For a Special Issue of Pure and Applied Geophysics on the Aitape Tsunami

The Aitape 1998 tsunami: Reconstructing the event from interviews and field mapping.

Abstract

On the evening of 17 July 1998, on the Aitape coast of Papua New Guinea, a strongly felt earthquake was followed some 10-25 minutes later by a destructive tsunami. The tsunami comprised three waves, each estimated to be about 4 m high. The second of the three waves rose to a height of 10-15 m above sea level after it had crossed the shoreline and caused most damage. Maximum wave heights and greatest damage were recorded along a 14-km sector of coast centered on Sissano Lagoon. In this sector the wave fronts moved from east to west along the coast; all structures were destroyed, and 20-40 percent of the population was killed. Partial destruction extended 23 km to the southeast and 8 km to the northwest, and effects of the tsunami were felt as far as 250 km to the west-northwest, beyond the international border (Joku, this volume). More than 1600 people are known to have died, with some estimates as high as 2200; 1000 were seriously injured, and 10,000 survivors were displaced. Information presented in this paper was gathered in the course of a public awareness campaign in 1998-2002, from interviews with eye-witnesses and from mapping of damage and inundation. These sources provided new information on the height, shape and timing of the waves; on the possible escape of petroleum and other gases from beneath the sea floor before and during the tsunami; on unusual sound effects that preceded the waves, and lighting effects that followed; on possible deep circulation (to 250 m) of sea water in the waves. We also recorded 50-70 cm of subsidence of the coastal sand barrier in the sector of most destruction and noted the resilience and potential protective capacity of certain species of trees. Eye-witness accounts indicate that the tsunami reached the shore at between 09:00 and 09:08 UT, which is earlier than is permitted by published models of the timing and location of the source of the tsunami.

1. Introduction

In the early evening of Friday 17 July 1998, following soon after a strongly felt earthquake, three great waves struck the coast west of Aitape, Sandaun Province, Papua New Guinea (PNG; Fig. 1). The waves came almost without warning and there was little people could do to save themselves, other than struggle to survive in the turbulent water. Entire villages were destroyed almost without trace, and survivors were carried inland into lagoons, mangrove and sago swamp, and forest (Fig. 2). People died by drowning or by impact with hard objects. In all, more than 1600 people lost their lives, with some estimates of fatalities as high as 2200, another 1000 were seriously injured, and 10,000 were displaced to new villages.

		version
1. Locality map and map of the Aitape coast. Bathymetry is from multibeam survey (Matsumoto and Tappin, this volume: solid lines ) and from spot depths on a navigation chart supplementary to chart Aus 389: dashed lines. The sunken reefs are at depths of 460 m and 82 m, respectively. Sissano Lagoon is the body of water inland from Warapu and Arop. The Sissano Government Station was at D and Nimas village (the easternmost Sissano village, mentioned in the text) at C. Sissano Mission was at B. Re-settlement villages are Rainbrum, Olbrum, Rowoi, Wuipom, Barupu, Wauroin, Arop-1 (new), Amu, Tainyapin, Aindrin, Aipokon, Teles and Lambu. The original Teles and Lambu villages were on the coast directly seaward of the new villages of the same name. Aitape jetty is just south of the letter A of 'AITAPE'. The main focus of the wave was on the 14-km sector of coast from Mak to just beyond Warapu, DE. Here the wave heights were 10 m or more, all buildings were destroyed, and destruction extended as far as 500 m inland. From the mouth of the Bliri River to Sissano (AB), and from Mak to Tarau Point (FH), wave heights were not more than 4 m above sea level. In sectors AB and GH, only structures close to the water's edge were affected by the waves.

2. Debris floating in Sissano Lagoon, Sunday 19 July 1998. Photograph by Fr Z Mlak.

The first scientific investigation of the tsunami was carried out by an international team that entered the area on 1 August 1998. The team remained in the field for a week and reported their findings to the authorities in Aitape, and then in Port Moresby, the national capital (ITST, 1998). Other investigations followed, including a second visit by an international team in September 1998, and four major marine surveys. Results of the investigations have been presented by Kawata, Tsuji et al. (1999), Kawata, Benson et al. (1999), Tappin et al. (1999, 2001), Synolakis et al. (2001) and Matsumoto et al. (this volume), amongst others.

From experience at the time of the Rabaul seismic crisis in 1983 and the eruption that followed in 1994, we were aware of the need to get reliable information to the survivors and disaster managers, not only because of concerns for their safety, but also for peace of mind. At the briefing in Port Moresby, the international team confirmed that there was such a need at Aitape. Accordingly we mounted a public information program, on a small scale, starting 13 August 1998, and continuing today. While carrying out the program there was opportunity to interview many of the survivors and, in later months, to spend some time in geological investigations (Davies, 1998a, b; J.M. Davies et al., 1999; Davies et al., 2000; 2001). Information from the interviews and field investigations is brought together in this paper.

2. The Setting

The island of New Guinea (Fig. 1) lies athwart the northern margin of the Australian continent; is 2500 km long and up to 800 km wide; and has a central mountainous spine with peaks to just under 5000 m. The mountains are flanked on the south side by broad plains, and on the north by riverine plains and coastal ranges. The western half of the island is the Indonesian province of Papua (was Irian Jaya); the eastern half of the island, together with adjacent islands, forms the Independent State of Papua New Guinea (PNG).

The island was formed by successive collisions between the northward-advancing Australian craton and exotic and para-autochthonous terranes. Currently, in the border region, the Pacific and Australian plates are converging at a rate of 11.1 cm/yr on an azimuth of 070^o (Pegler et al., 1995). Crustal shortening is accommodated by oblique subduction at the New Guinea Trench (Fig. 1), by faulting in the hanging wall plate, and by faulting and folding in a foreland thrust belt on the south side of the central ranges.

Aitape is on the north coast, 160 km east of the international border. This is an area of low-lying coastal plain broken by isolated hills of basement rocks: Oligocene volcanic arc rocks and associated limestone. Further inland, foothills give way to the steep-fronted Bewani and Torricelli ranges. The coastal plain is bounded seaward by a coastal sand barrier that stands 1-2 m above sea level and is typically a few hundred meters across (Fig. 3). The sand barrier is highest at the beachfront and slopes gently downwards away from the sea -- a common morphology for coastlines where sea level is rising relative to the land, and where there is a steady supply of sand distributed along the coast by longshore drift (e.g., see Komar, 1998; Figs. 2-22 and 2-25). Much of the sand barrier is planted with coconut palms and there are occasional large trees (kalopilam Callophylum inophyllum; breadfruit or kapiak Atocarpus altissima; talis Terminalia catappa; and mango Magnifera indica) and thickets of yar (Casuarina equisetifolia).

3. Aerial view of Sissano Lagoon and the sand barrier, September 1998. Photograph by H. Davies.

Surf beaches of grey sand form the front of the sand barrier. Beyond the beaches, sandbanks extend for 100-130 m, and beyond the sandbanks a gently-sloping shelf extends out to sea for 10-15 km. From the shelf edge an irregular broken slope descends to 4000 m water depth, in the floor of the New Guinea Trench (Fig. 1). A submarine canyon cuts through the shelf in the east.

The beaches are not protected by reefs or islands and for much of the year waves 1-2 m high break on the near-shore sandbanks. The surf is stronger in the Northwest Monsoon season, December to March, when ocean swells roll in from the open ocean, and is calmer in the season of Southeast Tradewinds, May to September. There is a longshore drift from the northwest in the Monsoon season, and from the southeast in the other season.

Aitape is a small commercial center, district administrative headquarters and port, served by regular air services and accessible by road from Wewak in the east, and by 4-wheel-drive track and road from Vanimo in the west, when the weather is dry. Minor roads connect to some of the nearby villages and the coastal villages are accessible by powered dinghy.

Before the tsunami, about 12,000 people lived in the coastal villages west of Aitape, from Malol to Sissano (Fig. 1). Most houses were of traditional materials, and most were within a few hundred meters of the waterfront and on land that was not more than a few metros above sea level. Each village extended for a kilometer or more along the coast.

3. The earthquakes of 17 July 1998

The events of 17 July 1998 began with a strongly felt earthquake of M_W 7.0 at 6.49 pm local time (0849 UT), followed by a series of aftershocks. There was a main shock at 08:49:13 Universal Time and a widely felt aftershock, itself composed of two events at respectively 09:09:32 (m_b = 5.6) and 09.10:02 (m_b = 5.9). In between, two small aftershocks are documented by the National Earthquake Information Center at respectively 09:02:06 (m_b = 4.4) and 09:06:03 (no magnitude reported):

(Synolakis et al., 2002). The epicenter of the main shock was about 4 km north of the Sera villages (Fig. 1) and the epicenters of the aftershocks lay along a line east from the main shock, the first due north of Sissano-Warapu and the second due north of Malol (Hurukawa et al., this volume). The focal mechanism of the main shock can be interpreted as either a steep reverse fault dipping to the north (e.g., Geist, 2000), or a low-angle thrust dipping to the south-southwest (Hurukawa et al., this volume).

The main shock was sufficiently vigorous and prolonged that at Malol, Arop and Warapu people left their houses and moved into open space. At Arop and Warapu cracks opened in the ground, and water squirted upwards, house foundation posts shook and water rose around the posts, and there was a smell of hydrogen sulfide. At Sissano Mission the earthquake caused minor damage to the 62-year-old church, and in the nearby villages some houses collapsed. At Malol the shaking was strong enough to cause concern that the water tanks at the Mission might collapse. At Vanimo, 140 km from the epicenter, the earthquake was described by one long term resident as stronger and more prolonged than any he had experienced.

A weak foreshock that preceded the main shock by about ten minutes was felt by observers at Malol, Arop, Warapu and Sissano, but was not recorded by observatories at Jayapura and Port Moresby. The foreshock had a gentle sideways motion, and people referred to this as a 'normal' earthquake, unlike the sharp shaking and fast up-and-down movement of the main shock that followed.

Surveys of the felt intensity of the earthquakes by Y. Tsuji and others (unpublished data) and Ripper et al. (1999) confirmed that the felt intensities were such as would be expected from a M_W 7.0 earthquake. In the Aitape area, and inland from Aitape at Nuku, the aftershocks were felt more strongly than the main shock.

4. The tsunami

The main shock was followed, some minutes later, by a loud boom, as though of thunder; this was heard from Sissano to Malol. A few minutes or up to five minutes later there was a roaring sound, variously described as the noise of a low-flying heavy jet plane, the approach of a large ship, or as the woop-woop-woop of a heavy helicopter. The sound progressed eastward along the coast then back again to the west, and was heard all along the coast from Sissano in the west to Aitape High School (near Tadji airport, Fig. 1) in the east.

Although the sun had set at 6.37 pm, there was still sufficient daylight that the day's activities were continuing. Men were painting a canoe, young people were playing touch football and their elders were moving around in the villages. People went to the beach to investigate the unusual noise and observed that the sea was 'boiling' or bubbling, and had receded by 50 m or so, exposing the nearshore sea bed. They then saw a wave develop in the distance, as a dark line on a sea surface that otherwise reflected the light of the sky. The wave approached and, when 200-300 m from the beach, started to break, rolling from the top. 'Smoke' or haze rose from the top of the wave, and many saw a red glow in the top of the wave.

One observer (John Sanawe, a former Colonel in the PNG Defense Force) reported that he first saw the sea on the skyline rise and explode, sending spray high in the air where it caught and diffracted the late afternoon sunlight into rainbow colors. He then heard a sound like distant thunder. He wondered at hearing thunder on a day when the sky was clear, then linked this sound to the explosion. Then there was a sound like a heavy helicopter, or such as can be heard when a bottle is held under water, and the sea started to retreat from the shore. The rhythm of the helicopter noise slowed as the retreat of the sea slowed. Then there was silence for 4-5 minutes, followed by the noise of a low-flying jet aircraft. Sanawe looked to sea and saw that a wave had formed at or near the site of the explosion. The wave then approached at great speed.

People ran from the approaching waves but almost all were caught. A few escaped by climbing trees, or pushing their boats into the lagoon. People in the waves were vigorously tumbled and turned in water that was laden with sand and debris. They were stripped of their clothing, lost skin by sand abrasion, were battered by hard objects and some cut or impaled by timber and metal objects. Those who were fortunate were carried into the lagoon and were able to cling to floating debris. An infant was deposited miraculously on the floating roof of a house. Those less fortunate were carried into swampland or into the mangroves that fringe the lagoon where some were impaled or were buried under piles of logs and debris. Some who had survived the initial impact were swept out to sea as the waters receded. Most had ingested water from the waves.

By 7.20 or 7.25 pm the water had retreated, though much standing water remained. At this moment, according to survivors, the sky was filled with a yellow or yellow-red glow that provided sufficient light for people to start searching for family members. They said that without the glow this would not have been possible (see also 5.6 Unusual lights).

At Sissano Lagoon a low haze had advanced with the wave and this now blotted out the stars, so that it became pitch dark, so dark that people moving inland, away from the lagoon, held on to each other to maintain contact.

Rescue began that night, the survivors helping each other. The first outside help arrived 16 hours after the event on the Saturday morning, and a major rescue effort began a day later, 40 hours after the event. By Monday evening (72 hours) all surviving injured were in hospitals at Vanimo, Raihu (Aitape) and Wewak (Taylor et al., 1999; Davies, in prep.).

4.1 The shape of the waves

The shape of the tsunami waves can be reconstructed from interviews with survivors, and from mapping the damage and the trails of debris left by the wave (Davies et al., this volume). The mapping made use of low level aerial photographs that were flown three weeks after the tsunami (National Mapping Bureau, 1998).

The tsunami was seen by observers as an initial lowering of sea level followed by three large waves. It caused damage along 45 km of coastline (AH in Fig. 1), but maximum energy was focussed on a 14-km sector (DE in Fig. 1). Here wave heights, on shore, were 10-15 m above sea level and there was extensive damage for distances of up to 500 m from the coast. Damage was less on either side of the 14-km sector. Towards the periphery, in sectors AB and GH, wave heights were lower and only structures close to the water's edge were damaged.

Three waves, about 2 m high, caused minor damage at towns and villages along the coast to the west-northwest for a distance of 250 km (Joku, this volume) and smaller waves were recorded 170 km to the east-southeast at Wewak, and 500 km to the east-northeast at Bipi Island, Manus Province. A wave with a peak-to-peak amplitude of 7 cm was detected on the tide gauge in Rabaul harbor, 1000 km to the east, and smaller waves were detected on tide gauges in Japan (Tanioka, 1999).

Within the 14-km sector, the frontal wave was observed to approach the coast obliquely from the east, striking the coast first at or near Mak (E in Fig. 1; Figs. 4, 5) and progressing to the west, so that the wave struck Arop before it struck Warapu (Fig. 1). At Warapu the wave front was curved, or C-shaped in plan view, with the result that the western and eastern ends of the village were hit before the center (Fig. 4).

	version
4. Progress of the wave along the Aitape coast as determined from interviews and damage mapping. The heavy lines indicate the approximate alignment of the wave front as it intersected each section of the coast. The numbers indicate the time the wave arrived at the coast, shown as minutes after the earthquake of 08:49 hours Universal Time.

5. Computer model of the wave front from Synolakis et al. (2002), reproduced by permission of The Royal Society.

People in a powered dinghy returning from Aitape observed the westward approach of the wave which hit the beach behind them. They accelerated to maximum speed of perhaps 35-40 km/hr, trying to outrun the wave but, as the boat approached Arop the wave loomed and broke 200-300 m out to sea, caught the dinghy and carried it ashore, killing all but one occupant.

Eye-witnesses at Arop and Warapu described how the first wave came ashore as a broken wave or bore, about 1 m high. People were knocked over by the wave and by the cushion of air moving in front of the wave, but houses were not much damaged. Jerome Kinisa, who was on low land just east of the mouth of the lagoon, described how the wave broke offshore then rushed at speed across the peninsula, as a 1-m-high mass of foaming water, sand and logs. He was carried by the wave into the middle of the lagoon. At Arop and Warapu, the second wave reached the shoreline before the first had withdrawn, loomed to tree-top height and then crashed down upon the villages. This was the wave that caused most destruction. At Warapu the third wave followed closely behind the second, and a weak fourth wave followed after that.

West of the 14-km sector the wave progressed westward, and east of the 14-km sector the wave progressed eastward. This pattern: the wave moving outward away from 14-km sector in both directions, suggested that the source of the tsunami was directly offshore from DE (Davies, 1998a, p.28).

At Aitape the three waves approached around the Aitape headland, from the northwest. The first wave came ashore at Namba 2 Pasis (Fig. 1) and was reflected back to meet the second wave at the headland. The waves then progressed to east and southeast. The wave crests were estimated to be 500-600 m apart (observations by Dickson Dalle). As observed here, and at Sissano Mission, the three waves were of about equal height.

4.2 The damage

The greatest damage was in sector DE (Fig. 1). Here the villages of Arop and Warapu were removed almost without trace, leaving only the concrete foundation slabs of churches and classrooms (Fig. 6). House foundation posts were lifted out of the ground and removed, except for a few at Warapu, and massive concrete blocks at Warapu were carried for distances of 50-60 m.

6. Three concrete slabs (right) are all that remain of the class rooms at Warapu school. Photograph H. Davies.

In contrast to the destruction of man-made objects, mature coconut palms and well established, deep rooted trees with extensive lateral rooting, such as the many casuarinas (Casuarina equisetifoli) and the occasional callophylum (Calophyllum inophyllum; Fig. 7), kapiak or breadfruit (Atocarpus altissima), talis (Terminalia catappa), and mango (Magnifera indica), withstood the waves. A robust callophylum at Warapu provided some protection for Camillus Kanuvo and family from the worst of the first wave. People who climbed trees and palms at Warapu and at Arop-2 escaped the wave (Fig. 8). The less salt-tolerant of these trees, notably the mango trees, died some months later.

7. A callophylum tree on the coast of Tumleo Island. Note the blocks of coral reef material carried ashore by the tsunami. Callophylum is a deep rooted and salt-tolerant tree that can provide protection from tsunamis.

8. A survivor re-enacts his escape from the wave by climbing a small tree on one of the islands just inside the mouth of the lagoon. Photograph by H. Dennett.

The beachfront was eroded by the wave so as to create a low escarpment 1-2 m high at the front edge of the vegetated sand ridge. This effect was seen from Malol in the east to Sissano in the west. The low cliffs were modified by the return flow of the water which carved channels through the beachfront berm.

At the back of the sand ridge, on the lagoon side, the waves caused scouring to depths of several meters, and scours were then partly filled with sand (Fig. 9). The scouring occurred along the line where the second wave dropped on the houses, and reflects the great turbulence at this point. Scouring was most severe where the ground already had been disturbed, e.g., where house posts had stood, or in the graveyard near the Warapu church; or where there was any steepening in the slope of the sand ridge back towards the lagoon. The scouring and its causes are discussed by Matsutomi et al. (2001).

9. Scouring by the tsunami waves on the inner side of the sand barrier at Warapu village. Photograph H. Davies.

Further west at Nimas, the easternmost of the Sissano villages (CD in Fig. 1), the damage was marginally less severe. Although most houses were destroyed, there was scouring in only a small area at the seaward end of the village and house posts remained standing, all canted inland by the force of the water. Four houses , 400 m from the coast, were preserved intact. In the other Sissano villages, at BC, many houses remained standing, but at the western end of BC the waves destroyed buildings at Sissano Mission (B in Fig. 1), including the great church that had stood since 1926. At the Mission, buildings more than 400 m from the coast were not damaged.

Dramatic evidence of the power of the waves was preserved in the low country inland from Arop and Mak, at the eastern end of DE, where clumps of wild sago palm were lifted and overturned by the waves, and tall trees were uprooted and transported inland for hundreds of meters, like so much match wood (Fig. 10). This swathe of destruction tailed off east of Mak, in sector EF.

10. A 500-m-wide swathe of destruction between Arop Community School and the beach. Photograph by H. Davies.

At Malol, seaward of the lagoon, houses within 50 m of the water's edge were destroyed but most other houses were left standing. The mission station and the adjacent villages , 200-300 m from the beach, were not damaged; this included Tainyapin village, which is on a low-lying island in the lagoon. East of the Yalingi River, the greater part of the Malol villages of Lambu and Teles was destroyed leaving only the western part standing. This was protected by the headland of the Yalingi river mouth. Lambu and Teles were on a low, poorly vegetated sand ridge, less than a meter above high water mark, that slopes downward away from the beachfront berm. There was no protection and almost no hindrance to the waves which swept in from the northwest. Even so, the damage extended only a limited distance inland, and houses on the inland side of the coast road (about 150 m from the waterfront) were undamaged.

From the Waipo River to Tarau Point (GH in Fig. 1), and from Sissano Mission to the mouth of the Bliri River (AB in Fig. 1), the energy of the wave was less and only houses close to the water's edge were damaged.

4.3 Height of the waves

Wave height information was collected by the international team (ITST, 1998) and reported by Kawata, Tsuji et al. (1999) and Kawata, Benson et al. (1999). They found that wave heights reached 10-15 m in the area of maximum devastation (DE in Fig. 1). These extreme wave heights were confirmed by survivors at Arop and Warapu, who spoke of waves as high as the tops of the coconut palms. At Arop and Warapu it was the second wave that reached these heights.

However, it is unlikely that the waves were 10-15 m high as they approached the beach. All estimates from eye witnesses who watched the waves approach the shore are that the waves were about 4 m high. This is the height estimated by observers on the beach at Arop and Warapu, and by the sole survivor of the powered dinghy that was caught by the wave offshore from Arop.

This is an important point because it bears upon the classification of the tsunami in terms of tsunami magnitude (see under Discussion), and because it revises a key data point in the modeling of the source and path of the tsunami. The published models have sought to explain a wave that had a height of 15 m as it approached the shore, not 4 m (e.g., see Fig. 6 in Tappin et al., 2001; and Fig. 9 in Synolakis et al., 2002, reproduced as Fig. 5 in this paper).

According to eye-witnesses the first wave started to break, rolling from the top, at a point 200-300 m from shore and reached the shore as a broken wave. (Our reconnaissance survey in January 2002 recorded water depths of 10 m at a point 200 m from the shore at Arop, and 11 m at 300 m from the shore.) The first wave then swept through Arop and Warapu villages and the lagoon mouth as a bore about 1 m high (which equates to about 2 m above sea level). It was the second wave that reared to great height and crashed down on the two villages.

Although we do not know for sure what happened, one can imagine that the second wave, travelling fast on the back of the flooding caused by the first wave, was deflected upwards when it reached the shore, because of the change in slope at the beach and the beachfront berm. The first wave had progressed across the sand barrier and into the lagoon, and so had not retreated before the second wave arrived. The effect of the flooding by the first wave would have been to reduce bottom friction and provide an elevated platform on which the second wave approached the beach. One eye-witness (John Sanawe of Arop) described the second wave as being deflected upwards when it reached the beachfront.

It is unlikely that the second wave, while at sea, was inherently bigger than the other waves. All eye-witnesses who saw the three waves as they approached the shore described them as being approximately equal in height.

East and west of DE, wave heights recorded by the international team were lower, at 2-4 m above sea level (Kawata, Benson et al., 1999). At only one location outside sector DE did these authors record a wave height of 10 m (at Uyang, the westernmost part of the Malol villages). Our investigations suggest that this was either an anomalous local effect, or an incorrect reading of the field evidence. The observation was not confirmed by Matsutomi et al. (2001).

4.4 The timing of the waves

Information about when the first wave hit different parts of the coast (Fig. 4) can help to define the location of the source of the tsunami. However, precise information is not easily obtained, because people in the villages generally do not concern themselves with knowing the time to the minute but, rather, take the time from the sun. In the late afternoon, village activities continue until after sundown -- people moving around, young people playing -- and when the light becomes insufficient to move around freely people sit down to the evening meal. There is no street lighting in the villages and, in fact, no electric light.

We sought information about the timing of the wave using two reference points: (1) did the wave arrive before or after the strongly-felt aftershocks (which occurred at 09:09 and 09:10 hours UT), and (2) how much daylight remained when the wave arrived. Of these, the first is the less subjective.

Our enquiries show that the wave reached villages in sector BE (Fig. 1) before the aftershocks. Eye-witnesses were unanimous about this. Some survivors at Warapu felt the aftershocks while they were in the water, for example Elizabeth Bade, who was carried from eastern Warapu to the islands just inshore from the lagoon mouth, felt the aftershocks while in the wave. The priest at Sissano Mission, Fr Otton, felt the aftershocks after having run 300 m to escape the wave.

At Malol too the waves may have arrived before the aftershocks (several reliable witnesses) but there are conflicting accounts. A group of young men, interviewed at Malol in October 1998, recalled that the wave reached Malol just as the shaking from the strongly-felt aftershocks ceased. At the time of that interview, there was no opportunity too cross-check with other Malol survivors, because all had dispersed to inland camps. The information from the youths was published (Davies, 1998a, p. 26) and was adopted as a key argument in at least one published paper (Tappin et al., 2001), but could be in error. More recent enquiries at Malol have not resolved the question. A number of survivors recount that the wave reached Malol some minutes before the strong aftershocks, but others maintain otherwise.

From the observers in sector BE (Fig. 1), we know that the waves reached all of the villages in this sector before the strongly felt aftershock at 09:09 hours UT. From other sources we know that the waves progressed westward along the coast, beginning at Mak, or possibly at Malol (Fig. 4). We do not know how much time was needed for the wave to progress from Mak to Sissano Mission, but do know that some significant time elapsed, because people saw the wave as it approached from the east, and had time to try to outrun the wave in a powered dinghy (in one instance), and, at Arop-2 and Warapu, to climb trees and to jump into canoes that were pulled up on the shore of the lagoon. If the progress from E to B took 5 minutes or more, as seems likely, then the wave first came ashore at Mak (or Malol) at or before 09:04 hours UT.

The youths at Malol recalled that the waves approached square on to the coast, i.e. heading due south.

On several evenings during recent fieldwork, we asked eye-witnesses at Malol and Warapu to estimate the time of arrival of the wave, based on their recollection of how much daylight remained. From this line of enquiry it appears that, at least at these villages, the first wave arrived after 09:00 UT (7 pm local time), possibly at around 09:05 UT (7.05 pm local time), and certainly before 09:09 hours UT.

15 km east of Malol, at Aitape jetty (Fig. 1), the arrival of the waves was recorded by several observers independently at 09.14 or 09:15 UT (7.14 or 7.15 pm), 2-3 minutes after the aftershocks. At Tumleo Island and at Pro (near Lemieng, Fig. 4) observers estimated that the waves arrived "about 5 minutes" after the aftershocks.

4.5 What was in the wave?

People who saw the broken wave approach the mouth of the lagoon described it as a dirty mass of broken water filled with dark sand, logs and other debris. The sand was deposited across much of the devastated area as a layer about 10 cm thick, thinning to a few cm close to the waterfront and towards the inland limit of deposition (Jaffe et al., 1998; Davies, Gedikile, Nongkas and Simeon, this volume). The sand deposit is coarse at the base and fine at the top, and includes rip-up mud clasts. Locally the deposits are much thicker, especially in areas where the ground surface had a steeper than normal negative gradient (sloped more steeply away from the waterfront), and where the waves had caused scouring (see Fig. 9; Davies, 1998a, pp.22-23).

Probably much of the sand was lifted from the beachfront and the immediate offshore sand banks as the wave came ashore (see also Matsutomi et al., 2001). At Teles-Lampu the sands included pebbles up to 1 cm across, that presumably were carried from the mouth of the Yalingi River.

The wave also carried much marine life and deposited this on the shore or in the lagoon, including sharks, quantities of fish, and some sea turtles. The fish species included Red Emperor (Lutjanus sebae), a bottom-feeding snapper species that normally inhabits depths of 30-50 m, and living specimens of a species of foraminifera that normally lives at depths of 250-300 m: a keeled form of Globorotalia. Foraminifera were found only in sands collected at Teles (Fig.1).

4.6 Hot and smelly

All survivors in the Arop-Warapu area reported that the water in the wave was hot, and many reported that it had a bad smell and stung the skin. Some described the smell as a 'chemical' smell, and others as 'rotten eggs' (hydrogen sulfide). Others, particularly at Arop, noted the smell of kerosene or petrol or oil. People speculated that the impurities that caused the stinging effect also caused wounds to become infected very quickly, and caused the skin on bodies floating in the lagoon to turn black. (Medical authorities disputed this and maintained that the blackening of the skin was to be expected in the normal course of events.) Survivors further to the east (e.g. at GH, Fig. 1) did not mention any unusual temperature or smell of the water in the wave.

The reports of oil in the water pointed to the possibility that explosive escape of gas may have been a factor in the tsunami (Davies, 1998a, p. 16, and see John Sanawe's observations under 4, and discussion under 5.2 and 5.3, below).

4.7 Other phenomena

Possible ground deformation. John Kimene of Nimas was one of a group that was fishing at a drowned reef 8-10 km from the coast at about 10 am on Thursday 16 July 1998. This probably is the reef marked on the map (Fig. 1) just inside the 200 m isobath, which stands at 82 m depth. As the party trolled 1-2 km west of the submerged reef they were surprised to run into a succession of 2-3 m waves that loomed and steepened as though about to break. They took this to be evidence of a new shoaling of the water in an area that previously had been quite deep. There was a smell of dead fish.

Escape of gas. On Friday 17 July 1998 at about noon Tom Kaisiera, a teenager from Nimas, paddled to the same general area and was surprised to find the sea bubbling with odorless gas. The area of bubbling was large, perhaps 100-200 m across. The canoe was drawn toward the center of bubbling area and it was only by paddling strongly that he could escape (Davies, 1998a, p.16).

Unusual lights. Three unusual lighting effects were reported. Many observers saw a red light on the horizon before the tsunami developed: "After the first earthquake, a long streak of red light like fire appeared just above the ocean on northern horizon, it flashed and then disappeared, then within seconds there was a loud bang". Also, many observers described a red glow or "fire" in the top of the wave.

After the wave passed, observers at widely separated locations (Warapu, Malol and Raihu) saw a yellow or yellow-red glow in the sky over the sea. "The sky lit up after the wave had destroyed the villages" (observer at Malol) and "after I climbed down from the tree I saw a big light over Arop and in the direction of Aitape" (observer on an island near the lagoon mouth). The Sisters at Malol recall that after the waves had passed they looked seaward and saw a calm golden sea. Warapu survivors recall that the yellow glow in the sky helped light their search for survivors.

4.8 Subsidence

There has been 50-70 cm of subsidence in an area from Nimas (C in Fig. 1) to Arop, and possibly beyond Arop towards Mak. This is evidenced in the flooding and erosion of the tip of the Warapu peninsula at the mouth of the lagoon (Fig. 11), and in the high water table at the former village sites of Warapu and Arop. Subsidence is thought to be the cause of the dying back of some vegetation a few hundred meters inland from Nimas. The subsidence may have been partly co-seismic and perhaps mostly post-seismic.

11. Subsidence at the former site of Warapu village is permitting erosion of the Warapu peninsula at the mouth of the Sissano Lagoon. Photograph by H. Davies.

While in the field in the first week of August 1998 the international team searched for evidence of subsidence but found none. We too saw no proof of subsidence on several visits in August and September 1998. Another team that entered the area in August saw and photographed, on 17 August 1998, the partly submerged bole of a tree on the lagoon shore at Arop, and grass that had been flattened by the high tide and cited this as proof of subsidence (Goldsmith et al., 1999; McSaveney et al., 2000). Because no other observers saw this, it seems that the flooding seen on 17 August 1998 may have been an ephemeral effect due to an influx of river water into the lagoon and restricted outflow through the lagoon mouth. (Flow through the lagoon mouth was restricted by newly-formed sand banks for five to six months following the tsunami.)

The present Sissano Lagoon was formed by an earlier, co-seismic subsidence event, in 1907 (Neuhauss, 1911; Welsch, 1998).

5. Discussion

The Aitape tsunami was the worst natural disaster in the history of the Independent State of Papua New Guinea, and compares with the catastrophic eruption of Mount Lamington in January 1951, in which 3000 people were killed. Since the beginnings of written history in this region, the Aitape tsunami is exceeded in impact by only the 1888 tsunami, triggered by the collapse of Ritter Island volcano (Everingham, 1977).

The tsunami happened at a time when the nation was ill prepared. There had been no major tsunami in Papua New Guinea since 1930 (Everingham, 1977), sufficiently long ago that recollections had been lost or dulled. Recent catastrophic tsunamis in Indonesia that should have triggered a state of increased awareness, including Flores in 1992 and Biak in West Papua in 1996, had passed without drawing much attention in the media or in disaster management circles. In addition, the promotion of tsunami awareness by Government had lapsed in recent decades, since the efforts of Everingham (1977), and the government agency responsible for disaster management was under-manned and temporarily without telephones.

Despite these disadvantages, a response to the disaster was cobbled together, and was followed by a strong recovery program. Vital to the response and the recovery were the contributions of individuals and groups within country and overseas, the generous donations of funds, goods and services, and the concerted efforts of key people on the ground.

The scientific investigation of the tsunami, similarly, was an amalgam of international and locally-based contributions. Key amongst these was the arrival of the international team two weeks after the disaster, their rapid investigation of the situation, and their reporting back to the authorities at the end of the third week. The members of the team made all information available to the PNG-based scientific community, as photocopies of field data initially and as a bound report later in the year (ITST, 1998; Kawata, Tsuji et al. 1999). After their departure they continued to provide support in the way of advice and encouragement.

The other major contribution to the investigation of the tsunami was the funding of four research cruises in 1999-2001 by the Japanese government, through the Japan Marine Science and Technology Corporation (JAMSTEC; Tappin et al., 2001; Matsumoto and Tappin, this volume). The cruises were carried out at the request of the South Pacific Applied Geoscience Commission (SOPAC), acting on behalf of the PNG Government.

For PNG-based scientists the first objective was to provide reliable information to the public, especially those in the disaster area. There was no umbrella organization nor budget for this, and some of those involved in management of the disaster did not accept the need for the program. So it was necessary to begin the program without funds, to establish the credibility of, and need for, the program, and then to seek funds so it could continue. Because of the low profile of the program initially there were difficulties arranging transport in the field and time was wasted. Over several months the program gained support, particularly amongst the managers and victims in the field at Aitape, sponsorship was found, and operations became more efficient.

In the course of the public information program it was inevitable that information would be gleaned from the survivors and that, over time, a considerable store of information about the tsunami would be accumulated. However, these enquiries were never the main thrust of the program and were never pursued exhaustively. More could have been learned with a concerted program of enquiry. Also, stories could have been cross-checked and independently verified. The uncertainty about the time of the arrival of the wave at Malol is a case in point.

Although the PNG-based scientists had very limited financial resources, they did have the advantage that, unlike their overseas counterparts, they could make repeated visits to the field area (18 visits in 3 years), and gather information in a more leisurely and systematic way, long after the emergency had eased. This has resulted in, for example, a better understanding of the timing and path of the wave and better documentation of the damage. The repeated visits also were vital to the success of the public information program because they provided an opportunity to build linkages with community leaders and to identify and respond to community concerns. A prime concern was to persuade the communities that the tsunami was a natural event, and not man-made. In this way, the program has been a factor in helping the communities return to normalcy.

There are still a number of issues about the tsunami that are unclear or unresolved as are discussed below.

5.1 The cause of the tsunami

Because the wave heights of 10-15 m were greater than should be expected from an earthquake of magnitude 7, members of the international team suggested that there may have been an augmentation of the wave by some secondary process, such as a submarine landslide (as reported in Davies, 1998a, p.7). Subsequently the discovery of submarine landslide deposits (Sweet and Silver, this volume) and of extensional fissures in the sea floor sediments (Tappin et al., 2001; Matsumoto and Tappin, this volume) has provided supporting evidence. The hypothesis that the Aitape tsunami was generated by a slump or landslide of sea floor sediments has become widely accepted (Tappin et al., 1999, 2001; Synolakis et al., 2002).

Synolakis et al. (2002) concluded that the slump occurred at 09:02 hours UT (7:02 pm local time), at a location where fissures in the sea floor had been mapped, about 30 km north-northeast of Arop. The slumping generated a T-wave signal that was picked up by hydroacoustic and seismic stations across the Pacific to the north and northeast of Aitape. A wave generated at this source would deliver maximum energy to the devastated area, DE in Fig. 1, and would approach the coast obliquely from the east. Depending on the dimensions of the slump block and the distance that the slump moved, the wave would have a maximum elevation of 10-15 m at the coast (Fig.5).

The model matches the field data with the exception that it requires the wave to reach Arop at 09:12 hours UT (7.12 pm local time), which is after the aftershocks and is 4-5 minutes later than our best estimate of the arrival time, based on survivor accounts. This raises two possibilities: (a) that the 09:02 UT seismic event does not mark the initiation of the tsunami; or (b) that the source of the 09:02 UT seismic event was much closer to the shore, perhaps as little as 10-12 km from Arop. The sea floor at this possible source area has not been mapped.

5.2 The possible involvement of gas

An escape of gas may have accompanied or even initiated the tsunami. The evidence includes the vigorous bubbling of gas encountered by a lone fisherman on the day of the tsunami; John Sanawe's description of what appeared to be an explosion in the wave that threw spray high in the air; the recollections by a significant number of the survivors that there was a smell of kerosene, petrol or oil in the wave; and Sanawe's observation that there was a film of light oil on vegetation at Arop on the Saturday morning.

The sediments that underlie the Aitape coast are known to contain petroleum. Seeps have been mapped in the Bewani and Torricelli Mountains (Hutchison and Norvick, 1980), and weak gas seeps are present at one location in Sissano Lagoon and at another offshore from the Waipo River (offshore from G in Fig. 1; information recorded during our investigations). In addition, there is a newly-recorded oil seep offshore from Lemieng (Fig. 4; information from staff of the Diocese of Aitape, January 2002). Gas seeps also were found in water depths of 1000-2000 m during the post-tsunami marine surveys; the seeps are marked by chemosynthetic faunal communities (Matsumoto and Tappin, this volume).

Gas escape on a more modest scale caused the 'boiling' of the ocean that was reported by many eye-witnesses. This was the escape of gas that is normally held in equilibrium in the sea floor sediments. The gas escapes when pressure at the sea floor is reduced at the time of the extreme lowering of sea level that precedes the arrival of the positive tsunami wave (M. Hovland, pers. comm. 1999; Hovland and Judd, 1988).

Soter (1999) has suggested that earthquakes may be triggered by gas ascending from depth. Escape of some of this gas to the surface would explain why gas venting might precede an earthquake.

5.3 Hot and smelly

One of the most puzzling features of the Aitape tsunami was that most or all witnesses in the Sissano Lagoon area described the water in the waves as unusually hot. This remains unexplained. Possibly this too was linked to the escape of gas.

The heating effect of escaping gas was demonstrated in the Gulf of Corinth in the eastern Mediterranean where Hasiotis et al. (1996, cited by Soter, 1999) recorded that water temperatures in the Bay of Patras increased by as much as 6^oC in several pulses before the 1993 Patras earthquake. They attributed the heating events to the escape of gas from mapped vents on the sea bed. Soter (1999) cited other references to the heating of ocean water at the time of the 1817 Aigion earthquake and tsunami, in the same area.

Water in the deeper, more stagnant parts of Sissano Lagoon -- parts that are not flushed by tides or by fresh water from the rivers -- also can become uncomfortably hot, as we found while free diving from the shoreline at old Arop. The heat presumably is due to the decay of vegetable matter. However, this warm water probably is not present in sufficient quantity to explain the warming of the large volume of water that was in the waves. Nor would it explain warm water in waves that arrived directly from the sea.

Hydrogen sulfide in the tsunami wave is more easily explained. Hydrogen sulfide that could be smelled during the first strong earthquake was released from sub-surface sediments as they started to liquefy. H₂S is widely present in the lagoonal mudflats and in the sediments just below the floor of the lagoon (our observations), and probably also is present in the sediments below the sea floor which are, at least in part, mudstones that accumulated in a lagoonal environment, before subsidence and shoreline retreat (e.g., see Komar, 1998, Figs. 2-22, 2-25).

Another puzzle is that the water stung the skin. This remains unexplained. Did the water become weakly acidic, and if so what might have caused this? Perhaps carbon dioxide gas escaping from the sediments beneath the sea floor or beneath the lagoon caused acidification.

5.4 The magnitude of the tsunami

Although there was clear evidence for maximum wave heights of 10-15 m onshore in sector DE (ITST, 1998; Kawata, Tsuji et al., 1999; Kawata, Benson et al., 1999), it is almost certain that the height of the waves as they approached and reached the beach was about 4 m (see under 4.3, above). The wave heights of 10-15 m were likely caused by upward deflection after the wave reached the coast. This has implications for computer modeling of the source, and for the classification of the tsunami. A maximum wave height of 4 m indicates a tsunami magnitude of 2, rather than magnitude 3 for a wave height of 10-15 m (Imamura-Iida scale of tsunami magnitude; Iida, 1958, cited by Shuto, 1999).

5.5 The thunderous boom and the roaring noise

Almost all observers in the Sissano Lagoon area recall a thunderous boom or "pairap" (Tok Pisin) as though of thunder, that was heard after the main shock, and that this was followed some minutes later a roaring noise as though of a low-flying heavy jet plane or a large ship. Some heard a woop-woop sound, as of a low-flying heavy helicopter, before the roaring noise. The jet like roaring sound traveled eastward initially (it was heard as far east as Lemieng (Fig. 4) and possibly beyond), then returned to the west. At the high schools east and southeast of Aitape it was heard as a whistling sound that progressed to the east, then to the west. Sanawe observed that the woop-woop sound coincided with the retreat of the sea from the shoreline, and slowed as the retreat of the water slowed.

Loud explosions have been heard at the time of other earthquakes and may be the effect of the P wave reaching the surface and releasing sound waves at frequencies greater than 15 cycles per second (Bolt, 1999). The delay between feeling the earthquake and hearing the sound can be explained because shock waves travel through rock faster than sound waves travel through air. Alternatively, as was suggested by John Sanawe (see under 4), the thunderous boom may have been caused by the gas-driven(?) explosion that he had witnessed.

Shuto (1997) reviewed many instances where a roaring sound had been heard at the time of tsunamis in Japan and concluded that the sound can be caused by the breaking wave, or by the rolling of pebbles as the wave retreats. Neither solution seems apt for the roaring sound that preceded the tsunami at Aitape. The source of the sound remains a mystery, as does the fact that the sound traveled along the coast to the east and then back to the west. Perhaps the sound was generated by displacement of air at the sea surface by the fast-moving low-amplitude tsunami wave, while it was propagating in deep water.

Soter (1999) notes that roaring sounds heard at the time of an are sometimes a precursor of the earthquake.

Survivors of the Aitape tsunami interpreted the thunderous boom and the roaring sound to be evidence that the tsunami was caused by a bomb that had been dropped by a low-flying jet plane. Persuading people against "the bomb theory" was one of the prime objectives of the public information campaign, because the hypothesis worked diametrically against people accepting the tsunami as a natural phenomenon, and so prolonged the period of uncertainty and rumor that followed the tsunami.

5.6 Unusual lights

Three kinds of light effects were reported by different observers at different points. Firstly, many observers saw a flash of red light along the seaward horizon at or near the time of the main shock. Was this a sunset effect (in the northern sky) or somehow connected to the earthquake? Then, as the wave approached, many saw a red glow in the top of the wave. This was seen not only at the area of most destruction (DE in Fig. 1) but also as far away as Aitape. This may have been bioluminescence, as suggested by Kawata, Benson et al. (1999), though eye-witnesses dispute this. Finally, after the waves had passed and all was quiet, many saw a luminous yellow glow in the sky. What might have caused this?

Unusual lights at the time of an earthquake have been widely reported elsewhere (e.g., Soter, 1999; Bolt, 1999, p. 202) but remain unexplained. Soter (1999) suggested the possibility that friction amongst particles entrained in a discharge of gas might cause a glowing electrostatic discharge.

5.7 Water circulation in the wave

The tsunami sands deposited at Teles included living foraminifera of a species that characteristically lives at 250 m water depth. This appears to indicate a surprisingly great depth of water circulation in the tsunami wave.

5.8 How dangerous is this coast?

Since written records began, at the time of the first European settlement in 1896, there has been no major tsunami on this coast other than the 1998 event. According to contemporary records (the account by Neuhauss (1911) and the diaries of A.B. Lewis (Welsch, 1998)) there was no tsunami at the time of the 1907 earthquake, and a tsunami described by McCarthy (1963) at the time of a major earthquake in 1935 was not severe and no deaths were recorded. On the evidence of that relatively short record, the recurrence interval of major tsunamis may be at least 100 years and possibly more. Currently we are investigating the pre-historic record.

The 1998 tsunami demonstrated that for villages in sector DE any tsunami is potentially dangerous, both because there are inadequate escape routes and because any tsunami generated near here will become focussed on this sector of coast, by virtue of the focussing effect of the shape of the sea floor (Tappin et al., 1999, 2001; Synolakis et al., 2002; Matsumoto and Tappin, this volume). Plantings of deep-rooted, salt tolerant trees (Fig. 7) will provide some protection from future tsunamis on this coast and elsewhere in PNG.

6. Conclusions

Information obtained in continuing investigations since 1998 builds upon and modifies information collected by the international soon after the tsunami. We present new information about the timing and westward progress of the waves, which arrived at the coast 4-5 minutes earlier than current models allow; and the maximum height of the waves while still offshore. The extremely high waves that left traces of their passage at 10-15 m above sea level onshore, as recorded by the international teams, are thought to have developed when a 4-m broken wave crossed the beach and was deflected upwards. Given that the coastal topographic profiles are known (Davies et al., this volume), it should be possible to test this idea in the laboratory. The revised wave height needs to be taken into account in models of the source of the tsunami. The escape of gas may have played a part in the tsunami and can be investigated by searching for gas blow-out structures in the as yet unmapped areas of the sea floor.

Acknowledgements

Field operations were facilitated by the Catholic Diocese of Aitape and specifically Bishop Austen Crapp, Father Tim Elliott, Balthasar Maketu, the Presentation Sisters, Philip Turner, and driver Oderich; by the Controller of the Emergency, the late Ludwick Kembu and the Deputy Controller, Vincent Tutu with Major John Sehi; Robert and Margaret Parer, Brother Stephen Garry and Stephen Matawe; District Office staff Dickson Dalle, Tuatureme Bodde and Martin Selmatin; and members of the Local Government Council. Rondi Davies drafted the map (Figs. 1 and 4) and Simon Saulei advised on species and names of trees. Members of the international tsunami science community provided essential advice at different times during the investigation, notably Fumihiko Imamura, Frank Gonzalez, Costas Synolakis and Yoshinobi Tsuji. Professor Imamura provided seed funding for field operations, and Professor Synolakis, funding to attend the 1998 Fall Meeting of the American Geophysical Union. The paper was improved by critical reviews from Eli Silver and Professor Imamura and by (other reviewers). Orogen Minerals Limited provided the initial financial support for field operations, and the National Disaster Committee provided funds subsequently.

References

Bolt, B.A. (1999): Earthquakes. New York, Freeman (4th edition)

Carey, S.W. (1935): Preliminary notes on a recent earthquake in New Guinea. Australian Geographer 2(8), 8-15.

Davies, H.L. (1998a): Tsunami PNG 1998 - Extracts from Earth Talk. Waigani: University of Papua New Guinea.

Davies, H.L. (1998b): Sissano tsunami: A reconstruction of the events of July 17, 1998, based upon interviews with survivors. 1998 Fall Meeting American Geophysical Union, Supplement to EOS, Transactions AGU 79(45) p.F572.

Davies, H.L. (this volume): The damage caused by the 1998 Aitape tsunami.

Davies, H.L., Davies, J.M, Lus, W.Y., Perembo, R.C.B., Joku, N., Gedikile, H., and Nongkas, M. (2000): Tsunami at Aitape - "Is it safe to go back?" (abstr.). Proceedings 15th Australian Geological Convention, Geological Society of Australia Abstracts 45.

Davies, H.L., Gedikile, H., and Nongkas, M. (this volume): The Aitape 1998 tsunami: coastal geology and tsunami sedimentation.

Davies, H.L. ( in prep.): Report on the Aitape 1998 Tsunami. Waigani, PNG National Disaster Management Office.

Davies, H.L., Davies, J.M., Lus, W.Y., Perembo, R.C.B., Joku, N., Gedikile, H., and Nongkas, M. (2001): Learning from the Aitape tsunami. Proceedings of the International Tsunami Symposium ITS2001, Seattle.

Davies, J.M., Davies, H.L., Joku, N., and Gedikile, H. (1999): The Aitape tsunami -- reconstructing the events and planning for the future at PARTIC. 1999 Fall Meeting American Geophysical Union, Supplement to EOS, Transactions AGU 80 p.F750.

Everingham, I.B. (1977): Preliminary catalogue of tsunamis for the New Guinea/Solomons region, 1768-1972. Bureau of Mineral Resources, Australia, Report 180.

Geist, E.L. (2000): Origin of the 17 July 1998 Papua New Guinea tsunami: Earthquake or landslide?, Seismological Research Letters 71:344-351.

Goldsmith, P., Barnett, A., Goff, J., McSaveney, M., Elliott, S., and Nongkas, M. (1999): Report of the New Zealand reconnaissance team to the area of the 17 July 1998 tsunami at Sissano lagoon, Papua New Guinea. Bulletin NZ Society for Earthquake Engineering 32, 102-118

Hasiotis, T., Papatheodorou, G., Kastanos, N., and Ferentinos, G. (1996): A pockmark bed in the Patras Gulf (Greece) and its activation during the 4/7/93 seismic event. Marine Geology 130, 333-344.

Holian, A. C., and Keith, P. P. (1998): Orthopaedic surgery after the Aitape tsunami. Medical Journal of Australia 169:606-609.

Hovland, M., and Judd, A.G. (1988): Seabed Pockmarks and Seepages. London, Graham & Trotman.

Hurukawa, N., Tsuji, Y. and Waluyo, B. (this volume): The 1998 Papua New Guinea earthquake and its fault plane estimated from relocated aftershocks.

Iida, K. (1958): Magnitude and energy of earthquakes accompanied by tsunami, and tsunami energy. Journal Earth Sciences, Nagoya University 6:101-112.

ITST (1998): Preliminary report on the tsunami caused by the earthquake at Sissano, Wesy Sepik, PNG, on 17 July 1998. International Tsunami Survey Team unpublished report, September 1998.

Jaffe, B., Gelfenbaum, G., Benson, B., Davies, H., and Nongkas, M. (1998): Sedimentation, erosion and flow in the July 17, 1998 Papua New Guinea tsunami. 1998 Fall Meeting American Geophysical Union, Supplement to EOS, Transactions AGU 79(45) p.F572.

Kawata, Y., Tsuji, Y., Matsutomi, H., Imamura, F., Matsuyama, M., and Takahashi, T. (1999): Field survey of the 1998 tsunami in the northwestern area of Papua New Guinea. Report to Ministry of Education, Science, Sports and Culture, Japan.

Kawata, Y., Benson, B.C., Borrero, J.C., Borrero, J.L., Davies, H.L., deLange, W.P., Imamura, F., Letz, H., Nott, J., and Synolakis, C.E. (1999): Tsunami in Papua New Guinea was as intense as first thought. EOS Transactions of the American Geophysical Union 80, pp. 101, 104-105.

Komar, P.D. (1998): Beach Processes and Sedimentation. New Jersey, Prentice Hall, 2nd edition, 543 pp.

McCarthy, J.K. (1963): Patrol Into Yesterday. Canberra, Cheshire, p. 156

McSaveney, M.J., Goff, J.R., Darby, D.J., Goldsmith, P., Barnett, A., Elliott, S., and Nongkas, M. (2000): The 17 July tsunami, Papua New Guinea: evidence and initial interpretation. Marine Geology 170: 81-92.

Matsutomi, H., Kawata, Y., Shuto, N., Tsuji, Y., Fujima, K., Imamura, F., Matsuyama, M., Takahashi, T., Maki, N., and Han, S.S., 2001, Flow strength on land and damage of the 1998 Papua New Guinea tsunami. In G.T. Hebenstreit (ed.)., Tsunami Research at the End of a Critical Decade. Dordrecht: Kluwer

National Mapping Bureau (1998): Aitape Disaster Area West Block and East Block, aerial photographs at 1:10,000 scale. Port Moresby: National Mapping Bureau.

Neuhauss, R. (1911): Deutsch-Neu-Guinea. 3 vols. Berlin: Dietrich Reimer

Pegler, G., Das, S., and Woodhouse, J.H. (1995): A seismological study of the eastern New Guinea and the western Solomon Sea regions and its tectonic implications. Geophysical Journal International 122:961-981

Ripper, I.D., Letz, H., and Moihoi, M. (1999): Felt effects of the 17 July 1998 Sissano Lagoon tsunami earthquakes. Geological Survey of Papua New Guinea Report 99/13.

Shuto, N. (1997): A natural warning of tsunami arrival. In G. Herbenstreit (ed.), Perspectives on Tsunami Hazard Reduction. Dordrecht: Kluwer, pp. 157-173.

Shuto, N. (1999): A short history of tsunami defense works in Japan. In Study on Regional and Historical Change of Tsunami Disasters -- In Commemoration of the 1993 Hokkaido Nansei-oki Earthquake Tsunami. Research Project, Grant-in-aid for Scientific Research (B) no. 09044126 (1998-1999), Faculty of Policy Studies, Iwate Prefectural University pp. 35-67.

Soter, S. (1999): Macroscopic seismic anomalies and submarine pockmarks in the Corinth-Patras rift, Greece. Tectonophysics 308:275-290

Sweet, S., and Silver, E.A. (2001): Seismic reflection images of the source region of the 1998 Papua New Guinea tsunami. In: Watts, P., Synolakis, C.E., and Bardet, J.P., (eds) Prediction of Underwater Slump and Slide Hazards. Rotterdam, Balkema.

Synolakis, C., Bardet, J-P., Borrero, J., Davies, H., Okal, E.,Silver, E., Sweet, S., and Tappin, D. (2002): The slump origin of the 1998 Papua New Guinea tsunami. Transactions of the Royal Society (in press).

Tanioka, Y. (1999): Analysis of the far-field tsunamis generated by the 1998 Papua New Guinea earthquake. Geophysical Research Letters 26:3393-3396

Tappin, D.R., Matsumoto, T., Watts, P., Satake, K., McMurtry, G.M., Matsuyama, M., Lafoy, Y., Tsuji, Y., Kanamatsu, T., Lus, W., Iwabachi, Y., Yeh, H., Matsumoto, Y., Nakamura, M., Moihoi, M., Hill, P., Crook, K., Anton, L., and Walsh, J.P. (1999): Sediment slump likely caused the 1998 Papua New Guinea tsunami. EOS Transactions of the American Geophysical Union 80, pp. 329, 334, 340.

Tappin, D.R., Watts, P., McMurtry, G.M., Lafoy, Y., and Matsumoto, T. (2001): The Sissano, Papua New Guinea tsunami of July 1998 -- offshore evidence of the source mechanism. Marine Geology 175: 1-23.

Taylor, R P, Emonson, D L, and Schlimmer J E (1998): Operation Shaddock - the Australian Defense Force response to the tsunami disaster in Papua New Guinea. Medical Journal of Australia 169:602-606.

Welsch, R.L. (1998): An American Anthropologist in Melanesia. A.B. Lewis and the Joseph N. Field South Pacific Expedition 1909-1913. 2 vols. Honolulu: University of Hawaii Press.