Another phase of strong Strombolian eruptions took place at Tavurvur in April. There were both gas-rich explosions and a small but complex lava flow on the upper S flank of the cone. The lead-up to these eruptions was characterized by episodes of relatively gas-rich, low-pressure vapor emissions punctuated by occasional moderate to large explosions. The explosions seemed to increase in strength from late March to early April. One of the largest explosions took place on 28 March and its cloud rose ~4.3 km above the vent. One on 5 April rose ~5.3 km.

In early- to mid-April Tavurvur displayed a variety of notable eruptive phenomena: odd-colored gases, unusual lava-venting styles, a small pyroclastic flow, and the ejection of very large bombs. The water-tube tiltmeter at Sulphur Creek (3.5 km from Tavurvur) showed inflation at a rate of 0.4 µrad/day in the direction of Tavurvur until about 8 April when a reversal occurred.

COSPEC measurements indicated low outputs of SO₂, 100-200 metric tons/day (t/d) following the higher output of the mid-March eruption (BGVN 22:03). During 29 March-2 April, briefly elevated SO₂ outputs (400-600 t/d) were measured. There was a decline in SO₂ flux in early April, reaching a low of 220-370 t/d when measurements ceased on 8 April. Unfortunately, the COSPEC was removed from Rabaul at a critical time, denying the opportunity to monitor SO₂ during the subsequent eruptive cycle.

Small swarms of low-frequency earthquakes preceded several of the 9-12 April explosions. An explosion on 10 April ejected a plume that may have risen as high as 7 km, and generated small pyroclastic flows on Tavurvur's upper flanks. Heavy ballistic fallout on the N flank also triggered a small landslide.

Eruption of 12 April. At 0829 on 12 April a Strombolian eruption was accompanied by a strong explosion. This explosion generated a very dense and dark ash cloud. Part of the ash cloud may have been projected at a low angle eastward, or alternately, the E part of the eruption column may have collapsed. In any case, the end result was the generation of a small pyroclastic flow on Tavurvur's E flank. A very dense, dark ash fall-out cloud hung over the area NE of Tavurvur until about 1000 on 12 April. About this time, the ash content of emissions dropped markedly.

During this episode, from about 0830 the RSAM level rose very sharply, peaking at 1000 at about 850 RSAM units. The seismicity remained at this level until about 1100 when it started to decline.

At 1030, nearby observers saw ejecta that looked incandescent and quite fluid. Accompanying explosions repeated at intervals of 1 second or less. The maximum height reached by ballistic ejecta was a little over 1 km. Some of the ejecta were tens of meters across. Orange gas clouds of unknown composition accompanied some explosions.

From about 1130 it seemed as though lava collecting in the crater had risen, although no lava flow had commenced. By about 1200 strong fumes escaped from Tavurvur's upper S flank, which began to bulge outward. Incandescence was seen occasionally on the flank as chunks of the bulging part of the cone tumbled downslope. At 1215 a breach in the S rim of the crater developed in an area that had been undermined by the subsurface passage of lava. Black lava without visible incandescence was moving down the S flank at 1218. By 1226, a broad flow-front (several hundred meters wide) reached a mound about half-way down the S flank. Soon after, by 1229, a W lobe developed.

At 1236 on 12 April, following 1-2 minutes of silence, a lava fountain rose rapidly several tens of meters above the crater. It then expanded at its top, forming a large ball of lava ~50 m across. The lava ball then exploded, showering the S flank with huge clots of brightly incandescent lava. A similar event followed at 1245 when a large mound of lava (50 m across) rose rapidly in the crater before exploding.

During the early afternoon, the multi-lobed front of the slowly descending lava flow headed mostly S and W. No incandescence could then be seen in the frontal parts of the flow, but weak incandescence was visible in the 20- 30 m channel through the breach in the S rim of the summit crater. Observers first saw an E lobe at about 1420, but it had certainly developed prior to that time.

An aviation report around 1335 on 12 April (issued by Darwin VAAC) described an ash cloud to 4 km that was "pale gray and low in ash content." The report also noted that ash clouds blew NE and satellite imagery lacked clear evidence of an ash cloud. At 2123 the Rabaul Volcano Observatory issued a warning to aviators that "a large ash column was ejected forcefully up to several kilometers."

During the early afternoon there was a gradual decline in seismicity, although some individual explosions accompanied very strong signals. By 1800 the RSAM level had dropped to about 640 RSAM units.

The night-time activity on 12 April remained spectacular even though the strength of the eruption had markedly declined. A particularly large bubble of lava formed in the crater at 1816 and shattered, causing a brilliant shower of incandescent ejecta, accompanied by a loud cracking detonation. Many similar explosions occurred through the night but none was as large or as bright as the one at 1816.

The outline of the new lava flow was difficult to see when viewed from ~1 km away on the night of 12 April. Only the part of the flow front on the S flank remained incandescent, and it lacked detectable movement.

Although there were occasional, powerful explosions through the night of 12 April, an overall decline in eruptive and seismic activity continued. At about 1000 on 13 April, seismicity reached a low of about 150 RSAM units. Strong explosions continued through the 13th, but they took place less frequently than they had previously. A key feature of the explosions was the very low ash content in emission clouds. Fluid or plastic incandescent ballistic lava fragments were the principal solid products from the explosions. The gas component was a white to pale gray fume emitted in moderate to large quantities.

The vigor of Tavurvur's outbursts gradually declined through the remainder of the month. Crater glow was still present at month's end.

Post-eruptive activity and observations. Strong, caldera-wide harmonic tremor was a significant feature of the post-Strombolian activity. Commonly, a pulsing and roaring noise accompanied the tremor; the noise suggested that the tremor may have been linked to a near-surface degassing process. On many occasions a large explosion took place followed by a prolonged period of degassing and tremor. There were instances, however, when tremor would commence without an attendant explosion and audible sounds. Tremor mostly ceased by 24 April, but isolated periods still occurred at the month's end.

In contrast to earlier Strombolian eruptions at Tavurvur, tiltmeter readings at Sulphur Creek lacked offset, an observation that appeared consistent with this eruption's relatively small volume. By late April there was evidence from the Sulphur Creek tiltmeter that inflation of the Tavurvur area had resumed, presumably in conjunction with a possible forthcoming Strombolian eruption.

Inspections on the E, N, and S sides of Tavurvur revealed abundant "cow-dung" lava bombs. The largest one seen had landed ~1 km from the vent and measured ~8-9 m across. The lava had obviously been very fluid, and after impact the bomb displayed delicate surface features. No fine ash-fall deposit was found around Tavurvur's base. This was consistent with the absence of ash in the emissions from about 1000 on 12 April, yet the early part of the eruption generated much ash. A search for deposits from the pyroclastic flows found only a light dusting of very fine gray ash in a gully at Tavurvur's NE foot.

The lava flow produced in this eruption differed slightly from the earlier flows seen in October 1996, and January and March 1997. The April flow was ~3-4 m thick (compared to earlier flows that were 4-5 m thick). The most recent flows only covered a relatively small area (10⁵ m², compared to 10⁶ m² for each of the earlier flows).

Inspection on 23 April revealed that the E lobe of the flow actually overrode the E margin of the initial flow. Thus, part of the April flow was compound and in this respect unlike the simple flows of the earlier eruptions.

Analysis of one sample from the 12 April eruption indicated a significant change in lava chemistry. The April sample was an andesite having ~59% SiO2 and 4.4% MgO. In comparison, the 1994-95 samples had the compositions of high-silica andesites to low-silica dacites. As with earlier samples, the April one showed evidence of mixing of the caldera's resident dacite with a basalt.

Geological Summary. The low-lying Rabaul caldera on the tip of the Gazelle Peninsula at the NE end of New Britain forms a broad sheltered harbor utilized by what was the island's largest city prior to a major eruption in 1994. The outer flanks of the asymmetrical shield volcano are formed by thick pyroclastic-flow deposits. The 8 x 14 km caldera is widely breached on the east, where its floor is flooded by Blanche Bay and was formed about 1,400 years ago. An earlier caldera-forming eruption about 7,100 years ago is thought to have originated from Tavui caldera, offshore to the north. Three small stratovolcanoes lie outside the N and NE caldera rims. Post-caldera eruptions built basaltic-to-dacitic pyroclastic cones on the caldera floor near the NE and W caldera walls. Several of these, including Vulcan cone, which was formed during a large eruption in 1878, have produced major explosive activity during historical time. A powerful explosive eruption in 1994 occurred simultaneously from Vulcan and Tavurvur volcanoes and forced the temporary abandonment of Rabaul city.

Information Contacts: B. Talai, H. Patia, D. Lolok, P. de Saint Ours, and C. McKee, Rabaul Volcano Observatory (RVO), P.O. Box 386, Rabaul, Papua New Guinea; Bureau of Meteorology, Northern Territory Regional Office, P.O. Box 735, Darwin, NT 0801 Australia.