growth rate estimates are n o t unreasonable, whether or n o t one accepts the hypothesis that plagioclase zoning is tidally triggered. We also assume that the magma lost most of its enthalpy of crystallization by conduction through its walls. A b o u t one-fourth of the enthalpy of crystallization was used to increase the temperature of the magma from 1030 to 1050°C. Taking the heat capacity to be 0.3 cal g-~ °C-~, 6 cal/g are used to heat the magma, whereas about 24 cal/g are released by 30% solidification. The enthalpy required to exsolve the H20 vapor depends upon the details of the process but probably is small at pressures greater than a few hundred atmosphere (Burnham and Davis, 1974; see also Harris, 1977). We neglect the effect of the exsolution of the H20, but include that due to warming of the magma by using 3/4 of the usual latent enthalpy of fusion {that is: 0.75 X 80 cal/g = 60 cal/g). Conductive cooling through initially cold walls can be estimated according to equation 60 given by Jaeger (1968): H t = 2 T c ( K p c t / ~ ) ~' where H t is the total quantity of heat per unit area up to time t conducted across the contact between dike and c o u n t r y rock, Tc is the temperature of the contact, and K, p and c are the thermal conductivity, density, and specific heat of the country rock. The temperature of the contact is assumed to remain constant up to time t, which is a good approximation as long as part of the magma remains in a partly molten state. We consider that 3/4 of the heat of crystallization for the 30% of phenocrysts was lost in three months time as suggested by the tidal triggering hypothesis. The total a m o u n t of heat to be lost per cm 2 (Ht) is (}.75 X 0.3 X 80 cal/g X 2.3 g/cm 3 X a cm = (41 cal/cm 2) × a = Ht where a is half of the thickness of the dike. According to the above H t = 2 T c ( K p c t / ~ ) ~ = 2 X 500(0.005 X 2.6 X 0.3 X 60 X 60 X 24 X 90/n) ~ = 98250 cal/cm 2. Thus, a = (98250/41) cm = 24 m; this corresponds to a dike about 48 m wide. The estimated thickness is a m a x i m u m value because the mathematical model assumes instantaneous emplacement of magma into cold wall rock yielding a m a x i m u m thermal gradient at the initial boundary. Two factors will lead to a smaller thermal gradient for the situation in question. The first is prior warming of the wall rock in an active sub-volcanic system. The second is the position of the interface between fluid magma (with suspended crystals) and rigid magma (interlocking crystals with interstitial melt). The above interface determines what material is extruded and what is left behind. In the present case we are concerned with the lnagma which erupted. The thermal gradient at the interface between fluid magma and rigid magma is significantly less than that at the initial contact at all times subsequent to emplacement. According to an approximation which neglects the enthalpy of fusion (Jaeger, 1968, equation 9 and fig. la) the temperature gradient at the isotherm defined by T = 0.9 To (To = initial temperature of intrusion) is about half that of the gradient at the contact. Assuming that the isotherm T = 0.9 To can proxy for the boundary between fluid magma and rigid mush, the total heat release (/It) from and the thickness of the fluid portion of the dike estimated above are too high by at least a factor of two. In sum, we conclude that about

46

20 m is a probable upper limit to the average thickness of the fluid portion of the b o d y of magma. A dike-like b o d y of magma 20 m thick is consistent with a vent diameter of a b o u t 7 m inferred above from the volumetric rate of extrusion. For a dike to pass upward into a cylindrical vent, the b o t t o m , narrowest part of the vent must have a diameter equal to the width of the top of the dike. Thus, it is plausible to associate the diameter determined from extrusion rate with the width of the feeding dike. In addition, it is reasonable to expect the average width of the dike to be greater than the diameter of the vent which constricts the rate of extrusion. The similarity within a factor of 3 in the vent diameter and dike width lend support to the general validity of our approach. Further considerations of the shape of the magma b o d y depend upon the relations between its volume, thickness, and vertical dimension. We need to show that the volume of the extrusion (0.1 km s) is a good approximation to the volume of the b o d y of magma and to consider whether the extruded b o d y of magma occupied the entire range of depth indicated for crystallization or only a portion of that range. Except for solidified magma attached to the walls of the magma body, the extruded magma probably occupied a similar volume as a quasi-static b o d y before extrusion began. Alternatively we might consider that each pulse of magma occupied smaller regions for a few hours or days preceding each eruptive pulse. The following evidence favors the former alternative: (1) the parental melt of each pulse had a composition equal to that of the bulk of the eruption, b u t magma of each pulse differs in composition; (2) the number of oscillatory zones on plagioclase phenocrysts are mostly greater than 28, suggesting according to the tidal triggering hypothesis, a growth time longer than the interval of time between separate eruptive pulses; (3) the size and percentage of vesicles and the concentration of adsorbed sulfur suggest eruption from successively greater depths; (4) the time scale to generate the different compositions of the successive pulses from the same initial starting composition appears to be greater than a few days if the individual pulses derive from independent, small bodies of magma. In sum, the above features, together are best in accord with the existence prior to eruption of one differentiated b o d y of magma approximately equal in volume to that extruded during the nine-day eruption. The extruded magma probably occupied only a small part of the total depth range of solidification before the eruption begvn. If the magma occupied the entire depth range, then considerable differences in crystallinity would be expected. Although the variation in average crystallinity is only 13% and is irregular with time (Table 3), the variation for each pulse is 15--27%. The high K20 contents 9f inclusions of glass in some phenocrysts indicate that some extruded portions of the magma b o d y were at least 67% solid. Consequently the range in solidification of extruded products was from a b o u t 30% to 67%. A vapor-saturated high-alumina basalt magma beginning to crystallize isothermally at a b o u t 2 kbar would be 16% of the temperature interval between liquidus and solidus at 1 kbar and 33% of the distance at 500 bar,

47

according to Yoder and Tilley (1962). The relation between temperature below liquidus and crystal fraction is not known, but the data of Peck et al. (1966) indicate that it is approximately linear for basalts if the major minerals are crystallizing. Thus, we would expect a two-fold increase in crystal fraction for a magma body roofed at 500 bar and extending to 1000 bar. The range in phenocryst mode may partly reflect proximity to cold walls, but the similarity in quenching temperature (except for the H20 discussed above) suggests that the temperature range of the extruded parcels of magma was less than 50°C. It is likely that variations in crystal fraction depend more upon H:O than temperature, but our data are inadequate to be certain. In sum, we tentatively favor the interpretation that each pulse of magma excavated material from a depth range corresponding to a few hundred bars and that the total depth range of the magma body at the time of eruption was less than about 2 km. The estimated times for growth of plagioclase phenocrysts (a few months) and for sorting of olivine phenocrysts (a few months to a few days) overlap. Most of the olivines are comparatively early and the proportion of olivine to plagioclase forming probably declined with time. Therefore it is reasonable for the sorting time for olivine to be at least as great as the crystallization time for plagioclase. The implication of a month-long settling time for olivine is that the magma body was vertically elongate during most of the time that settling of olivine occurred. A dike-shaped conduit less than about 20 m wide and passing upward into a cylindrical vent 7 m or more in diameter is consistent with the available geological and geophysical evidence. The twin volcanoes Fuego and Acatenango and smaller vents are aligned transverse to the main WNW-trending chain of active volcanoes in central Guatemala (Stoiber and Cart, 1973). The alignment is consistent with an underlying fault or dike trending NNE. The epicenters of microearthquakes which occurred during March through September of 1975, near Fuego, fall on three intersecting lines, two of which trend north, the third trends east (Harlow, 1976). The lines are each about 10 km long and one of the intersections is less than 3 km south of the summit of Fuego. The earthquakes occurred five months after the end of the 1974 major eruption of Fuego, but intermittent minor eruptions continued beyond this period (Table 1). It is likely that the same feeding conduit was active during the recorded microearthquake activity as during the major eruption of 1974. The located microearthquakes are on the periphery of the net of seismic stations, and their distribution, although suggestive of intersecting faults or dikes, is also consistent with a single irregular region of microearthquake activity in view of the uncertainty of the locations of the epicenters (D.M. Harlow, personal communication, 1977). On Unalaska Island in the Aleutians, which is tectonically and structurally similar to Central America, the Miocene Unalaska Formation consists of interbedded siltstone and flows, sills, domes, and pillowed units of basalt, andesite and dacite which are intruded by two swarms of basaltic and andesitic dikes (Drewes et al., 1961). Locally, the dikes comprise 70--100% of the rock unit. Volcanic vents in batholiths are generally

48

identified with cylindrical pipes and stocks a few tens of meters to a few kilometers in diameter but dikes also occur. Commonly such vents are clustered near the margin of the pluton and are more mafic than the bulk of the pluton (see discussions in Fiske et al., 1963, p. 59; and Tabor and Crowder, 1969, pp. 12--23). Intrusive forms of basaltic rocks are predominantly sheet-like. In sum, a dike-like conduit for the 1974 eruption of Fuego volcano is suggested by our interpretation of the texture of the extruded magma and is consistent with geological and seismic data. We can n o w infer an image of the body of magma, with a volume of 0.1 km 3, a thickness of 0.02 km, a height of 2 km, an average breadth of 2.5 km. Our best estimate of the shape of the b o d y of magma is that of a coin-shaped, vertical lens. The batch of magma, almost completely liquid, may have entered a vertical crack at a depth of a b o u t 8 km. During the next three months the magma ascended the crack, effervesced most of its gas, crystallized, differentiated and became slightly warmer. Ascent probably was triggered by tidal stresses. Sometimes during ascent the b o d y of magma probably lost dynamical communication with its parent body. After about three months the magma attained a level in the crust where its b u o y a n t overpressure was adequate to reopen the vent of Fuego. Probably this level is near the base of the volcano. During the next ten days the magma body emptied, but a significant portion of the initial batch of magma probably remains below, frozen onto the walls of the crack.

Deep source o f Fuego magma Mantle peridotite is a compatible source material for Fuego's parent magma only if the magma has fractionated olivine and clinopyroxene before approaching the surface to give Cr and Ni abundances sufficiently low to match the Fuego concentrations. If this suggestion is correct, then the Fuego magma has undergone fractionation at a deeper level prior to its entrance into the shallower, dike-like conduit. Gill's (1974) model calculations seem to eliminate the possibility of Fuego's basalts being directly derived from eclogite being underthrust under Central America. High degrees of partial melting of eclogite would be required to account for the Ni, Co and Cr concentrations of Fuego relative to midocean ridge basalts, while low degrees of partial melting seem dictated by the elevated K, Sr, Rb, Ba and REE data. Chondrite-normalized REE plots for the 1974 Fuego ashes, the first published for Quaternary volcanic rocks of Central America, show slightly lightREE-enriched patterns, Eu/Eu* of nearly 1.0, and total REE abundances to 5--25 times chondrite values (Fig. 31). These are characteristics noted in mafic lavas from other orogenic volcanic regions such as the Lesser Antilles (Arculus, 1976}, New Britain (Arth, 1974), and the Aleutians {Kay, 1977}.

49

E~

:

f

field of

Q) Oct. 14 [] Oct. 17 i Oct. 19 Oct.23

REE ~ ch Q

~

x~~ \

~.

![ '.... ~ ....

I.O-

L'0

groundmassseparates

c.

~'~\

'..,,, "%,,

~.

s"

feldspars

A

I

T

bLu

Fig. 31. C h o n d r i t e - n o r m a ] i z e d ( L e e d y L - 6 ) R E E p l o t s f o r F u e g o ashes, g~oundmass samples

and feldspar of 1974.

Volcanic activity and great earthquakes The two most recent clusters of volcanic eruptions at Fuego (Fig.2) followed great shallow-thrust earthquakes, which occurred in 1853 and 1942 (Carl 1977). The historic earthquake record is inadequate to judge if such events preceded earlier eruptive clusters. The cause of clustering of eruptions could be related to large-scale deformation of the convergent plate boundary before, during and after great earthquakes. CONCLUSIONS More than 0.1 km 3 of homogeneous high-alumina basalt, possibly derived from olivine and pyroxene fractionation of mantle-derived basalt at deeper levels, began to crystallize in a body thicker than several tens of meters at a depth of at least 5 km beneath Fuego volcano. Magma worked its way upward, through a dike-like conduit during the few months prior to eruption. Tidal forces were probably important in the movement of magma within the conduit. Crystallization of plagioclase, olivine and magnetite began at 1030 -+ 50 °, as well as traces of clinopyroxene and pargasitic hornblende. Crystals about 2 mm in diameter underwent gravitational sorting within the magma body,

50

aided by the low viscosity of the melt, which contained about 3% H20. Aided by bubble-rafting, floating of plagioclase apparently occurred at the top of the magma body, which may have reached approximately the level of the base of Fuego's cone (3 km from the surface). The eruption consisted of four distincL pulses of ultravulcanian activity, each lasting 4--17 hours. The onset of each pulse coincided with luni-solar tidal minima. The magma was quenched at temperatures of 1050 +- 30°C and had different proportions of groundmass and four phenocryst phases in each successive pulse. These differences and bulk chemical data on the ashes are consistent with crystal fractionation and the inversion, by eruption, of the magma body. Gases released during the eruption were differentiated also, with higher relative S content during the most explosive phase of the eruption, when the most highly differentiated basalt was being brought to the surface. ACKNOWLEDGEMENTS

Financial support for this work came from the National Science Foundation, through grants DES74-19025, GA-26026, GA-38435, GA-35074, and EAR76-15016. The Phoenix Memorial Laboratory, through John Jones and Tom Meyers, provided the neutron activation analyses. The U.S. Geological Survey, through Bob Zielinski provided U and Th analyses. Tim Pearson, Karen Henry and Marsha Houghton did much of the laboratory work. David Harris helped with some of the calculations of magma dynamics and contributed valuable discussions and comments on the manuscript. Henry Pollack gave us a copy of his computer program for tidal calculations. Discussions with Norman K. Grant, Richard Stoiber, Thomas Crafford, Gregory A. Hahn and Donald C. Noble are gratefully acknowledged. Jean Polk and Julene Erickson helped edit the manuscript and typed the several drafts. Michael Carr, Stephen Self and Brian Baker read and commented on our lengthy early drafts. ANALYTICAL METHODS

Major elements in ashes were analyzed by atomic absorption after L i B O J H N O 3 fusionsolution (Medlin et al.,1969). Trace elements in ashes were determined by neutron activation (Cs, La, Ce, Sm, Eu, Tb, Yb, Lu, Th, Hf, Co, Cr, Sc), X-ray fluorescence (Rb, Sr, Zr, Mn), graphite furnace atomic absorption (Ba, Cu, Zn, Ni) and "slow neutron" techniques (U, Th). U.S.G.S. rock standards W-1 and BCR-1 were used as standards in all but the U and Th determinations (Flanagan, 1973). Ash leachates were obtained in solution of distilledwater (Rose et al.,1973, p. 350) and the elements were analyzed by specific ion electrode and graphite furnace atomic absorption methods. Sulfur was determined o n leached and unleached ash by an induction furnace/titration procedure. Precision of trace element determinations generally vary between 3 and 15% of the amount determined. Electron microprobe analyses of glass inclusions and minerals were performed at the University of Chicago following the methods outlined by Anderson (1974, pp. 1485--1586).

5]

Densities of plagioclase phenocrysts were determined with the aid of heavy liquids. W e hand-picked euhedral, millimeter-sized crystalsand ground off most of the adhering glassy groundmass by manipulating the crystals with tweezers on abrasive paper under a dissecting microscope. W e attempted to fillany voids which might be connected to the outside of the crystals by vacuum impregnating the crystals with a 2.6-g/cm 3 density mixture of diiodomethane and dimethyformamide. Subsequently we determined the densities of the crystals by their sink-floatbehavior in similar liquid mixtures calibrated by comparison with standard density glass tablets and by weighing 10-ml aliquots of liquids. The most successful method used a set of calibrated liquids of uniform depth kept in stoppered bottles which we briefly opened to insert a crystal with tweezers. The time for a crystal to float from the bottom or to sink from the top was determined for the bracketing liquids. The density of the crystal was calculated by interpolation assuming that the two liquids have identical viscosities.The formula is: Px

=

t~

t~-

t2

P~--

t2

- tl-

t~

P2

where t, and t 2 are the times' t o float (negative sign) or sink (positive sign) in liquids w i t h densities p~ and p 2. The estimated accuracy is 0.02 g/cm 3.

REFERENCES

Anderson, A.T., Jr., 1974. Chlorine, sulfur and water in magmas and oceans. Geol. Soc. Amer. Bull., 85 1485--1492. Anderson, A.T., Jr.,in preparation. Plagioclase phenocrysts from Fuego. Anderson, A.T. and Sans, J.R., 1975. Volcanic tempe~'ature and pressure inferred from inclusions in phenocrysts. Int. Geothermometry Geobarometry Syrup. Pennsylvania State Univ., Oct. 1975 (abstract). Arculus, R.J., 1976. Geology and geochemistry of the alkalibasalt-andesiteassociation of Grenada, Lesser Antilles island arc. Geol. Soc. Am. Bull., 87: 612--624. Arth, J.G., 1974. Rare earth elements in the basalt-andesite-dacite-rhyolitesuites of Talasea and Rabaul, N e w Britain, Geol. Soc. A m . Abstr. Progr., 6: 638. Arth, J.G., 1976. Behavior of trace elements during magmatic processes -- a summary of theoretical models and their applications. J. Res. U.S. Geol. Surv., 4: 41--47. Bonis, S.B., 1974. Fuego volcanic eruption. Smithsonian Inst. Center Short-Lived Phenomena, Event 134-74, Cards 1965, 1967, 1971. Bonis, S.B. and Salazar, O.D., 1974. The 1971 and 1973 eruption of Volcan Fuego and some socio-economic considerations for the volcanologist. Bull. Volcanol., 37: 394--400. Bottinga, Y. and Weill, D.F., 1970. Densities of liquid silicatesystems calculated from partial molar volumes of oxide components. Am. J. Sci., 269: 169--182. Buell, W.C. and Stoiber, R.E., 1976. Eruption of Volcan Fuego, October 14, 1974. Bull. Volcanol., 38: 861--870. Burnham, C.W. and Davis, N.F., ~974. The role of H~O in silicatemelts, If. Thermodynamic and phase relations in the system NaAISi30-H20 to 10 kilobars, 700°--1100°C. Am. J. Sci., 274: 902--940. Cart, M.J., 1977~ Volcanic activity and great earthquakes at convergent plate boundaries. Science, 197: 655--657. Crafford, T., 1977. Remote sensing of SO2 in the plume of Fuego volcano, Guatemala. Bull. Volcanol., 39: 536--556. Davies, D.K., Quearry, M.J. and Bonis, S., 1978. Glowing avalanches, Volcan de Fuego, 1974. Geol. Soc. Am. Bull., 89: 369--384. Davies, D.K., Combs, M.J. and Bonis, S.B., 1978. Airfallfrom the 1974 eruption of Volcan de Fuego, Guatemala. Geol. Soc. A m . Bull., 88 (in press).

52

Deger, E., 1932. Der Ausbruch des Fuego in Guatemala a m 21. Januar 1932 und die chemische Zusammensetzung seiner Auswurfmaterialien. Chem. Erde, 7 : 291--297. Drewes, H., Fraser, G.D., Snyder, G.L. and Barnett, H.F., Jr., 1961. Geology of Unalaska Island and adjacent insular shelf, Aleutian Islands, Alaska. U.S. Geol. Surv. Bull., 1028-3: 583--763. Eggler, D.H., 1972. Water-saturated and undersaturated melting relations in a Paricutin andesite and an estimate of water content in the natural magma. Contrib. Mineral. Petrol., 34: 261--271. Eggler, D.H. and Burnham, C.W., 1973. Crystallization and fractionation trends in the system andesite-H:O-CO2-O: at pressures to 10 kbar. Geol. Soc. Am. Bull., 84: 2517-2532. Ewart, A., Bryan, W.B. and Gill, J.B., 1973. Mineralogy and geochemistry of the younger volcanic islands of Tonga. J. Petrol., 14: 429--465. Fiske, R.S., Hopson, C.A. and Waters, A.C., 1963. Geology of Mount Rainier, National Park, Washington. U.S. Geol. Surv. Prof. Paper, 444: 1--93. Flanagan, F.J., 1973. 1972 values for international geochemical reference samples. Geochim. Cosmochim. Acta, 37: 1189--1200. Gill, J.B., 1974. Role of underthrust oceanic crust in the genesis of a Fijian calc-alkaline suite. Contrib. Mineral. Petrol., 42: 29--45. Harlow, D.M., 1976. Instrumentally recorded seismic activity prior to the main event (February 1976 earthquake). U.S. Geol. Surv. Prof. Paper, 1002: 12--16. Harris, D., 1977. Ascent and crystallization of albite and granitic melts saturated with H20. J. Geol., 85: 451--458. Haughton, D.R., Roeder, P.L. and Skinner, B.J., 1974. Solubility of sulfur in mafic magmas. Econ. Geol., 69: 451--467. Hoffman, D.J. and Rosen, J.M., 1976. Observations of stratospheric particulate matter following the eruption of Fuego. Trans. Am. Geophys. Union, 57: 346. Holloway, J.R., 1973. The system pargasite-H20-CO2: a model for melting of a hydrous mineral with a mixed-volatile fluid, I. Experimental results to 8 kbar. Geochim. Cosmochim. Acta, 37: 651--666. Jaeger, J.C., 1968. Cooling and solidification in igneous rocks. In: H.H. Hess (Editor), Basalts, Vol. 2. Interscience (Wiley), New York, N.Y., pp. 611--627. Kay, R.W., 1977. Geochemical constraints on the origin of Aleutian magmas. In: M. Talwani and W.C. Pitman III (editors), Island Arcs, Deep Sea Trenches and Back Arc Basins. Am. Geophys. Union, Maurice Ewing Ser., 1: 229--242. Kirkpatrick, R.J., 1976. Towards a kinetic model for the crystallization of magma bodies. J. Geophys. Res., 81: 2565--2571. Kuno, H., 1960. High-alumina basalt. J. Petrol., 1 : 121--145. Leeman, W.P., 1976. Petrogenesis of McKinney (Snake River) olivine tholeiite in light of rare earth and Cr/Ni distributions. Geol. Soc. Am. Bull., 87: 1582--1586. Longman, I.M., 1959. Formulas for computing the tidal accelerations due to the moon and the sun. J. Geophys. Res., 64: 2351--2355. Lofgren, G., 1974. An experimental study of plagioclase crystal morphology: isothermal crystallization. Am. J. Sci., 274: 243--273. Medlin, J.H., Suhr, N.M. and Bodkin, J.B., 1969. Atomic absorption analysis of silicates using LiBO~ fusion. At. Absorpt. Newslett., 8: 25--29. Meinel, A.B. and Meinel, M.P., 1975. Stratospheric dust-aerosol event of November 1974. Science, 188: 477--478. Melson, W.G. and Saenz, R., 1973. Volume, energy and cyclicity of eruptions of Arenal volcano, Costa Rica. Bull. Volcanol., 37 : 416--437. Molnar, P. and Sykes, L.R., 1969., Tectonics of the Caribbean and Middle America regions from focal mechanisms and seismicity. Geol. Soc. Am. Bull., 80: 1639--1684. Mooser, F., Meyer-Abich, H. and McBirney, A.R., 1958. Catalogue of the Active Volcanoes of the World, Part VI. International Association of Volcanology, Naples, 146 pp.

53

Peck, D.L., Wright, T.L. and Moore, J.G., 1966. Crystallization of Alae Lava Lake, Hawaii. Bull. Volcanol., 29: 629--656. Pollack, H.N., 1973. Longman tidal formulas: resolution of horizontal components. J. Geophys. Res., 78: 2598--2600. Pough, F.H. and Mulford, J.W., 1957. The Cranbrook Central American volcano expedition. Cranbrook Inst. Sci. Newslett., 27: 2. Robie, R.A. and Waldbaum, D.R., 1968. Thermodynamic properties of minerals and related substances at 298.15°K (25.0°C) and one atmosphere (1.013 bars) pressure and at higher temperatures. U.S. Geol. Surv. Bull., 1259: 1--256. Roeder, P.L., 1974. Activity of iron and olivine solubility in basaltic liquids. Earth Planet. Sci. Lett., 23: 397--410. Rose, W.I., Jr., 1977. Scavenging of volcanic aerosol by ash: atmospheric and volcanologic implications. Geology, 5: 621--624. Rose, W.I., Jr., Bonis, S., Stoiber, R.E., Keller, M. and Bickford, T., 1973. Studies of volcanic ash from two recent Central American eruptions. Bull. Volcanol., 37: 338--364. Rose, W.I., Jr., Grant, N.K., Hahn, G.A., Lange, I.M., Powell, J.L., Easter, J. and DeGraff, J.M., 1977. The evolution of Santa Maria volcano, Guatemala. J. Geol., 85: 63--87. Self, S., 1977. Characteristics and mechanisms of vulcanian eruptions. IAVCEI Syrup., Durham. Shaw, D.M., 1970. Trace element fractionation during anatexis. Geochim. Cosmochim. Acta, 34: 237--243. Shaw, H.R., 1972. Viscosities of magmatic silicate liquids: an empirical method of prediction. Am. J. Sci., 272: 870--893. Sparks, R.J.S. and Wilson, L., 1976. A model for the formation of ignimbrite by gravitational column collapse. J. Geol. Soc. London, 132: 441--452. Stoiber, R.E. and Cart, M.J., 1973. Quaternary volcanic and tectonic segmentation of Central America. Bull. Volcanol., 37: 304--325. Tabor, R.W. and Crowder, D.F., 1969. On batholiths and volcanoes -- intrusion and eruption of Late Genozoic magmas in the Glacier Peak area, North Cascades, Washington. U.S. Geol. Surv. Prof. Paper, 604. Taylor, P.S. and Stoiber, R.E., 1973. Soluble material on ash from active Central American volcanoes.-Geol. Soc. Am. Bull., 84: 1031--1042. Tilley, C.E., Yoder, H.S., Jr. and Schairer, J.F., 1967. Melting relations of volcanic rock series. Carnegie Inst. Washington Yearb., 65: 260--269. Volz, F.E., 1975. Volcanic twilights from the Fuego eruption. Science, 189: 48--50. Wickman, F.E., 1966. Repose period patterns of volcanoes. Ark. Mineral. Geol., 4: 291-367. Wilson, L. and Sparks, R.J.S., 1976. Plinean eruption columns and ignimbrite formation. Trans. Am. Geophys. Union, 57: 347. Wright, T.L. and Doherty, P.C., 1970. A linear programming and least squares computer method for solving petrologic mixing problems. Geol. Soc. Am. Bull., 81 1995--2008. Wright, T.L., Peck, D.L. and Shaw, H.R., 1976. Kilauea lava lakes: natural laboratories for study of cooling, crystallization and differentiation of basaltic magma. Am. Geophys. Union, Geophys. Monog~., 19: 375--390. Yoder, H.S. and Tilley, C.E., 1962. Origin of basaltic magmas: an experimental study of natural and synthetic rock systems. J. Petrol., 3: 342--532.