RELACS - Imaging the Rabaul volcano

Rabaul is a town on the island of New Britain in Papua New Guinea, which lies is inside the caldera of one of the major volcanoes of the New Britain island arc. In late 1994 the Rabaul volcano started erupting causing much damage and disruption in Rabaul town and the surrounding area. Volcanic activity continues, but is now waning. The activity of the Rabaul volcano has been monitored for the past 45 years by the Rabaul Volcano Observatory, and the object of the RELACS project is to improve the monitoring capability in the future.


RELACS stands for Rabaul Earthquake Location and Caldera Structure. The collaborative project involves the seismology group at RSES, the Australian Geological Survey Organisation, Hokkaido University, the University of Wisconsin and the Rabaul Volcano Observatory. The project is funded by AusAID and JICA, the foreign aid organisations of Australia and Japan, respectively. The aim of the project is to aid the Rabaul Volcano Observatory with the monitoring of the Rabaul volcano by conducting an experiment which will improve the observatory's ability to locate earthquakes associated with the volcano and to interpret their potential relationship to volcanic activity.

1997 experiment

The experiment was successfully conducted in 1997. A combined instrument pool of 75 instruments from all five participating institutions was deployed during the latter half of the year in the vicinity of Rabaul. The instruments recorded local, regional and distant earthquakes as well as artificial pressure-wave sources (see Figure 9). The timing of the waves emanating from the different sources as they arrive at the various seismometers throughout the Rabaul area contains information about the way in which seismic wave speed varies in the volcano's subsurface. These data will be used to build a tomographic image of the interior of the Rabaul volcano based on wave speed. Using that image we can improve the location of earthquakes which occur within the volcano and relate their location to features of the tomographic image so as to better understand them. Figure 10 shows an example of an earthquake recording at Rabaul.

An important aspect of the RELACS project is to re-analyse recordings of earthquakes just before the onset of the 1994 eruption in light of an improved velocity model and improved methods of earthquake location in order to take full advantage of the past recordings of the Rabaul Volcano Observatory.

In addition to providing the Rabaul Volcano Observatory with practical information to aid them with their social task of warning the public around the volcano about impending eruptions, the RELACS project is of much basic research interest. A detailed tomographic image can reveal the geometry of the volcano's magma storage and improve our understanding of the physical processes of magma emplacement in the crust and of eruption. The distribution of seismicity around the volcano can likewise tell us much about the processes of a volcano.

Figure 1: a) The configuration of the RELACS deployment. Solid circles represent one component stations, triangles are three-component stations. b) A subset of the events recorderd by the array (solid circles) together with the artificial sources set off for profiling and shot into the RELACS array (stars).

1998 analysis

Figure 2: Example of an earthquake recording at Rabaul using the new solid-state recorder system developed at RSES. This event occurred some 120 km away from Rabaul and was recorded using a short-period, vertical seismometer. Both a P wave and an S wave are clearly visible.

During 1998 we have developed software for the error checking of data, archiving, and general manipulation of the data from the RSES solid-state recorders and from the broad-band stations. An archive has been built in standardised form containing about one thousand of the most significant of the 2300 events detected during the RELACS survey. The data archive includes the thirty explosions carried out during the survey which provide sources with high precision of timing and location. The interpretation of the seismic data is now in its preliminary stages.

An example of data recorded during RELACS is presented in fig 3 below. Record sections along an east-west striking profile which crosses the Rabaul caldera are displayed. The array of recording stations is approximately the same in all cases with the shot point moving along profile. This transect was designed to map shallow structure around the volcanic caldera (top 5 km). Clear indications of heterogeneity are evident from the temporal relationships in the sections and the structure of the wavefield itself. For example, wavespeeds are slow on the east side of the caldera relative to the west as seen from later arriving energy at a comparable distance.

Figure 3: Record sections along an east-west traverse through the Rabaul caldera. The array of stations is kept approximately fixed in space while the shotpoint (black dot) moves from west at top to east at bottom. Distances are true distances. The location of the caldera is indicated with the letter C. Times have been corrected for station elevation, shot depth and water depth beneath each shot. Travel-time is shown using a reduction velocity of 5.5 km/s.

1999 analysis

Figure 4 Images from the 3-D model of the velocity structure in the Rabaul region. The top panel shows a depth slice at 4 km depth dominated by the central low velocity anomaly beneath Blanche Bay. The lower panel shows a cross section to 10 km depth along the profile indicated in the map view above.

The RELACS project was completed during the year. Among the project's results are detailed two-dimensional velocity models along two profiles crossing the northwest Gazelle Peninsula, one crossing the Rabaul caldera complex, the other crossing the Tavui caldera to the northwest of Rabaul. The project also resulted in a detailed three-dimensional velocity model for the top 15 km and an area within 25 km from Rabaul. The distribution of seismicity within the Rabaul volcano was also reexamined using the three-dimensional velocity model.

A clear low-velocity anomaly was found to lie at 3-5 km depth directly beneath the Blanche Bay caldera, the youngest and seismically active collapse structure of the volcanic complex. This low-velocity zone is about 5 km across and its margins underly a ring of volcanic vents which have been active after the formation of the collapse structure. The data cannot resolve the velocity within this low-velocity zone clearly, but do constrain it to lie in the range from 3.0 to 4.5 km/s. This represents a major anomaly compared to adjacent velocities of 5.5 - 6 km/s at the same depths. This anomaly lies beneath the seismicity of the Rabaul volcano and is interpreted to be a shallow magma reservoir.

The surface velocities above the magma reservoir, within the caldera structure are found to be very low, between 1.5 and 2 km/s. These low velocities may be explained by the fracturing of shallow rocks caused by the collapse event and by soft sedimentary infill.

Tavui caldera is not underlain by a shallow crustal magma reservoir. At depths comparable to the magma reservoir beneath Blanche Bay the Tavui caldera is underlain by fast velocities. These may be the cumulates of a crystallized magma reservoir which then would have to be interpreted as being very old (more than one hundred thousand years). The implications for risk associated with this volcanic structure are clear, however, a deeper (> 10 km) crustal magma reservoir beneath Tavui cannot be ruled out.

The seismicity of the 25 years leading up to Rabaul's last eruption in 1994 is arranged in a roughly oval pattern underlying the Blanche Bay collapse structure. This seismicity had been determined to lie in the top four km, but considerable uncertainty had to be assigned to that estimate due to the uncertain velocity structure around the volcano. The RELACS project has significantly constrained this velocity structure and thus tightened the constraint on the depth extent of seismicity. The result is that the seismicity extends to 3-4 km depth and that while the oval pattern stays largely intact it has slightly shrunk and shifted. These changes of pattern of seismicity are verified by corrected mislocation of explosions conducted by the RELACS project.

Figure 4 highlights the main results from the RELACS project on the velocity structure around the Rabaul volcano. The top panel shows a depth slice at 4 km depth dominated by the central low velocity anomaly beneath Blanche Bay. The lower panel shows a cross section to 10 km depth along the profile indicated in the map view of the panel above. Again the model is dominated by the confined low- velocity anomaly beneath Blanche Bay and other structures associated with the volcano. The shallow fast velocities to the northwest of the caldera (beneath point A) are probably due to shallow lying carbonates (limestone outcrops in the area). The shallow fast velocities around the western rim of the caldera may be due to a high volume fraction of intrusives.

Updated:  13 September 2010/Responsible Officer:  Director, RSES /Page Contact:  Web Admin