Global Volcanism Network Bulletin v. 19, no. 9, September 1994

Rabaul (Papua New Guinea)  Tavurvur remains active; details of
      September eruptions

New Britain, Papua New Guinea (4.27S, 152.20E)
All times are local (= GMT + 10 hours)

New eruptions began on 19 September 1994 (Bulletin v. 19, no. 8),
ending a repose period of ~51 years. Following the pattern of the
last two eruptive episodes (1878 and 1937-43), there were almost
simultaneous outbursts on opposite sides of the caldera as the
intra-caldera cones Tavurvur and Vulcan began erupting at 0605 and
0717, respectively. The eruption at Vulcan was the more powerful
and included a brief phase of strong Plinian activity soon after
its onset. Vulcan's eruption ended on 2 October. The eruption at
Tavurvur, after peaking during the first five days of activity,
exhibited a slow decline. However, moderate to weak activity
continued as of 28 October. By mid-late October, eight new
3-component seismic stations and two tilt stations had been
installed by volcanologists at the Rabaul Volcano Observatory (RVO)
with the assistance of USGS scientists. Many stations had been
damaged or destroyed by tsunami, vandalism, or heavy ashfall during
the eruption. The following report is from RVO.

Precursory activity. "A levelling survey along the usual route from
the Rabaul Town area to Matupit Island was completed on 15
September. Compared with the previous survey on 19 July (see
Bulletin v. 19, no. 7), the greatest change was uplift of ~25 mm at
the S extremity of the island. This rate of uplift is similar to
the long-term rate observed during 1973-83, prior to the 'Rabaul
Seismo-Deformational Crisis Period' of 1983-85.

"For most of the time in the preceeding few months, seismicity gave
little or no warning of the coming eruptions. The normal
(high-frequency) seismicity on the caldera ring-fault was at a low
level. Some low-frequency events were recorded, but their origin
and significance are not yet known.

"The eruptions were immediately preceded by 27 hours of vigorous
and fluctuating seismicity, which was initiated by two caldera
earthquakes (max ML 5.1) at 0251 on 18 September. These earthquakes
were located in the E part of the caldera seismic zone, near
Tavurvur, at a depth of 1.2 km. The earthquakes were felt very
strongly throughout the town and a small localized tsunami was
generated. Seismicity over the following four hours took place near
Vulcan and showed a general decline. Through this period, the
pattern of seismicity appeared to be similar to many previous
swarms of earthquakes on the caldera fault system. During the next
ten hours (0600-1600), earthquakes continued at a steady rate,
still concentrated near Vulcan. From about 1600 on 18 September,
seismicity increased and reached a peak at about 0200 on 19
September; at this time, earthquakes were felt every few minutes.
Seismicity then showed a slow decrease. Earthquake epicentres were
concentrated in the Vulcan area until about 0430, when the focus
shifted to Tavurvur.

"Soon after dawn on 19 September (0600), it was clear that an
eruption was imminent because offshore areas had emerged. The most
obvious uplift was at Vulcan, where a tide gauge was almost out of
the water, indicating an estimated uplift of 6 m. The W and S
coasts of Matupit Island had also been raised and the S shoreline
was shifted ~70 m S.

Evacuation. "In consideration of the increased seismicity after
about 1600 on 18 September, RVO recommended the declaration of a
Stage 2 alert (eruption expected within weeks to months) around
1800. This was subsequently issued at 1815. Throughout the late
afternoon a voluntary evacuation of the town had developed, but the
release of the Stage 2 alert accelerated the process. At midnight,
RVO advised the Provincial Disaster Committee that an eruption was
imminent. By this time, people had congregated in Queen Elizabeth
Park in the centre of Rabaul Town. Transport was mobilised, and
during the next few hours people were ferried from the town area to
beyond the caldera rim. RVO recommended a Stage 3 alert (eruption
expected within days to weeks) in the early hours of the 19th, but
the Disaster Committee refrained from a declaration because the
evacuation appeared to be proceeding well. It was feared that
announcement of a higher stage of alert might be
counter-productive. The evacuation went smoothly and by around 0700
on the 19th, the town and high-risk areas were virtually deserted.

Outbreak of eruptions. "An aerial inspection had been arranged for
early morning on the 19th. While waiting on the Rabaul airstrip, a
small white emission cloud was noticed above the W rim of
Tavurvur's summit crater at about 0603. Three minutes later, ash
was seen in the emissions which appeared to originate from the SW
part of Tavurvur's 1937 crater. The intensity of the emissions was
low as billowing, grey, cauliflower-shaped ash clouds rose slowly
and with little sound (figure 3). The ash clouds rose only a few
hundred metres and were driven towards Rabaul Town by moderate SE
winds. At about 0618, the ash plume had reached the S limits of the
town. The strength of the eruption remained low over the next hour
as darkness descended on Rabaul.

"The eruption of Vulcan commenced at 0717 on 19 September with
relatively small explosions on the N flank of the Vulcan 1937 cone.
However, activity intensified rapidly, and by 0737 low-density
pyroclastic flows were being generated and the eruption column was
rising rapidly. Run-out distances of ~2 km were common for these
early pyroclastic flows. At 0743, ballistic ejecta were seen
landing in the water up to 1 km from the E shore of Vulcan. At
about 0745 a phase of very strong activity commenced. Continuous
explosions generated a Plinian eruption column that attained a
height of about 20 km. The sounds of this activity were of dull
thudding, quite a contrast to the sharp, loud reports of electrical
discharges around the eruption column. By 0830, Rabaul Town and
surrounding areas were enveloped in darkness by the spreading ash
canopy. The phase of Plinian activity had ended by about 0830, but
strong ash emission continued.

"A number of tsunami were generated, probably by the Vulcan
activity. The largest of these rose ~5 m above high water. The SW
and W parts of Matupit Island were hit numerous times by tsunami,
washing inland as far as several hundred metres. Small boats were
carried inland ~60 m at the head of Rabaul Harbour.

Continuing eruptions. "The activity at Tavurvur increased through
the 19th and the eruption column was estimated to have reached a
maximum height of ~6 km. Only one vent was active. The eruption
column was very dense and the moderate SE winds drove the ash plume
directly over Rabaul. No pyroclastic flows were generated at
Tavurvur. Over the next few days activity at Tavurvur waned
slightly. The eruption column was usually ~1-2 km high. The dense
dark grey-brown ash clouds fed a plume that continued to blanket
Rabaul Town with fine ash.

"At Vulcan, at least four vents were active. The main vent was at
the point of the eruption outbreak. Another vent slightly to the N
was active briefly. A vent in the crater of the 1937 Vulcan cone
and one on its SW flank also were active. Two more phases of
Plinian activity took place at Vulcan in the evening of 19
September between about 1830 and 1930. The intensity of this
activity was considerably weaker than the first Plinian phase.
Pyroclastic flows were formed throughout the first few days of the
eruption. The largest of these extended ~3 km. Pumice from Vulcan
formed a large raft that covered most of Simpson Harbour.

Sequence of felt earthquakes and decline of eruption. "On 23
September, between about 1850 and 1900, there was a sequence of
strongly felt caldera earthquakes. The largest of these had an
estimated magnitude of 3.5. Most of the seismic stations had been
lost during the first day of the eruption, so it was not possible
to locate any of these earthquakes. However, most of them appeared
to originate from the SE part of the caldera. These earthquakes may
have been due to structural re-adjustment of the caldera to the
eruptive removal of significant quantities of magma. On the morning
of 24 September, a marked decline was evident in the activity at
Vulcan, and a lesser decline was seen at Tavurvur. This may have
been connected with the sequence of earthquakes the previous
evening. The eruption at Vulcan ended on 2 October, but Tavurvur
continued erupting, generating an eruption column 1-2 km high and
a plume ~20 km long.

Lava flow at Tavurvur. "A small lava flow was first noticed in the
summit crater of Tavurvur on 30 September. The aa lava was emerging
from a sub-terminal vent on the W flank of the growing ejecta cone.
The flow rate was extremely low as the lava slowly advanced towards
the W rim of the summit crater. On 5 October, a new lava lobe was
seen overriding the first lobe in the summit crater of Tavurvur.
This lava lobe also advanced very slowly and eventually reached the
nose of the first lobe. The length of these lobes was ~100 m. Lava
continued to be fed into these lobes after they had stopped
advancing, causing them to thicken. Eventually, on 8 October, a
breakout occurred on the W side of the original lobe. A more fluid
black lava emerged, ponding between the earlier lava flows and the
W crater rim. On 12 October, following a considerable growth of the
body of lava within the crater, lava began spilling over the crater
rim and descending Tavurvur's W flank. A second lava breakout from
the earlier bulky flows within the crater took place on 14 October.
This became the main feeder for the slowly advancing lava flow on
the W flank of the cone. It remained active until about 25 October.

Tephra from Vulcan and Tavurvur. "The tephra from Vulcan was pale
grey-brown pumice and ash, probably of dacitic composition. In
contrast, Tavurvur's tephra was dominated by very fine-grained ash.
Accretionary lapilli were abundant throughout both sequences and a
number of ash units were extremely hard, apparently having
self-cemented on deposition. The base of the Tavurvur sequence was
marked by a blue-grey very fine ash that appeared to be rich in
sulphides. This material probably originated as a hydrothermal clay
on the crater floor. Late in the Tavurvur sequence was a pumiceous
unit that may be sub-Plinian. During 8-18 October, strong
explosions ejected ballistic material as far as 1.5 km from
Tavurvur's summit. Large blocks (to ~1 m size) were found partially
buried in the road around the N and E foot of Tavurvur. These
ejecta included a mixture of dense glassy lava blocks, porphyritic
lava blocks, and pumiceous bombs.

Sulfur dioxide emissions. "SO2 emission rates from Tavurvur were
measured in the period from 29 September to 6 October by Stan
Williams (Arizona State University). Preliminary results indicated
a progressive decline from ~30,000 to ~3,000 tonnes/day.

Ground deformation. "Tilt measurements, which started at Matupit
Island on 24 September, indicated a large deflation (~930 frad) of
the central part of the caldera compared with pre-eruption values,
and a slowly reducing rate of deflation during the eruption. The
rate of deflation declined from ~10 to ~2 frad/day between 24
September and 25 October. Sea-shore levelling measurements, which
started in late September, indicated minor subsidence over most of
the caldera compared with pre-eruption levels. The greatest
subsidence was ~80 cm in the area of Rabaul Airport, between
Matupit Island and the town. About 3 m of uplift was recorded at
the E shore of Vulcan and slight uplift was recorded at the S end
of Matupit Island. Geodetic levelling from outside the caldera,
through Rabaul Town, and onto Matupit Island, confirmed these

Effects of the eruption. "The official death toll from the
eruptions and associated events was 5; four of which were due to
house roofs collapsing. One person was killed by lightning. Over
50,000 people have been displaced by the eruptions and were in care
centres in safe areas of the Gazelle Peninsula as of the end of

"The rapid accumulation of ash on Rabaul Town caused collapse of
some buildings within a few hours of the onset of the eruptions.
Ashfall from Tavurvur in the first few days of the eruption caused
widespread damage in Rabaul Town; virtually every building in the
S part of town collapsed. Serious structural damage was sustained
by most buildings in the ashfall zone within 8 km of Tavurvur. All
housing in the immediate area of Vulcan (to ~2 km) was destroyed
within ~1 hour of the start of the Vulcan eruption by a combination
of pyroclastic flows and heavy ashfall.

"Heavy rainfall during the first day and night of the eruption
exacerbated the effects of heavy ashfall. Mudflows and floods were
widespread in the Rabaul Town area, near Vulcan, and immediately
outside the Rabaul Caldera to the NW. The most serious floods were
NW of the caldera, where the heavy ashfall caused rapid runoff and
eventual deep erosion and migration of stream channels. The
obliteration of rainforest cover around Rabaul will present a
serious risk of flash floods and mudflows at times of heavy
rainfall. The wet season in Rabaul normally starts in early

Satellite imagery. "The westwards-spreading ash plume from Rabaul
was clearly visible from Earth-imaging satellites. A wide-angle
plume (90 deg) was seen on a series of Japanese Geostationary
Meteorological Satellite (GMS) images as a triangular area at 0903
of 19 September, spreading at different wind levels in a fan
extending from Rabaul. The N edge of the plume trended NW, and the
S edge to the SW, extending across the E Bismarck Sea and moving
down the N coast of New Britain.

"A similar spreading pattern was seen on images (IR channel 4) from
the NOAA-12 polar orbiting satellite (see Bulletin v. 19, no. 8).
The SE margin of the cloud at 1800 on 19 September was seen curving
S over the Solomon Sea and SE New Guinea, with the NE margin
extending past Manus Island. All parts of Papua New Guinea to the
W of these margins were covered by the eruption cloud. The strongly
sheared cloud seen on subsequent images was being driven S and then
E by high-level winds towards the Fiji region.

"AVHRR imagery from the Nimbus-7 satellite showed similar ash-cloud
dispersal patterns. However, computation of the temperature
differences recorded between AVHRR IR channels 4 and 5 at 1905 on
19 September and 0747 the next day yielded unexplained patterns in
which negative temperature differences (T4-T5), thought to be
indicative of ash-bearing clouds, were restricted to <1 deg of
latitude W of Rabaul (F. Prata, pers. comm. to RVO). In addition,
the SO2 signature seen on TOMS images at 1520 on the 20th and 1503
on the 21st (see Bulletin v. 19, no. 8) were restricted to the E
corner of the Bismarck Sea W of Rabaul, or over the general Rabaul
area. Both of these aspects of the satellite imagery require
further consideration and study."

Jim Lynch (NOAA Synoptic Analysis Branch) provided the following
satellite interpretation. NOAA and GMS satellite imagery clearly
depicted the volcanic plume during the first three days of the
eruption (19-22 September). The size and shape of the plume during
the first 18 hours is shown on figure 4. By correlating plume drift
with available wind data, the maximum height of the original plume
was estimated at 21-30 km altitude, well into the stratosphere. The
eruption maintained the plume to this altitude for ~12 hours before
tapering off to 12-18 km. After the first 56 hours of continuous
activity there was apparently a 6-hour respite, after which the
eruption resumed at a moderate intensity, generating a plume to <=6
km altitude. The stratospheric plume (>21 km) blew W and WNW toward
Borneo and Southeast Asia; however, the plume became too diffuse to
track beyond 1,300 km from the volcano. The upper tropospheric
plume (12-18 km) tracked SW, then S, and finally SE for ~1,000 km
around an upper-level ridge before it became too diffuse to track
with standard infrared imagery. The denser, more opaque portion of
the plume remained within ~400 km of the volcano. Analyses of
visible, infrared, and multispectral imagery from NOAA-12 and GMS
satellites definitively depicted an ash plume only within 1,000 km
of the volcano. Analysis of TOMS data revealed a relatively small
amount of SO2 (80 kilotons) close to the volcano (see Bulletin v.
19, no. 8). The fact that a dense plume of ash and aerosols did not
remain in the upper atmosphere suggests that the ash plume was
composed mostly of large particulates that fell out of the
atmosphere near and just downwind from the volcano.

Information Contacts: Chris McKee, with contributions from RVO
Staff and R.W. Johnson, Rabaul Volcano Observatory, P.O. Box 386,
Rabaul, Papua New Guinea; Jim Lynch, NOAA/NESDIS Synoptic Analysis
Branch, Room 401, 5200 Auth Road, Camp Springs, MD 20746 USA; Dan
Dzurisin and C. Dan Miller, USGS Cascade Volcano Observatory, 5400
MacArthur Blvd., Vancouver, WA 98661 USA (Email:

Figure 3. Photograph of Tavurvur taken from a helicopter at 0611 on
19 September 1994, just after the onset of activity. Note the 1878
cone (right foreground) being eaten away. View is approximately
towards the ENE. Courtesy of Rod Stewart, Rabaul Volcano

Figure 4. Areal extent and propagation of ash from Rabaul by
upper-level winds from 0830 on 19 September to 0230 on 20 September
1994. Isochrones are based on analysis of GMS infrared satellite
imagery. Courtesy of Jim Lynch, NOAA.