Feature

Tsunami Lessons

Last year's tsunami in Papua New Guinea holds sobering lessons for New Zealand.

Willem de Lange

On 17 July 1998, a moderate earthquake occurred in the Saundaun province of north-western Papua New Guinea. This rather insignificant event was recorded by the Pacific Tsunami Warning Centre and warnings were issued of the possibility of small local sea level oscillations.

However, what is now called the Saundaun Tsunami was a catastrophic event locally, killing over 2,000 people and making it one of the most deadly tsunamis this century. Early reports described particularly gruesome scenes with human remains dangling from trees, or mauled by dogs and crocodiles.

The magnitude of the devastation was surprising given the moderate size of the earthquake (M_w = 7.1) and the extreme geographic concentration of the affected area. The location of the epicentre was identified variously as significantly inland and offshore, with the final revision indicating the epicentre was practically on the coastline. The aftershock distribution indicates that the earthquake may have been associated with a gently dipping fault immediately offshore of Sissano Lagoon. Preliminary characteristics of the source of the earthquake, obtained by Japanese seismologists, suggested a fault area of 30 by 15km, with a slip of about 2m.

International Investigators

An International Tsunami Survey Team (ITST) with 13 scientists, a medical specialist and two film crews from Australia, Japan, New Zealand and the US was deployed. In different, earlier incarnations, the ITST has performed inundation surveys for nine major tsunami catastrophes in the past decade, occurring in Indonesia, Japan, Mexico, Nicaragua, the Philippines, Peru and Russia.

Inundation surveys involve measuring tsunami inundation height and inland penetration distances, whenever watermarks and other indicators can be found. These data are highly ephemeral and can be easily lost due to storms or recovery operations. The ITST uses standard surveying gear, GPS receivers for locating the inundation marks consistently on maps, corers for sediment sampling, and portable seismometers. For the Saundaun Tsunami, literacy and language difficulties meant interviews rather than written surveys were conducted.

The team split into two and headed for Aitape by various routes, taking measurements along the way. Travel was difficult with stormy weather and unbridged river-crossings blocked by swollen rivers.

From Aitape, the teams headed for Sissano Lagoon by boat, helicopter and four-wheel drive. Sissano Lagoon is fronted on the ocean side by two narrow sand spits with a fairly narrow mouth and limited ebb-tide delta close to the western end of the lagoon. The lagoon is almost semicircular with the back shore about 4 km from the mouth. The villages of Arop were located on a sand spit at the eastern end of the lagoon, and the Sissano villages lay about 1 km west of the lagoon entrance.

Profiles measured by the ITST indicated that the sand spits had a maximum elevation Casuarina species and a tree known locally as laulau, most vegetation on the eastern spit was severely to totally destroyed. There was no evidence of any of houses or their remains anywhere along the sand spits, other than a few inclined or tilted rows of foundation poles.

The average flow depth over the spit near the lagoon mouth was found to be 10 m, while near Arop it was 15 m. The team inferred from the damage and sedimentary structures left behind that the water current induced by the tsunami over the sand spits was at least 10 m/s, probably peaking to twice this value; eyewitness reports suggested 15-20 m/s current velocities. It should be noted that the force of the tsunami current on an object is 1,000 times greater than the force on the same object by a wind of the same speed.

Three team members sailed inside the lagoon to try to determine how far the wave had penetrated. Navigation was difficult because of the amount of trees, stumps and building debris. Debris from the community on the sand bar reached to the back of the lagoon, which is swampy and fronted by mangroves. The height of the wave there was probably less than 1m, as there were several mostly intact houses on stilts that were taken to have been unaffected by the tsunami.

The teams flew helicopter sorties, when equipment could be spared from the relief effort, to determine inland penetration in the immediate vicinity of the lagoon for areas that were inaccessible on foot due to the very adverse conditions. They also interviewed numerous eyewitnesses who helped put together the sequence of events.

In total, 80 inundation data points were measured and 30 different topographic transects, covering densely an area of more than 40km, with several points measured at Vanimo about 100km west of Sissano and Wewak, 180km east. Generally, the inundation heights diminished rapidly about 10km east and west from the worst hit area which extended between Arop and Sissano. At the village of Serai, about 15km from the lagoon mouth, the inundation height was about 4m, and there was no damage in the village.

It is clear that most of the tsunami was fairly narrowly focused onto a 40km strip of coastline between the Rainbaum (Arnold) River and Aitape, and diminished rapidly to either side. This narrow focus of the tsunami energy was surprising for a seismic tsunami, and collected data suggests that the tsunami spread radially from a source almost directly off Sissano and then dissipated rapidly.

The first wave arrived within 5-10 minutes from the mainshock. It caused a noticeable recession of the water, and led some people to move towards the sea. The elevation wave following was reported as a wall of water, making a thunderous noise resembling a jet aircraft.

The first wave was followed shortly after by another two waves, the third of which was clearly smaller than the first two. The three waves occurred only five minutes apart, suggestive of a highly dispersive wave train, rather than individual waves generated by strong aftershocks or sequential rupture.

It was not possible to confirm the likely cause of the tsunami, but most team members felt that a secondary mechanism such as a submarine landslide was the probable cause, rather than the earthquake itself. Further work will continue to model the source and determine the potential of future tsunami from the same subduction zone, and to find out whether the possibility for a trans-Pacific tsunami exists.

Recommendations

The team made a number of recommendations to local authorities.

People should not relocate to locales which are fronted by water and backed by rivers or lagoons. Memorials should be built at the worst stricken sites as a reminder and to discourage future habitation of high -risk locales.

Schools, churches and other critical facilities should never be located closer to 400m from the coastline, and preferably 800m in at-risk areas.

The local Casuarina tree species withstood the tsunami wave attack significantly better than coconut palms, and Casuarina forests should be planted in front of coastal communities, whenever possible.

There should be evacuation drills annually on the disaster's anniversary so that all people in at-risk areas know that if they feel the ground moving they should run as far from the beach as possible.

Every family an at-risk area should have a designated Casuarina tree with a ladder or carved steps to allow vertical evacuation of the able, when there is no other option.

Local authorities reported that the tsunami had brought out many superstitious beliefs, and had been blamed by some on impiety. Following an announcement about the possibility of closely spaced pairs of tsunami events, there were rumours that a second tsunami was imminent.

The team was asked to explain to survivors the causes of tsunami, and discussed some simple warning signs that a tsunami may be imminent. They stressed that ground motion is not always a precursor: anybody who lives close to the coastline should be on the lookout for unusual water motions.

The team also tried to explain some of the unusual phenomenon seen. Reports of the sea bubbling and of foul smelling gas and warm water stinging the eyes were attributed to the tsunami stirring up the stagnant bottom waters of Sissano Lagoon. The lagoon is normally calm, and quite possibly a layer of vegetable matter would have accumulated on the bottom, building up an oxygen-poor environment where noxious gases may have developed in the sediments.

Some eyewitnesses described the tsunami as "a water-fire infernal mountain of water with fire sparkles flying", and this was cited by locals as an explanation of severe burns observed among the dead and some survivors. The team explained that most likely the tsunami had triggered bioluminescence, a phenomenon also known as "sea-fire" where dinoflagellates and other marine organisms emit light when stirred. The team had observed dramatic examples of bioluminescence in the wake of their boat over several nights while onboard, and speculated that this may have created the appearance of sparks flying as the wave approached.

The burns reported were not from heat but from friction, which probably caused significant skin loss as the victims were dragged over hundreds of meters among debris and trees. Other victims had the skin flayed from exposed portions of their bodies, giving the appearance of being sand-blasted. This was attributed to the sediment carried by the tsunami waves.

A review of historical data indicates that the Sissano area has experienced severe tsunami in the last 200 years. Here and elsewhere on the north coast there is a mythology of tsunami that doubtless stems from pre-historic experiences. However, in historic time, which dates from the first mission settlements in the 1880s, tsunami appear to have been few and far between and, until the Saundaun Tsunami, none had caused serious damage.

Understanding this event will hopefully lead to the production of inundation maps for the north coast of PNG. Having access to inundation maps helps the local authorities appropriately locate schools, hospitals, and other critical facilities.

The Sediment Speaks

Of particular interest to New Zealand was the examination of the sedimentary deposits left by the tsunami. When sediment is deposited by a tsunami and preserved, a geologic record of that tsunami is created. By looking at the sedimentary sequence in an area, the occurrence of prehistoric tsunami can be identified to extend the relatively short or non-existent historical record. Because it is not yet possible to predict when a tsunami will occur, obtaining a record of prehistoric events may be one of the only means to assess future risk.

A second ITST group looked at sediment deposited and its characteristics, measuring land elevation, flow depth, flow direction and tsunami deposit thickness and character along cross-shore transects at four sites.

Tsunami deposits were common and were identified as grey-coloured sand typically overlying a brown soil containing many roots. In places, plants were bent over and buried by the sand, or removed completely. The lower part of the tsunami deposit sometimes included rip-up clasts of the underlying muddy soil. The recently deposited tsunami sand probably came from offshore, as numerous sand dollars were found near the surface of the deposit. Another common characteristic of recently deposited tsunami sand was normal grading, where the size of the sand grains decreases from the bottom to the top in the deposit.

The Arop School transect was chosen as a good site to look down into the sedimentary record for evidence of past tsunami. Away from sandy river or ocean sources, the depositional environment at this site was probably that of a quiet water lagoon. Several long push cores were taken at 135m from the shoreline. The top metre of each core was characterised by a 5-10cm thick normally-graded sandy layer at the surface (deposited by the July 17 tsunami); this is underlain by 20-30cm of a brown muddy soil. Beneath the soil is a uniformly gray muddy sediment. A thin coarse silt/fine sand layer is present approximately 120cm below the surface. This layer is 3-4cm thick and was found at a similar depth in each of the cores, probably deposited by a past tsunami. Material just below this fine-sand layer was obtained for possible dating to determine the approximate age of this thin layer.

Implications for New Zealand

Clearly the type of infrastructure present along the New Zealand coast is quite different to that along the least-developed coastal area in PNG, so we would not suffer the same number of casualties if a similar event were to occur here. Nonetheless there are a number of important aspects of this disaster that are relevant to New Zealand.

Firstly there are significant similarities between the physiography of the affected area and portions of the New Zealand coast. The east coast of the North Island and northern South Island has a narrow continental shelf that drops steeply into a submarine trench, similar to the Saundaun province of PNG.

Only one definite landslide tsunami is known from the east coast of the North Island: a 15 m high surge caused by a landslide during the 1931 Napier Earthquake. However surveys of the offshore bathymetry and stratigraphy have shown the presence of very large prehistoric submarine landslides off Hawke Bay. These landslides involve volumes at least an order of magnitude larger (50km³) than that postulated for the Saundaun Tsunami. An analysis of tsunami hazard undertaken for the region by the Hawkes Bay Regional Council was based on earthquake-induced fault motions, and did not consider landslide effects.

The West Coast of the South Island also has a narrow continental shelf that drops off to abyssal depths. Historical records of New Zealand tsunami events show that several landslide-generated tsunami have occurred along this coast. There are also a couple of unexplained events that were probably caused by landslides. So far these events have mostly involved subaerial landslides that have produced smaller tsunami than the Saundaun event.

Historical and other evidence indicate that New Zealand has experienced landslide-generated tsunami in the past. These have been quite large, but localised in their extent. It is probable that New Zealand will experience landslide generated tsunami in the immediate future -- within the next 30 years. It is likely that this event will be in the range 5-15m, but in some areas it is possible that the waves can be much larger, greater than 25m.

Most of the New Zealand coast has a higher relief than the Saundaun province. Coastal dunes that I am familiar with are typically 6-15m above mean sea level. However there are regions where development has greatly reduced dune height to provide views and access to the beach. These regions have a correspondingly greater hazard.

New Zealand construction standards are more stringent than most of those evident in the affected region. Some of the buildings such as the school and mission at Sissano seem to have been built to a similar or better standard than holiday homes along the New Zealand coast --the tsunami waves completely destroyed these buildings.

The damage to the buildings was caused by three main processes:

• floating and subsequent collisions
• the direct impact of the wave and the associated currents
• impact by debris carried by the wave

The buildings were mostly unaffected when the flow depths were less than 1 m, as in many cases the main bearers of the floors and walls were above this level. Many New Zealand houses are closer to the ground. The buildings suffered increasing damage as the flow depth increased from 1m to 3m, and at greater depths the buildings were totally destroyed. Given the height of most of the New Zealand coast, the flow depths produced by a similar event would mostly be less than 3 m.

A significant proportion of the casualties could have been avoided by timely evacuation. This was not possible for the population on the narrow sand spits, but it was an option for the rest of the affected region. Unfortunately, lack of tsunami awareness meant that most stayed in the hazardous region to watch the water recede, and many were unable to outrun the following wave.

It is probable, given historic behaviour in New Zealand, that a significant number of people would also fail to evacuate in time here. The standard Civil Defence publicity associates tsunami with strong earthquakes. In areas where submarine landslides are considered a realistic possibility, it is desirable to revise this to include all felt earthquakes and any unusual water motions.

Finally, although the actual cause of the Saundaun Tsunami has not yet been determined, it is evident that the triggering earthquake did not display any special characteristics that would have allowed prediction of the resulting tsunami. The PTWC assessed this event as only causing small localised sea level disturbances.

A similar event here would produce the same assessment. The ITST could not determine any remote method that could have provided a suitable warning for the affected region. The two difficulties were the absence of any earthquake signature that the ITST could associate with a large tsunami, and the short travel time (15 minutes or less). The same would apply in New Zealand.

Dr Willem de Lange Waikato University