Detection of magma bodies beneath Krakatau volcano (Indonesia) from anomalous shear waves

Journal of Volcanology and Geothermal Research, 39 (1989) 335-348 335 Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

Detection of magma bodies beneath Krakatau volcano (Indonesia) from anomalous shear waves

HERY HARJONO 1'2, MICHEL DIAMENT 1, LASSAAD NOUAILP and JACQUES DUBOIS 1'* ' Laboratoire de G~ophysique (U.A. du CNRS 730), Bat. 509, Universit~ de Paris Sud, 91405 Orsay, France

2 Puslitbang, Geoteknologi - LIPI, Jl. Cisitu 21/154D, Bandung 40135, Indonesia

(Received November 21, 1988; revised and accepted May 31, 1989)

Abstract

Harjono, H., Diament, M., Nouaili, L. and Dubois, J., 1989. Detection of magma bodies beneath Krakatau volcano (Indonesia) from anomalous shear waves. J. Volcanol. Geotherm. Res., 39: 335-348.

The seismograms of 14 local earthquakes recorded by a temporary network in the Sunda Strait were used to con- struct an image of shear-wave attenuating bodies in the vicinity of the Krakatau volcanic complex. Seismograms for all ray paths going beneath the Krakatau complex show a diminution of the amplitude of S waves. These diminutions cannot be attributed to source mechanisms or station effects. Combination of all ray paths, although in limited num- ber, demonstrates that the attenuating body is located in two zones, which are probably disconnected, although the limited set of data available does not allow us to draw a definite conclusion. The upper zone is about 9 km deep and the lower one is at least 22 km deep. The upper zone contains probably numerous irregular pockets of magma. The deep zone appears to be a very wide body, related to the extensional nature of the Sunda Strait. The existence of a large, deep reservoir should be taken into account for the mitigation of future hazards.

Introduct ion

The deep and shallow structures beneath volcanoes can be inferred from an inversion of seismological data (e.g., Aki et al., 1977; Walck and Clayton, 1987; Walck, 1988; Sanders et al., 1988). If such inversion is not possible due e.g. to a limited number of stations, useful infor- mation can be obtained from a visual analysis of the seismic signature on analog seismograms (e.g., Dubois, 1971; Barazangi et al., 1974; McCaffrey, 1983 ). With such an approach, sev- eral authors have detected and located magma bodies or zones of molten rocks beneath volcan- oes since these features are characterized by a

*Present address: Institut de Physique du Globe de Paris, 4, Place Jussieu, 75252 Paris, France.

high attenuation of shear waves (e.g., Matu- moto, 1971; Ryall and Ryall, 1981; Sanders and Ryall, 1983; Sanders, 1984; Sanders et al., 1988). In this paper, we examine local earthquake sig- natures in order to constrain a zone of anoma- lous propagation of shear waves beneath or near the Krakatau complex.

The Krakatau complex is situated in the Sunda Strait between Java and Sumatra is- lands, Indonesia. It consists of four islands: Sertung, Panjang, Rakata and Anak Krakatau (Figs. 1 and 2). Anak Krakatau (child of Krak- atau in Indonesian) is an active volcanic cone born in August 1927, 44 years after the famous 1883 explosion. The last eruption occurred in March 1988 when a small quantity of andesitic lava was extruded (Suparto and A. Sudradjat, pers. commun., 1988). This volcano is charac-

0377-0273/89/$03.50 © 1989 Elsevier Science Publishers B.V.

336 H. HARJONO ET AL

104°E [

S U M A T R A

106°E

5oS

I N D I A N "~,' ~ -

O C E A N

J AVA

7os

Fig. 1. Map of the studied area. The hatched band is the volcanic line of Panaitan (P), Krakatau complex (KC), Sebesi (Si), Sebuku (Su), Rajabasa (R) and Sukadana (Sa). Bathymetry contour in km {after Larue, 1983 ).

terized by a bimodal basalt-dacite composition, different from the common pattern of volcan- oes in Indonesia which generally show a uni- form composition (Hutchison, 1982). The eruptions cycles of Krakatau always begin with a basalt and evolve into dacite composition (Van Bemmelen, 1942; Camus et al., 1987). This is very probably related to the tectonic set- ting of the Sunda Strait. Indeed, according to the results of recent geological and geophysical investigations, the Sunda Strait is an exten- sional area (Lame, 1983; Huchon and Le Pi- chon, 1984; Renard et al., 1985; Harjono, 1988; Harjono et al., 1988; Lassal et al., 1989; Lame, in prep.; Larue and Foucher, in prep.). There- fore, the Krakatau complex appears to be a very unique volcano of the Indonesian arc and this explains the vast amount of investigations car- ried out there (e.g., Decker and Hadikusumo, 1961; Zen, 1969; Yokoyama, 1981; Camus and Vincent, 1981; Nishimura et al., 1986; Camus et al., 1987) including the Krakatau Program of the Indonesian-French cooperation in Oceanology.

The Krakatau complex lies on a N-S vol- canic lineament going from Sukadana to Pan- aitan island (Fig. 1) which is outlined by seis- micity, as revealed by worldwide seismological data (Harjono et al., 1988). The Anak Kraka- tau is presently the only active volcano of that lineament. Recently, a detailed seismological survey was carried out in and around the Sunda Strait in order to better constrain the seismic- ity of the area. One of the main results of our survey was to reveal a zone of seismicity con- centrated beneath the Krakatau complex (Fig. 2; Harjono, 1988). These events could be very precisely located, i.e. with an average accuracy of about 2 km, and a map of epicenters (Fig. 2b) shows that they are mainly located beneath the part of the new caldera (1883) which overlaps the older one (Camus and Vincent, 1983). It must be noted that this seismicity was recorded during a short period (3 months) when Anak Krakatau was not very active. A view in vertical cross section (Fig. 2c) reveals that the seism- icity is concentrated in a narrow vertical zone. Another important result of the detailed seis- mological survey was to show that these events have most probably a tectonic origin (Harjono, 1988), i.e. they could not be associated with tremors. This conclusion was not only based on the visual examination of seismograms which always show high frequencies, since, as has been pointed out by Minikami (1974), volcanic shocks might have the appearance of tectonic earthquakes. Our conclusion was based mainly on the fact that stations around the strait sys- tematically recorded up and down first motions which allowed the computations of focal mech- anisms for all events. An inversion of these mechanisms (Harjono, 1988) gave a stress ten- sor which could fit the mechanisms of the earthquakes located outside of the Krakatau complex. Furthermore, the direction of exten- sion (N130 ° ) given by the stress tensor was consistent with the one deduced from an anal- ysis of seismic reflection data in the Sunda Strait (Lassal et al., 1989). In fact the Kraka- tau complex is located at the intersection of the

DETECTION OF MAGMA BODIES BENEATH KRAKATAU VOLCANO 337

104°E 106°E

0

\ BTG

o o e o e • <

o I ~ o • o

c

~°oo o 0 ° o < 3 0 0 o o

0

o o

J

105.3 °

b

6.0 °

6.2 °

lOkm

• =

A K L I

5°S

;.DMe e e(-~ TBG ]

' uJ~* / 4~C.L

o • ~ . 7os

• o

,

105.5 °

® o

W k m

10

20

,, ~e.~ #. ~>~.~ ..... •

A

j * , . \

1Okra

volcanic lineament (Fig. 1) and of a fault ori- ented N230 ° which approximately links the Krakatau complex with the cluster of seismic- ity located in the graben (Fig. 2a).

The focal mechanisms (Fig. 2c) are domi- nated by compression for events with depths less than 4 km and by extension for larger depths. Such variation of focal mechanisms with depth beneath volcanic features have been re- ported elsewhere (e.g., the Long Valley Cald- era, California; Vetter and Ryall, 1983). The authors suggested that these observations could be explained by the presence of a magma body at shallow depth, the existence of which was proven by other geophysical studies. Although such an interpretation is tempting for this seis- micity, we believe that the variation of focal mechanisms with depth observed here is most probably related to structural effects (Harjono, 1988).

In fact the location of the magma body be- neath the Krakatau complex was never dis- cussed except that according to petrological and volcanological analysis, the magma reservoir for recent eruptions should be at shallow depth, less than 15 km beneath the Krakatau complex (De N~ve, 1983; Camus et al., 1987). It also has been suggested that the reservoir of the 1883 erup- tion should have a large volume, at least 10 km 3 or even more than 20 km 3 (Newhall, 1983; Ca- mus et al., 1987 and pers. commun. ).

We present now our method of detection and

Fig. 2a. Local seismicity recorded by our temporary net- work during Summer 1984. Black and open dots corre- spond to events located between 0-20 km and 20-50 km, respectively. Triangles are the stations of the network. KLI and KRK stations belong to the Indonesian Institute of Meteorology and Geophysics Agency and to the Volcanol- ogical Survey of Indonesia. The hatched area is presented with more detail in Figs. 2b, 7 and 9. Fig. 2b. Epicenter distribution close to the Krakatau com- plex. The pre 1883 caldera (thick) and the new caldera (thin) are drawn after Camus and Vincent, 1983. Fig. 2c. Cross section through Krakatau complex and some selected focal mechanisms (projected in lower hemi- sphere ), after Harjono (1988), note the variation of focal mechanisms with depth.

338 H. HARJONO ET AL.

the results concerning the location of a magma body beneath and/or close to the Krakatau complex from the analysis of records of our mi- croearthquake survey of 1984.

Method and data

Our method consists of detecting a possible attenuation of S waves for some ray paths trav- eling beneath the Krakatau complex from earthquakes located all around it, by examining the amplitude of the S waves on the seismo- grams. In fact this method is very similar to the one used for example by Sanders (1984) in or- der to locate magma bodies beneath the Long Valley caldera. Since we used analog stations in our network, we only performed a visual obser- vation of the attenuation of S-wave amplitude.

One can notice, however, that the effect of the magma chamber on the wave propagation is also to filter high frequencies of both P and S waves. Nevertheless, even though we system- atically inspected the spectral contents of the seismograms we did not rely on this observa- tion since ( 1 ) due to the field condition the fil- tering was not the same on all stations and (2) we believed that a visual estimate of the spec- tral contents is quite subjective except in very homogenous field conditions.

The data used in this study were selected from our survey in summer 1984 (Fig. 2). The net- work consisted of 11 MEQ-800 Sprengnether portable seismographs and one station of WWSSN (KLI) (Fig. 2a). KLI and KRK be- long to the Indonesian Meteorology and Geo- physics Agency (BMG) and to the Krakatau observatory of Voleanological Survey of Indo- nesia (VSI) respectively. All stations consisted of a single vertical seismometer; the events were recorded in analog form either on smoked or ink papers. The hypocenter locations were deter- mined using the Hypoinverse routine of Klein (1978) and assuming a crustal velocity struc- ture as shown in Table 1. A detailed description of the network and of the data analysis can be found in Harjono (1988).

T A B L E 1

Crustal velocity model used for the locations of earth- quakes in the Sunda Strait. This model was based on both refraction data (Fatwan, 1983) and the inversion of data of the microearthquake survey. After Harjono et al. ( 1988 ).

Depth to top of layer v. (km) (km s)

0 3.1 4 5.5 9 6.8

22 7.8

It must be noted that all stations did not give identical information. For example most of the data from KRK could be used only for the de- termination of first arrivals. This was due to some technical conditions required for the pur- pose of monitoring the volcano. So KRK sta- tion, which was of major importance in our net- work for the precise location of earthquakes (Harjono, 1988), could generally not be used for the present study. Also due to field condi- tions, the gain of some stations was limited and small earthquakes were not systematically re- corded at all stations with a signature clear enough to analyse the respective amplitude of P and S. It should be also noted that some strong events could not be used since we could not dif- ferenciate the S from P coda. We therefore re- jected poorly recorded events that were either too weak or too strong.

To select earthquakes for the present study we need to take events generating rays passing beneath the Krakatau complex in all azimuths. Due to the geometry of our network (absence of stations on the west side) and to the limited seismicity to the east of the Sunda Strait (see Fig. 2a), it was not possible to obtain paths coming from the northeast or the east. This limits the precision of the location of a possible attenuating body.

The error in the location of the magma bod- ies is mainly due to the uncertainty of the hy- pocenter locations. Several tests have been per- formed (Harjono, 1988) in order to obtain the

DETECTION OF MAGMA BODIES BENEATH KRAKATAU VOLCANO 339

T A B L E 2

Loca t ion of e a r t h q u a k e s u sed in t h e p r e s e n t s tudy . E R H a n d E R Z are t he e s t i m a t e d ho r i zon ta l a n d ver t ica l er ror respect ively.

M L is t he local m a g n i t u d e .

E v e n t Da t e T i m e Lat . Lon . D e p t h E R H E R Z M L 1984 ( U T ) ( ° S ) ( ° E ) ( k m ) ( k m ) ( k m )

1 Ju ly 23 0345 6 ° 06.84 ' 105 ° 26.13 ' 7.00 0.63 0.48 3.0 2 Ju ly 23 0354 6 ° 06.36 ' 105 ° 26.54' 7.70 0.59 0.42 3.4 3 Ju ly 24 1735 6 ° 07.31' 105 ° 26.10' 7.20 0.52 0.54 3.5 4 Ju ly 30 1233 6 ° 04.31 ' 105 ° 29.94' 20.30 0.57 0.57 2.9 5 Aug. 01 1030 6°08.01 ' 105°25.20 ' 9.40 0.65 1.39 3.4 6 Aug. 01 1038 6 °08.75 ' 105 ° 24.98' 6.80 0.57 0.84 3.5 7 Aug. 03 0934 6 ° 08.33' 105 ° 24.46' 8.20 0.75 0.96 2.8 8 Aug. 06 0102 7 °00.35 ' 105 ° 29.61' 19.00 3.07 1.28 3.5 9 Aug. 06 0115 6°07.68 ' 105°24.52 ' 3.40 0.73 0.39 3.4

10 Aug. 07 1817 6°39.33 ' 104°53.60 ' 69.00 3.11 6.69 3.6

11 Aug. 07 2143 6 ° 15.13' 104°57.04 ' 17.00 0.78 6.39 2.7 12 Aug. 13 1733 6°04.99 ' 105°24.70 ' 4.90 1.79 1.13 3.4 13 Aug. 13 1755 6°06.24 ' 105°25.87 ' 3.50 1.85 0.42 2.5 14 Aug. 27 0422 6 ° 32.07' 105 ° 01.49 ' 4.80 4.04 1.19 3.3

best accuracy possible on the hypocenter deter- minations. For example, the differences in the locations of earthquake hypocenters deter- mined using different crustal models were shown to be relatively small (less than 3 kin). For the present study we kept only earthquakes which were very well recorded and with hypo- centers as accurate as possible, that is with a lateral error (ERH) and a vertical error (ERZ) of the hypocenters less than 2 km for the earth- quakes beneath Krakatau and less than 7 km for the earthquakes located outside of the Krakatau complex. Note that in addition the location errors of the station positions were of a few tens of meters (Harjono, 1988). So, fi- nally we kept 14 representative events, 5 lo- cated outside the Krakatau complex and 9 in- side it. They are listed in Table 2.

In spite of our careful selection it must be noted that the events inside the Krakatau com- plex were generally poorly recorded in the most remote stations (TRM, BTG and KLI).

As shown in Table 2, all events except num- ber 10 are within the crust. According to our velocity model (Table 1 ) they generally give re- fracted waves except events 5 and 10. So the information on the downward extension will be

mostly limited to the depth of 22 km corre- sponding to the Moho. It should be noted also that almost all the ray paths from events inside the caldera, except the ray paths going to BTG and KLI, were refracted on the 9 km deep interface.

We classified the S waves on each seismo- gram according to their signatures. Our classi- fication is the following: 0 - for the normal propagation with very pronounced S waves; 1 - for events with slightly attenuated S waves; and 2 - for events with a small or even without S signature. Since this classification based on a visual observation can be quite subjective, we individually classified all the seismograms and cross-checked our classification. We finally kept only the unambiguous paths.

There is a possibility that the diminution of the amplitude of shear waves is due to the po- sition of the recording station relative to the SV radiation pattern (Sanders, 1984). So we sys- tematically determined focal mechanisms to check the position of the stations relative to the P nodal planes, since the amplitudes of S waves are minimum in a direction just between the two P nodal planes (e.g., Bullen and Bolt, 1987). To check whether the attenuation is due to an el-

340 H. H A R J O N O E T AL,

fect of the station, we also compared seismo- grams for many other earthquakes, having var- ious azimuths with respect to the station and not included in the present study.

Resul ts and d i scuss ion

Figure 3 shows the surface projection of the ray paths for the earthquakes located outside of the Krakatau complex. We observe clearly that some seismograms recorded by TBG, GRB, and CIL are characterized by small amplitudes of S or even absence (classified as 1 or 2) while the other stations received normal shear waves of quality 0. This figure also confirms that the at- tenuation is not a station effect, since clear shear waves are observed at TBG, GRB, or CIL for earthquakes giving rise to ray paths outside the Krakatau complex. Furthermore, as previ- ously mentioned in order to check a possible source effect we determined focal mechanisms. For example, Figure 4 presents a series of rec- ords for event number 10 and its focal mecha- nism. From this figure we can see that at GRB the amplitude of the shear waves is very small. If we then check its focal mechanism, we find that the position of GRB is close to the nodal plane: so normally the record at GRB should present a large amplitude of the shear wave. Thus, the at tenuat ion of S wave evidenced at GRB station cannot be attr ibuted to the source but must be related to an at tenuat ing body lo- cated between the source and the station.

The combination of all observations (Fig. 3) reveals that the at tenuat ion is most probably related to the Krakatau complex since only rays passing close to it show attenuation. This figure also gives some hints that the at tenuat ing body can extend rather far to the east or that another one should exist since the ray from event 8 to GRB shows attenuation. The distance between that ray and the Krakatau complex cannot be attributed to a mislocation of event 8. Note that in addition a shift of the epicenter of event 8 to the west will also shift the projection of the ray to TBG!

104oE 106OE

I ' AKLI , \ 5°S

BTG ', ~ T B G

~, I / GRB

U J K ~ 7°S

I t

Fig. 3. Surface projection of ray paths for the four earth- quakes outside the Krakatau complex. The striped and dot- ted lines correspond to the strongest attenuation {classi- fied as 2 ), dashed lines to the slight attenuation (classified as 1) and full lines to classification 0 or normal propagation.

In fact the at tenuating body revealed by this path (8-GRB) can be at any location between the source and the station. For example Figure 3 cannot exclude a magma body beneath the Krakatau complex and another one some place south of GRB, that is in the northern part of the volcanic line (Fig. 1). Although there is presently no surface manifestation of active volcanism close to GRB, some very recent ac- tivity (0.8 Ma) has been reported there (Nish- imura et al., 1986). Note that the existence of a magma body without surface activity has been reported elsewhere as for example in the Mount Katmai area, Alaska (Matumoto, 1971). Nevertheless, if such body exists, it must have a limited extension since neither gravity (La- me, in prep. ), magnetic (Kimpouni and Tuch- olka, 1987; pers. comm.) or seismological (see Fig. 2 ) data show an anomaly there.

If we exclude the existence of the magma body nearby GRB, that path (8-GRB) gives addi- tional information, since it is a refracted wave. The magma body so revealed must be about 22

DETECTION OF MAGMA BODIES BENEATH KRAKATAU VOLCANO 341

, - - - . _ .

_~ .__ _. - - =

/ \ . / " "._

p~ n_J

D

- .,, CIL

\ .

LO~

DGRB t

DTRM / /

" . . D D DBT G / ~.. --. TBG D s -1

"----_ _

i

180 ~ T w

2~ ~* ~ '~ ~! AkIF

65" 52" ~ PLRNE D D I L R T ,

Fig. 4a. Focal m e c h a n i s m of e v e n t 10. G R B a n d C I L s t a t i o n s are close to P nodal p l anes a n d shou ld no rma l ly record S waves .

km deep, that is near the bottom of our crustal model as previously mentioned.

Close to the Krakatau complex, Figure 3 shows that several ray paths are not attenuated (for example l l -KRK, 14-KRK or 4-TRM) in apparent contradiction with the attenuation observed on the ray paths 8-TBG and 10-GRB. To solve that apparent contradiction one must examine the third dimension. Ray paths arriv- ing at KRK are superficial in the vicinity of the Krakatau complex while according to our crus- tal model the ray 8-TBG must be at 22 km depth and the one 10-GRB at 32 + 6 km. So, this in- dicates that if we assume that the attenuation of ray paths 8-TBG and 10-GRB is due to a body located just beneath the Krakatau complex, i.e. where they intersect, that body would be at a depth at least 22 km and extend at least down to 32+6 km. Furthermore, ray paths from events 11 and 14 to KRK show that it must not

extend too far in the southwest while the rays from event 4 show that it must not extend in the northwest at a depth of about 20 km. Before discussing the extent of this body further, we now discuss the data from earthquakes located beneath the Krakatau complex. Figure 5 shows the ray paths for all the earthquakes inside the Krakatau complex; we also added event no. 4 which is very close to it.

All these events were well located with a high accuracy on their depth due to their location in the center of the network and to the existence of the KRK station nearby.

For all these events, records at CIL clearly show diminished S amplitudes. Again, we ruled out the possibility of attenuation due to a source or a station effect. Figure 6 displays a set of seis- mograms for one event of this set and its focal mechanism.

Figure 5 shows that paths toward GRB or

342 H. HARJONO ET AL.

Fig. 4b. Some seismograms of event 10 in various stations. Note the large a t tenuat ion of S waves at GRB as compared to other stations.

UJK, which are normal for events with depth less than 9 km (see Table 2), show an atten- uation for event 4 which is much deeper. For other stations such as CBL or PAS, the rays are either normal or attenuated, depending on the earthquakes. This observation must be ex- plained by the geometry of the attenuating body.

This pattern confirms also our assumption

that no reservoir exists south of GRB at a shal- low depth. These results compared with the previous ones deduced from path 8-GRB indi- cate that the magma body extends at least at 20 km east from Anak Krakatau (i.e. the distance between Anak Krakatau and the path 8-GRB) or parallel to the volcanic lineament.

Figure 7 shows all ray paths close to the

DETECTION OF MAGMA BODIES BENEATH KRAKATAU VOLCANO 343

105.4

\/'~ / ' (~ / ,

/,

0 /

13 i

Fig. 5. Surface projection of ray paths for earthquakes located beneath the Krakatau complex and for event no. 4 (the numbers correspond to Table 2 ).

E",/E!.iT "_ ,~

/ !

z !

• h " v m _ j I i

!

. . --" ._S

/ / -

. / m

I .............................. ~ IB ,, J s o a ~ -

- \

BTG

TRM

TBG

--._.. , . .

Ill

.... BaIL

[ ] •

P /

/ /

j j ~

! 80

/-

Lan

m T ! m T m o,RIKE ~ir-

186" 58" >( PLRNE Y PL~kE

• C05!F'.

EIILRT.

Fig. 6a. Focal mechanism for event 12.

344 H. HARJONO ET AL

TRM

SDM

C I L

Fig. 6b. Some seismograms of event 12. Note the attenuation of S waves at CIL.

DETECTION OF MAGMA BODIES BENEATH KRAKATAU VOLCANO 345

105.2 ° 105.4 ~ 105.6 ° W AK

b •

~ ~ . . ~ . :-;_.;..:.. . . . . . . . .

30

,iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii!iiiiiiii i

io Km

Fig. 8b. Cross section B.

Fig. 7. Ray paths close to the Krakatau complex for all events. Same notation as on Fig. 3. Lines A, B and C cor- respond to cross sections shown on Fig. 8.

NW A K

krn r Q

10 )- , : .

3O

1OKra L ~ j

.__ .L__ (,_ .... ~..".:Z:~,,,,

i.-i:.:.:i'."

Fig. 8. Cross sections beneath the Krakatau complex. Dots correspond to earthquakes. The rays in the vertical plane are shown with the same notation as in Figs. 3 and 7. Squares show intersection of a ray with the cross section, black squares are non-attenuated rays, half-black ones show rays of quality 1 and open squares show most attenuated rays. The dotted areas show the most probable attenuating zones. (a) . Cross section A.

Krakatau complex together with the projection of 3 vertical cross sections. The first (A) is par- allel to the direction KRK-SDM/BTG and is shown on Figure 8A. The B cross section is be-

I I . Km

C

10

20

30

0

AK

1OKra

gE

o~ mmm •

Fig. 8c. Cross section C. See Fig. 7 for location.

tween event 11 and CIL (Fig. 8B) and the C one is parallel to the direction GRB-UJK (Fig. 8C).

On these vertical cross sections we show rays which are very close or within the vertical plane and we indicate with squares the intersection of the plane with other rays. The black squares correspond to the rays of quality 0, the half black squares to those of quality 1 and the open squares to quality 2. Dots correspond to the earthquakes.

The three cross sections clearly display two zones of attenuation. A first one, extending be-

346 H. HARJONO ET AL.

~ 0 5 . 2 o 1 0 5 . 4 ° 1 0 5 . 6 °

r ~ i T r

Sebesi

" Pall an

[

6"3° I ~0 km

Fig. 9. 9- and 22-km deep l imit of the magma body. The thick line is the 22-km limit and is not constrained in the northeast or east of the Krakatau complex (see text for de- tails ). Striped areas correspond to the upper attenuating zone.

low our crustal model, is located at least at a depth of 22 km and extends at least down to 32 km beneath the Krakatau complex.

The second zone is much more superficial, at a depth of about 9 km and seems to have a much smaller extension than the deeper one. Fur- thermore, the W-E cross section (Fig. 8B) shows that the upper zone is most probably di- vided in several areas where varying numbers of passing rays show signs of attenuation. This area seems to consist of separated pockets of magma and fractures directly beneath and south of Anak Krakatau. The western limit of these features is well constrained by some direct ray paths (e.g., l l -KRK and 5-SDM) and is lo- cated 10 km from Anak Krakatau.

No precise indication can be given for the eastern limit of the superficial body due to lack of data.

The lack of data does not allow us to draw a definite conclusion concerning the relationship between the upper and lower attenuating zones. In fact, from our data one can neither prove that the two parts of the attenuating body are ver-

tically separated nor that their are connected. However, the presence of an epicenter at 20 km seems to indicate a high degree of rigidity and thus the separation of the magma-bearing area into two parts.

All results are summarized on Figure 9 where some limits of the attenuating body are given. The full line corresponds to the limit of the at- tenuating at the Moho, i.e. at a depth of 22 km and is constrained by the ray paths from 4 to TRM, 8 to KLI, 11 to GRB and 14 to CIL. We do not exclude an extension of the magma body at a depth larger than 22 km outside this line in any direction. The hatched area corresponds to the zone where the magma body is closer to the surface, at a depth of about 9 km. As with the deeper zone, the upper one also could extend to the east or northeast.

Comparison of our results with petrological or volcanological information and other seis- mological data suggests the following remarks.

We confirm that a superficial magma body exists in the vicinity of the Krakatau complex. According to already mentioned studies (de N~ve, 1983; Camus et al., 1987) this magma body is most probably the reservoir of the re- cent eruption of Anak Krakatau.

Our result also implies the existence of a very large, deeper reservoir. This large reservoir seems to be much larger than the one invoked for the 1883 eruption. The existence of a very large reservoir can be related to the extensional nature of the area. For example a very large magma reservoir has been reported in the Rio Grande Rift area (e.g., Brocher, 1981 ). There- fore it seems logical to assume that the upper and lower zones of attenuation correspond to two distinct reservoirs, probably connected by fractures. It is tempting to relate the existence of a large deep reservoir to the cataclysmic eruptions, although the superficial reservoir shown here seems large enough for the differ- entiation of the large volume (10-15 km ~) of dacite of the 1883 eruption (G. Camus, pers. commun., 1988). In fact, the existence of a large deep reservoir, providing an important source

DETECTION OF MAGMA BODIES BENEATH KRAKATAU VOLCANO 347

for basic magmas, should be taken into account for the mitigation of future hazards.

Note also that Figure 9 seems to confirm the southwestward migration of the activity o f Krakatau proposed by Sudradjat (1982), since in the vicinity of Krakatau the center of our 22- km limit is shifted southwest of the new caldera which is itself shifted with respect to the pre- 1883 caldera.

The limited amount of data used here do not permit us to discuss the relationship between the vertical seismicity observed beneath Krak- atau complex and the various zones of atten- uation. Clearly, further investigation of the structure beneath the Krakatau complex is necessary in order to better constrain the ge- ometry of the attenuating bodies. For example, a more dense seismological network including stations on the islands of the Krakatau com- plex and Ocean Bottom Seismographs (not available in our 1984 survey) would certainly provide new information.

A c k n o w l e d g e m e n t s

We regret that Jean Bloyet who greatly helped us during our field work left us too early to see the final version of this paper. This work was done in the framework of the French-Indone- sian Cooperation on Oceanology and we are in- debted to Prof. M.T. Zen who encouraged us to study this unique volcano. We thank the Vol- canological Survey of Indonesia and the Indo- nesian Meteorology and Geophysical Agency (BMG) for providing data from KRK and KLI. We also thank M. Larue and M. Regnier from ORSTOM and the technician group of Puslit- bang. Geoteknologi-LIPI for their assistance during the survey. The manuscript benefitted from the comments of P. Camus and of discus- sion with P. Tucholka. This research was sup- ported by A.T.P. "Gdologie et Gdophysique des Oceans" of C.N.R.S. by ORSTOM, IFREMER, LIPI and BPP, Teknologi.

References

Aki, K., Christofferson, A. and Husebye, E., 1977. Deter- mination of the three-dimensional seismic structure of the lithosphere. J. Geophys. Res., 82: 277-296.

Barazangi, M. Isacks, B., Dubois, J. and Pascal, G., 1974. Seismic wave attenuation in the upper mantle beneath the southwest Pacific. Tectonophysics, 24: 1-12.

Bemmelen, R.W. van, 1941. Krakatau. Bull. East Indian Volcanol. Surv., 18: 53-60.

Bullen, K.E. and Bolt, B.A., 1985. An Introduction to the Theory of Seismology, 4th ed. Cambridge University Press, New York, NY, 499 pp.

Brocher, T.M., 1981. Geometry and physical properties of the Socorro, New Mexico, Magma bodies. J. Geophys. Res., 86 (B10): 9420-9432.

Camus, G. and Vincent, P.M., 1981. Krakatau volcano. SEAN Bull., 10: 4.

Camus, G. and Vincent, P.M., 1983. Un si~cle pour com- prendre l'~ruption du Krakatau. La Recherche, 14: 1452- 1457.

Camus, G., Gourgaud, A. and Vincent, P.M., 1987. Petro- logic evolution of Krakatau (Indonesia): Implication for a future activity. J. Volcanol. Geotherm. Res., 33: 299- 316.

Decker, R.W. and Hadikusumo, D., 1961. Result of the 1960 expedition to Krakatau. J. Geophys. Res., 19: 167-173.

De N~ve, G.A., 1983a. Geovolcanology of the Krakatau group in the Sunda Strait region: Review of a hundred years development {1883-1983). Symposium on 100th year Development of Krakatau and its surrounding, Jakarta, August 23-27, 1983.

De N~ve, G.A., 1983b. Krakatau's earliest known activity: was it prehistoric? Berita Geologi, Bandung, 15: 39-44,

Dubois, J., 1971. Propagation of P-waves and Rayleigh waves in Melanesia: Structural implications. J. Geo- phys. Res., 78: 3431-3436.

Fatwan, S., 1983. Pendugaan bawah tanah permukaan den- gan metoda seismik bias di daerah Selat Sunda. S1 The- sis, Institut Teknologi Bandung.

Harjono, H., 1988. G~odynamique du d~troit de la Sonde (Indon~sie): apports des donn~es de microsismicit~ et implications volcanologiques. Thesis, Universit~ Paris- Sud, 262 pp.

Harjono, H., Diament, M., Dubois, J. and Larue, M., 1988. Sismicit~ du d~troit de la Sonde (Indon~sie): Presen- tation des r~sultats d'un r~seau local. C.R. Acad. Sci. Paris, 307: 565-571.

Huchon, P. and Le Pichon, X., 1984. Sunda Strait and Central Sumatra Fault. Geology, 12: 668-672.

Hutchison, C.S., 1982. Indonesia. In: R.S. Thorpe (Edi- tor), Andesites. Orogenic Andesites and Related Rocks. John Wiley & Son, Chichester, New York, Brisbane, Toronto, Singapore, pp. 207-224.

Klein, F.W., 1978. Hypocenter Location Program: Hypoin- verse. U.S. Geol. Surv., Open-file Rep. 78-694, 113 pp.

348 H. HARJONO ET AL.

Larue, M., 1983. CORINDON IX Cruise Report. Larue, M., 1988. Gravimetrie du ddtroit de la Sonde (In-

don~sie). (In prep.) Larue, M. and Foucher, J.P., 1988. M~sures et mod~lisa-

tion du flux de chaleur dans un graben du ddtroit de la Sonde (Indon~sie). (In prep. )

Lassal, O., Huchon, P. and Harjono, H., 1989. Extensian crustale dans le dgtroit de la Sonde (Indon~sie). Don- n~es de la sismique r~flexion (Campagne Krakatau), C.R. Acad. Sci. Paris, 309, II: 205-212.

McCaffrey, R., 1983. Seismic wave propagation beneath the Molucca Sea arc-arc collision zone, Indonesia. Tecton- ophysics, 96: 45-57.

Matumoto, T., 1971. Seismic body waves observed in the vicinity of Mount Katmai, Alaska, and evidence for the existence of molten chamber. Geol. Soc. Am. Bull., 82: 2905-2920.

Minikami, T., 1974. Seismology of volcanoes in Japan. In: L. Civetta, P. Gasparini, G. Luongo and A. Rapolla (Editors), Physical Volcanology. Elsevier, Amsterdam, pp. 1-27.

Newhall, C.G., Decker, R.W., Sudradjat, A., Tilling, R.I. and Peteson, D.W., 1983. A comparison of eruption and magma reservoirs of Krakatau, Mount St. Helens, and Galunggung. Symposium on 100th Year Development of Krakatau and its Surrounding, Jakarta 23-27 Au- gust, 1983.

Nishimura, S., Nishida, J., Yokoyama, T. and Hehuwat, F., 1986. Neo-tectonic of the Strait of Sunda, Indonesia. J. Southeast Asian Earth Sci., 1: 81-91.

Renard, V., Chamot-Rooke, N., Deplus, C., Diament, M., Dubois, J., Ganie, B., Harjono, H., Huchon, P., Jongsma, D., Kridoharto, P., Larue, M., Rigolot, P., Soekartadi- redja, H., Suhardi, I., Sumarno, S., Voisset, M., Widja- janegara, M. and Wirasantosa, S., 1985. Krakatau'85 Cruise Report.

Ryall, A. and Ryall, F., 1981. Attenuation of P and S waves in a magma chambre in Long Valley caldera, California. Geophys. Res. Lett., 8: 557-560.

Sanders, C.O., 1984. Location and configuration of magma bodies beneath Long Valley, California determined from anomalous earthquake signal. J. Geophys. Res., 89 (B10): 8267-8302.

Sanders, C.O. and Ryall, F., 1983. Geometry of magma bod- ies beneath Long Valley caldera, California determined from anomalous earthquake signals. Geophys. Res. Lett., 10(8): 690-692.

Sanders, C., Ho-Liu, P., Rinn, D. and Kanamori, H., 1988. Anomalous shear wave attenuation in the shallow crust beneath the Coso volcanic region, California. J. Geo- phys. Res., 93 (B4): 3321-3328.

Sudradjat, A., 1982. The morphological development of Krakatau volcano, Sunda Straits, Indonesia. Geologi Indonesia, 9 (1): 1-11.

Vetter, U.R. and Ryall, A.S., 1983. Systematic change of focal mechanism with depth in the Western Great Basin. J. Geophys. Res., 88 (B10): 8237-8250.

Walck, M.C. and Clayton, R.W., 1987. P wave velocity vari- ations in the Coso Region, California, derived from local earthquake travel times. J. Geophys. Res., 92 (B 1 ): 393- 405.

Walck, M.C., 1988. Three-dimensional Vp/Vs variations for the Coso region, California. J. Geophys. Res., 93 (B3): 2047-2052.

Yokoyama, I., 1981. A geophysical interpretation of the 1883 Krakatau eruption. J. Volcanol. Geotherm. Res., 9 (4): 359-378.

Zen, M.T., 1981. The state of Anak Krakatau in September 1968. Bull. Natl. Inst. Geol. Mining, Bandung, 2 ( 1 ): 15- 23.