A Preliminary Study into the Tsunami Hazard faced by Southwest Pacific Nations

Full text

A Preliminary Study
into the
Tsunami Hazard
faced by
Southwest Pacific Nations
Christopher Thomas, David Burbidge and Phil Cummins
Risk and Impact Analysis Group
Geoscience Australia
22 June 2007

Contents
1 Executive Summary 2
2 Introduction to Tsunami 6
2.1 CausesofTsunami................................ 6
2.1.1 Earthquakes................................ 6
2.1.2 Other Tsunamigenic Mechanisms . . . . . . . . . . . . . . . . . . . . 9
2.2 Historical Tsunami in the Southwest Pacific . . . . . . . . . . . . . . . . . . 10
3 Method 11
3.1 TheSubductionZones.............................. 11
3.2 TheEarthquakes................................. 11
3.3 Computation of the Sea Floor Deformation . . . . . . . . . . . . . . . . . . 14
3.4 Simulation of the Tsunami Propagation . . . . . . . . . . . . . . . . . . . . 14
3.5 ModelOutputPoints............................... 14
3.6 ModelValidation................................. 16
4 Results 17
4.1 Suite1:Mw8.5.................................. 17
4.2 Suite2:Mw9.0.................................. 20
5 Future Work 23
A Significant Sources by Nation 26
A.1 Suite1:Mw8.5.................................. 26
A.2 Suite2:Mw9................................... 29
B Validation: Kuril Islands, 15 November, 2006 32
C Validation: Chile, 22 May, 1960 36
1

1 Executive Summary
The Indian Ocean tsunami of December 26, 2004 and subsequent smaller events have in-
creased awareness among emergency management authorities throughout the Pacific of the
need for more information regarding the hazard faced by Pacific nations from tsunami. As
part of the establishment of the Australian Tsunami Warning System, the Australian Gov-
ernment has undertaken an effort to support regional and national efforts in the southwest
Pacific to build capacity to respond to seismic and tsunami information. As part of this
effort, Geoscience Australia has received support from AusAid to partner with the South
Pacific Applied Geoscience Commission (SOPAC) to assist Pacific countries in assessing
the tsunami hazard faced by nations in the southwest Pacific. This report forms part of
the preliminary assessment of this hazard.
The tsunami threat faced by Pacific island countries consists of a complex mix of
tsunamis from local, regional and distant sources, whose effects at any particular location
in the southwest Pacific are highly dependent on variations in seafloor shape between the
source and the affected area. These factors make the design of an effective warning system
for the southwest Pacific problematic, because so many scenarios are possible and each
scenario’s impact on different islands is so varied. In order to provide national governments
in the southwest Pacific with the information they need to make informed decisions about
tsunami mitigation measures, including development of a warning system, a comprehensive
hazard and risk assessment is called for. The preliminary hazard assessment described
in this report has been made by undertaking numerical modelling of both the sea floor
deformation due to potential tsunamigenic earthquakes and the propagation of the resulting
tsunami. These models are used to predict maximum wave amplitudes offshore, at a water
depth of 50 metres, of the islands under consideration. (Throughout this report the term
maximum wave amplitude is used to denote the maximum absolute deviation from mean
sea level, which can occur either as a high crest or a deep trough. The term maximum
wave height denotes the maximum height from mean sea level to crest.)
Being a preliminary study, the scope is limited as follows:
1. The study focuses on tsunami caused by earthquakes and, more particularly, earth-
quakes occurring in subduction zones. While tsunami can be caused by other types
of earthquakes, as well as asteroid impacts, landslides and volcanic collapses and
eruptions, earthquakes in subduction zones are by far the most frequent source of
large tsunami. They are therefore the only events considered in this study.
2. This study treats offshore wave amplitudes and not onshore inundation. While it
is tsunami inundation that is normally considered hazardous, inundation is strongly
dependent not only on the offshore tunami height, but also on factors such as shal-
low bathymetry and onshore topography. A study of inundation therefore requires
detailed bathymetric and topographic data and involves very intensive numerical
computations which are beyond the scope of this study. The object of this study
is to answer the broader question: which Pacific nations might experience offshore
wave amplitudes large enough to potentially result in hazardous inundation, and from
which subduction zones might these waves originate? This information will inform
subsequent studies of a more detailed nature.
3. The results are presented in terms of maximum credible offshore wave amplitudes
and do not treat the likely occurrence of such events. A probabilistic analysis will be
the subject of a later study.
2

4. The assessment was confined to a study of the hazard of the following nations:
American Samoa New Caledonia
Cook Islands Niue
Federated States of Micronesia Palau
Fiji Papua New Guinea
French Polynesia Samoa
Guam Solomon Islands
Kiribati Tonga
Marshall Islands Tuvalu
Nauru Vanuatu
Figure 1 indicates the region of interest.
Figure 1: Region of interest of the study.
Because the significance of the wave amplitude depends on the depth of the water
in which it occurs, the amplitudes have been normalised to a notional depth of 50 metres
in order to be able to compare results from different regions. To aid in interpretation of
the results, offshore tsunami amplitudes have been categorized into 5 ranges (Table 1).
Categories corresponding to a higher range of offshore tsunami amplitudes can presumably
be associated with a higher level of hazard on coastlines of similar type. This categorization
has been adopted throughout this report. It is important to note, however, that these
ranges do not reflect the inundation that normally causes damage and/or fatalities and
which can often vary widely depending on local bathymetry and topography. The categories
indicated in Table 1 are therefore best viewed as indicating a relative level of hazard over
areas of broad geographic extent. For example, a Category 5 tsunami along the coast of
Papua New Guinea represents a higher level of hazard than a Category 2 tsunami along
the coast of New Caledonia.
It is premature, however, to interpret Table 1 in terms of impacts. This is especially
true for low-lying atolls such as Tuvalu and Kiribati. On the one hand, the lack of any high
ground may appear to make these islands especially vulnerable to tsunami, so that even a
Category 2 tsunami may be a cause for serious concern. On the other hand, because such
atolls often have steep drop-offs in which ocean depths increase very rapidly with distance
from the fringing reef, there may not be a pronounced shoaling effect, so that these islands
may never experience a large tsunami. Such considerations require much more modelling to
address and are beyond the scope of the present study, although it is intended that they be
3

considered in a later phase of this project. In particular, the information presented in this
report should not be used as a guide for responding to tsunami warnings. The information
presented here is preliminary and is only intended as a rough guide for prioritizing work
in subsequent phases of this project.
Category Normalised Amplitude (cm) Colour
10 – 25
225 – 75
375 – 150
4150 – 250
5>250
Table 1: Categorization of offshore tsunami amplitudes, normalised to an equiv-
alent depth of 50 metres. The “Colour” column refers to the colour used for
amplitudes of this category which will be used on all the figures shown through-
out this report.
Tsunami generated by two suites of simulated earthquakes were studied: Suite 1,
consisted of 187 moment magnitude (Mw) 8.5 earthquakes, and Suite 2 comprised 39 Mw
9.0 earthquakes (Figures 5 and 7, pages 11 and 13). The results are summarised in Table
2. For both suites of earthquakes the nations most affected were in the south and west
of the study area, including Vanuatu, Papua New Guinea, Guam, Solomon Islands and
Tonga, each of which recorded Category 5 amplitude tsunamis from the Suite 1 (Mw 8.5
earthquakes). This is due to the proximity of these countries to the subduction zones and
the orientation of the fault lines which acts to direct the tsunami towards these nations.
Category
Suite 1 Suite 2
American Samoa 3 3
Cook Islands 2 4
Fiji 3 5
French Polynesia 2 3
Guam 5 5
Kiribati 2 3
Marshall Islands 2 3
F.S. of Micronesia 4 4
Nauru 1 2
New Caledonia 4 4
Niue 3 4
Palau 3 4
Papua New Guinea 5 5
Samoa 4 4
Solomon Islands 5 5
Tonga 5 5
Tuvalu 2 3
Vanuatu 5 5
Table 2: Summary of results. Categories represent the highest amplitude
recorded for that nation, and should be interpreted according to Table 1.
4

The countries of Fiji, Federated States of Micronesia, New Caledonia and Samoa were not
as badly affected by the Suite 1 tsunami but still experienced significant waves (Category
3 to 4). Nations in the north and east of the study area, such as Kiribati, Marshall Islands,
Nauru, Cook Islands, French Polynesia and Tuvalu were much less affected by the Suite 1
tsunami and only experienced Category 1 or 2 sized waves.
As is to be expected, Suite 2 events produced greater effects than those of Suite 1
on all nations. Notable in this respect is Fiji which experienced Category 5 amplitudes
from Suite 2 events. However it should be noted that without further investigation it is
not possible to say that even Category 2 amplitudes will not produce significant run-up at
some locations.
The figures in Appendix A show that Suite 1 events (Mw = 8.5) from the subduction
zones in the eastern and far northern rims of the Pacific did not produce effects larger than
Category 2 on any of the nations studied. However some Suite 2 events (Mw = 9.0) in the
Peru-Chile, Aleutians and Kuril subduction zones did produce significant (Category 3 or
above) effects for some nations.
5

2 Introduction to Tsunami
Tsunami is a Japanese word - tsu meaning ‘harbour’ and nami meaning ‘wave’. They are
caused when large masses of water in the ocean are suddenly displaced by some event.
Gravity acts to return the displaced water to its equilibrium position and the disturbance
propagates as a wave, possibly for a very long distance. In the open ocean tsunami are
not obviously dramatic phenomena. In deep water even very significant tsunami will have
amplitudes of only a few tens of centimetres. Tsunami differ though from wind generated
waves in that they have very long wavelengths (distance from peak to peak), exceeding 100
kilometres, they involve movement of the water all the way to the ocean floor, and they
travel very quickly, of the order of 600 to 700 kilometres per hour or more in deep water.
Even a very large tsunami is likely to pass undetected by the occupants of a boat in the
open ocean. However tsunami carry a great deal of energy and they are able to transport
this energy very long distances. When these waves reach shallow water they slow down
and “bunch up” (their wavelength decreases), and their height increases dramatically, a
process known as shoaling. The maximum amplitude of the 2004 Boxing Day event was
estimated to be around 0.6 metres in the open ocean (Song et al, 2005) but the tsunami
ran up to heights of ten metres along many coasts, even those thousands of kilometres from
the earthquake (for example in India, see Narayan et al 2005).
2.1 Causes of Tsunami
2.1.1 Earthquakes
The most common causes of tsunami are earthquakes along oceanic subduction zones.
Subduction zones occur where two tectonic plates are colliding, and one of the plates is
sliding (subducting) beneath the other (Figure 2). As this happens friction between the two
plates may cause the upper plate to stick to the subducting plate and to become distorted
by its motion. Eventually the stress associated with this deformation accumulates to such
an extent that it can no longer be sustained by the frictional force between the plates,
resulting in a sudden movement of the upper plate as it springs back into place. This is
known as a subduction zone earthquake. This movement causes a sudden displacement
of the water lying above the plate, producing a tsunami. Not all earthquakes occur in
subduction zones, and other types of earthquakes have been responsible for generating
tsunami. However subduction zones have the potential to produce the largest earthquakes
and the most significant tsunami. For this reason this study focuses exclusively on the
oceanic subduction zones that could produce tsunami which impact on the area of interest.
These zones are indicated in Figure 3.
The southwest Pacific area is surrounded by the “Ring of Fire”, a region of intense
tectonic activity. Numerous volcanoes ring the Pacific Ocean and it is the location of some
of the largest and most seismically active faults in the world. Until the 2004 Indian Ocean
tsunami it was also the location of some of the most damaging tsunami in recent history,
such as the 1960 Chile tsunami and the 1998 Aitape tsunami (in the Sandaun province
on the north coast of Papua New Guinea). Figure 3 shows the known plate margins in
the Pacific coloured according to type. Subduction zone plate margins (shaded blue in
Figure 3) are known to be the source of most of the largest earthquakes in history. The
1960 Chile tsunami was caused by a magnitude 9.5 earthquake along the subduction zone
off the Chilean coast (ChT in Figure 3). The resulting tsunami caused major damage and
deaths as far away as Hawaii and Japan and minor damage was reported throughout the
6

Figure 2: Mechanism for tsunami generation in an oceanic subduction zone.
Pacific (Alport and Blong, 1995). The USGS estimate of the death toll from this event is
more than 2000, including 61 deaths in Hawaii, 128 deaths in Japan and 32 dead or missing
in the Phillipines. For more information about this event, see Appendix C. The South
American subduction zone has a long history of hosting large, tsunamigenic earthquakes
and it will almost certainly continue do so in the future.
The southwest part of the Ring of Fire has also been known to produce large, tsunami-
genic earthquakes, although none so far has been as large as the 1960 Chile earthquake.
An earthquake, probably combined with a submarine landslide, produced the 1998 Aitape
tsunami, which was the most lethal tsunami in this area in historic times. The Aitape
tsunami devastated several villages in Papua New Guinea and killed 2200 people.
The subduction zone to the east of PNG near the Solomons Islands has also been
known to produce large earthquakes. As this report was being written, an earthquake of
magnitude 8.1 occurred to the southwest of the Solomon Islands, near the South Solomon
Trench subduction zone. The earthquake and resulting tsunami caused substantial damage
and scores of deaths in the Solomon Islands. There have also been some unconfirmed
reports of damage in Papua New Guinea from this event.
Further to the east and south of the Solomons, the subduction zone near the New
Hebrides Trench off Vanuatu is another plausible site for a great (magnitude greater than
8) tsunamigenic earthquake. The history of the tectonic uplift of this area, as preserved in
coral growth bands, suggests that only moderate earthquakes (less than magnitude 8) have
occurred along short segments of this subduction zone so far (Taylor et al, 1990). However,
the potential for these segments of the Vanuatu subduction zone to rupture together in a
single, very large, earthquake should not be discounted.
Further to the east and south again, the Tonga-Kermadec Trench subduction zone,
7

Figure 3: Map of major plate boundaries in the Pacific Ocean with subduction
zones labelled as follows: AlT - Aleutian Trench, ChT - Chile Trench, CsT
- Cascadia Trough, HT - Hikurangi Trough, IBT - Izu-Bonin Trench, JpT -
Japan Trench, KmT - Kermadec Trench, KrT - Kuril Trench, MT - Mariana
Trench, MAT - Middle America Trench, NT - Nankai Trough, NGT - New
Guinea Trench, NHT - New Hebrides Trench, PhT - Philippines Trench, PrT
- Peru Trench, PyT - Puysegur Trench, RT - Ryukyu Trench, SST - South
Solomons Trench, TnT - Tonga Trench (Bird, 2003). Subduction zone plate
margins are shown in blue and are the source of the largest earthquakes in
history.
stretching from the Tonga Islands to New Zealand, has historically experienced earthquakes
of magnitude 8.0 to 8.3 which have generated local tsunami. Recent work on an earthquake
that occurred in 1865, however, suggests that the potential for the generation of far-field
tsunami in the Tonga-Kermadec Trench may have been underestimated (Okal et al, 2004).
South of New Zealand, much of the relative plate motion is in the strike direction
(that is, in the direction of the fault line), so that even when large earthquakes occur the
vertical component of the slip is small and they typically generate only small tsunami.
There is, however, a section of plate boundary just to the south of New Zealand known as
the Puysegur Trench, along which subduction has been occurring for the past ten million
years, a very short time in geologic terms (Meckel et al, 2005). Subduction zones with such
short histories are rare and their potential to produce large earthquakes and tsunami is
8

unknown. No major tsunamigenic earthquake has occurred on this trench in the historic
past, which would suggest either that subduction is mostly aseismic and no large earth-
quakes will occur, or that the subduction zone has been accumulating strain energy for
over 200 years and has the potential to rupture in a major earthquake.
In the north Pacific, the subduction zone off Cascadia is thought to have hosted an
earthquake around magnitude 9 in 1700 which generated a large tsunami that impacted
Japan (Atwater et al, 2005). More recently, large earthquakes along the Aleutian Islands
have generated waves that were damaging as far as Hawaii. The 1964 Prince William
Sound earthquake (magnitude 9.2) created a damaging tsunami in Alaska. According to
the USGS, that event took 125 lives (110 from the tsunami and 15 from the earthquake).
There were also tsunamigenic events in 1965 (the magnitude 8.7 Rat Islands earthquake)
and 1957 (the magnitude 8.6 Andreanof Islands earthquake). Both were large enough to
create damaging tsunami in the Alaskan region.
Japan has a record of seismicity going back nearly one thousand years from the
subduction zones off its coast. Events along the Nankai and Kamchatka zones are known
to create very large local tsunami, but as yet this area has not experienced any earthquake
that we can be confident had a magnitude of 9 or above.
In summary, there are major subduction zones in the west, north and east of the
Pacific Ocean basin that either have produced damaging tsunami in the past or could
plausibly produce them in the future. Therefore there is a real prospect that any of the
nations in the southwest Pacific might be exposed to a significant tsunami hazard.
2.1.2 Other Tsunamigenic Mechanisms
Subduction zone earthquakes are not the only possible sources of tsunami. As mentioned,
the Pacific is rimmed with volcanoes, many of which are submarine or near the coast.
Should one of these volcanoes erupt violently or collapse suddenly into the sea, there is a
real prospect of it producing a tsunami. For example, in 1888 a large tsunami, caused by
the flank collapse of Mount Ritter in Papua New Guinea, ran up to 12 – 15 metres and
wiped out villages on the western coast of New Britain (Johnson, 1987). The probability
of this occurring is difficult to estimate without detailed study of the volcanoes concerned
and may be a topic for future work.
The Aitape tsunami demonstrated that there is also the prospect of a landslide source
generating a tsunami. The largest tsunami in history occurred in 1958 when an earthquake
triggered a landslide into the Lituya Bay fjord in Alaska. The tsunami reached an altitude
of 510 metres on the other side of the bay (Mader, 2002). However, tsunami generated in
fjords usually remain trapped within them and are rarely considered to be a major threat
outside of the local region.
Submarine landslides on the continental slope are also a genuine hazard. These can
be triggered by a nearby earthquake, or may happen without warning. Historically they
have tended to produce large, but local tsunami (for example the 1998 Aitape and 1953
Suva tsunami) but there is the prospect of a major tsunami that impacts the far-field if
the landslide is sufficiently large. Again, this may be a topic for future work.
The largest tsunami of all are likely to be generated by asteroid impacts. It is
known that major extinction events marking the transitions between geologic eras, such as
that between the Cretaceous and Tertiary periods 65 million years ago, are the result of
9

massive impacts of comets or asteroids of about ten kilometres in diameter. Objects capable
of causing worldwide catastrophes are most certainly associated with massive tsunami,
with wave amplitudes far exceeding any tsunami in historic times. There is considerable
uncertainty about the generation and propagation of tsunami waves from intermediate-
sized objects with diameters in the range 100 metres to one kilometre. Smaller objects
almost certainly do not generate tsunami. Larger objects are clearly capable of penetrating
to the ocean floor and generating long-period waves that travel across the ocean with little
loss of energy. These are likely to be quite rare, but potentially devastating, events.
2.2 Historical Tsunami in the Southwest Pacific
Historical observations of tsunami run-up heights in the southwest Pacific are illustrated
in Figure 4. Run-up height is the highest level reached by a tsunami on land, measured
with respect to the sea surface at mean sea level. Figure 4 shows that the maximum run-
up heights observed historically in the southwest Pacific region are about 15 metres, and
that run-up heights of 10 metres or more have been observed in Fiji, the Solomon Islands,
Vanuata, the Marquesas Islands, Pitcairn Island, and Papua New Guinea (Gusiakav, 2005).
The same data show that over 200 events have caused tsunami in the region, with about
60 having led to run-up observations of one metre or more, and that run-up heights of five
metres or more appear to occur with a frequency greater than once every 10 years. Note
that these heights are much greater than any of the offshore tsunami heights computed in
this study. In general run-up heights can be much larger than offshore tsunami heights,
due to the momentum of the wave and the shoaling and focusing caused by the shape of
the seafloor and coastline.
Figure 4: Observations of tsunami in the southwest Pacific (Gusiakov, 2005).
Yellow bars indicate height of observed run-up, and green bars indicate heights
measured on tide gauges. The largest observations are along the northern coast
of Papua New Guinea and are 15 metres in height due to the 1998 Aitape
tsunami.
10

3 Method
The object of this study is to provide a preliminary evaluation of the hazard faced by
nations in the South-West Pacific in terms of maximum credible offshore wave amplitudes
for tsunami generated by earthquakes along the major subduction zones of the Pacific
rim. Briefly, this has been done by simulating earthquakes along the entire extent of the
subduction zones shown in Figure 3, computing the amount of the sea floor deformation
that would result from each of these events, simulating the propagation of the resulting
tsunami, and recording the time series of the wave heights at a large number of locations,
called model output points, offshore of the nations in the region. More details about each
of these facets of the study are provided in the following sections.
3.1 The Subduction Zones
The subduction zones considered in this study are those in Bird (2003) that have the
potential to produce tsunami that could impact on the region of interest. That is, all the
subduction zones shown (in blue) in Figure 3
3.2 The Earthquakes
Two suites of simulated earthquakes have been examined, each covering the extent of the
subduction zones under consideration. Suite 1 consists of 187 Mw 8.5 earthquakes based on
a uniform slip model on 200 by 200 kilometre sub-faults (Figure 5). The strike (Figure 6)
Figure 5: Locations of the 187 Mw 8.5 earthquakes comprising Suite 1. The
200 by 200 kilometre sub-faults are shown alternately in red and yellow.
11

Figure 6: Fault parameters.
is determined by its location on the plate model of Bird. In each case the dip was set to
14 degrees, which is the average for a subduction zone (Bird and Kagan, 2004). The rake
was set to 90 degrees representing a pure thrust fault which, for a given magnitude, length,
width and dip, results in the greatest deformation of the sea floor. The shear modulus
was set at 40 Gigapascals (it can vary from between 30 and 50 GPa; 40 was chosen as an
average value). The slip was set to be a uniform 3.94 metres across the rupture plane,
giving the earthquake a moment magnitude of 8.5.
Suite 2 comprises the 39 Mw 9.0 earthquakes shown in Figure 7. In most cases these
consist of five adjacent 200 by 200 kilometre sub-faults from Suite 1. The exception is
the event simulated on the Puysegur Trench south of New Zealand which consists of four
adjacent sub-faults. Most of the other fault parameters are as for Suite 1. Slip, however,
has been set to a uniform 4.44 metres across the fault plane (5.54 metres for the Puysegur
event). This gives the earthquake an approximate moment magnitude of 9.0.
It is possible that earthquakes of these magnitudes would have higher mean slip
values if their area is smaller. This may result in higher tsunami amplitudes, particularly
close to the source. An investigation into the effect of smaller area but higher slip events
is beyond the scope of this preliminary study.
12

Figure 7: Location of the 39 Mw 9.0 earthquakes (shown alternately in yellow
and red) comprising Suite 2. To provide complete coverage of the subduction
zones some events overlap (shown in orange).
13

3.3 Computation of the Sea Floor Deformation
For each simulated earthquake the sea floor deformation resulting from the assumed fault
parameters was computed using a seven-layered elastic model of the earth (Wang et al,
2006). The elasticity values for the seven layers were inferred from some seismic reflection
surveys to be fairly typical values for a subduction zone. If the elasticity values in a given
area differ substantially then this could result in a significantly different amplitude and
wavelength for the initial tsunami wave. Unfortunately, these values are poorly understood
for most parts of the Earth. It is assumed that the sea floor deformation is instantaneous
and that the sea surface directly above the fault instantaneously changes to match the sea
floor shape. This shape is taken as the initial condition for the finite difference method used
to simulate wave propagation. Instantaneous deformation may seem to be an unrealistic
condition but tests indicate that the model results are rather insensitive to the time taken
for the deformation to occur.
3.4 Simulation of the Tsunami Propagation
In the open ocean, where the water depth is large in comparison with tsunami wave am-
plitude, the propagation of a tsunami is essentially a linear phenomena and can be well
described by relatively simple equations, known as the linear long wave or shallow water
equations (“long” refers to the wavelength of the tsunami, “shallow” to the fact that the
water depth is small compared with the wavelength). For this study the propagation of
each tsunami was simulated using software written by the URS Corporation that uses
a staggered grid finite difference scheme to numerically solve the equations of the long
wave theory, using the initial condition computed as described in Section 3.3. The wave
is assumed to start at rest, that is, the water particles have no horizontal velocity at the
first time step. Total reflection is assumed at the coast and in this preliminary study no
consideration was taken of the effects of Coriolis force or bottom friction.
As a tsunami approaches shallow water its wavelength decreases, its amplitude in-
creases and its behaviour is no longer well described by the simple linear long wave equa-
tions. For this reason the current study does not consider the behaviour of tsunami in
shallow water, and focuses on the amplitude at locations with a water depth of approxi-
mately 50 metres.
The propagation of a tsunami is critically affected by the bathymetry. This study
employs a down-sampled 250 metre grid produced by Geoscience Australia spliced with
the US Navy’s bathymetry model, DBDB2. The resulting model, which we refer to as
GADBDB2, is illustrated in Figure 8. The grid resolution used was 2 arc minutes (about
3.7 kilometres). This bathymetry model and resolution have been shown to work well
for tsunami simulated for a probabilistic hazard map for the western coast of Australia
(Burbidge et al, 2007). While it seems likely the model is also suitable for the southwest
Pacific, more detailed studies are needed to confirm whether GADBDB2 is the best model
available for simulating southwest Pacific tsunami, and this should be the subject of future
work.
3.5 Model Output Points
A total of 1258 model output points were selected offshore of the Pacific nations as locations
at which to record the wave heights of the simulated tsunami (Figure 9).
14

Figure 8: Bathymetry used in the study (GADBDB2). This was produced by
splicing a grid produced by Geoscience Australia with DBDB2, the US Navy’s
bathymetry model.
As a tsunami approaches shallow water its amplitude increases. Thus model results
at output points at different depths are not directly comparable. In order to be able to
compare the impact of tsunami at different locations it is desirable to choose the output
points at (or close to) a uniform depth, and the standard depth employed in this study is
50 metres. However the resolution of the bathymetry grid means that it in some locations
it is not possible to choose sufficiently many points with depth close to 50 metres. The
approach adopted has been to select points with depths as close as possible to 50 metres
but to allow a range of depths in order to achieve a sufficient density of output points.
Green’s Law (see Mei et al, 2005) has been used in order to compare the model
results for output points at different depths. The version used here relates the tsunami
amplitudes A0and A1at two points with depths d0and d1along the path of propagation
as
A1=d0
d1
1
4
A0(1)
By setting A0to the simulated amplitude observed at an output point at depth d0and
setting d1to 50, Equation (1) gives A1as the simulated wave amplitude normalised to a
standard depth of 50 metres.
In reality energy from a tsunami is dissipated by friction at the bottom of the ocean,
15

Figure 9: Distribution of the 1258 output points at which wave heights of the
simulated tsunami were recorded.
and at the shoreline during shoaling and run-up. Neither of these factors are accounted for
in the model used for this preliminary study, which ignores the effect of bottom friction and
assumes total reflection at the shoreline. Because of this lack of accommodation of energy
dissipation in the model, large reflected wave amplitudes may be predicted at the output
points many hours after the first arrival. Therefore the modelling procedures adopted in
this study are most likely to well represent the waveform at an output point in only a
limited period following the arrival of the first wave. For this reason, unless otherwise
indicated, the maximum amplitudes recorded for this study are taken over the three-hour
period immediately following the first arrival of the tsunami. Although three hours is an
arbitrary figure, in practise maximum amplitudes of actual tsunami are unlikely to occur
later than this cut-off time.
3.6 Model Validation
The model described above has been validated against data recorded from ocean bottom
pressure gauges following the Kuril earthquake and tsunami of 15 November 2006. Details
are given in Appendix B. For the pressure gauges considered in this validation exercise,
it was found that the model predicted the maximum amplitude of the first wave with an
error, on average, of about 23%.
16

4 Results
Figures 10 and 12 show the normalised maximum wave heights for all tsunami from Suites
1 and 2 respectively. This gives an overview of the tsunami propagation from all of the
faults modelled.
Figures 11 (Suite 1), and 13 (Suite 2) show, for each output point in the region of
interest, the maximum amplitude that is predicted by the model from any of the simulated
tsunami. These figures do not indicate which event produced the maximum amplitudes,
but information regarding the effect of different faults on each nation can be obtained from
Tables 3 and 4, and the plots that are shown in Appendix A.
4.1 Suite 1: Mw 8.5
The nations most affected by the events in Suite 1 are in the south and west of the study
area, including Vanuatu, Papua New Guinea, Guam, Solomon Islands and Tonga. This is
largely due to the proximity and orientation of the subduction zones of Tonga-Kermadec,
New Hebrides, South Solomons, New Guinea, Philippines and Mariana. Nations to the
north and east of the study area, such as Kiribati, Marshall Islands, Nauru, Cook Islands,
French Polynesia and Tuvalu are less affected, with amplitudes measured in the tens rather
than hundreds of centimetres. Note however that without a more detailed study involving
Figure 10: Maximum wave heights for the 187 tsunami of Suite 1 (Mw 8.5),
normalised to a standard depth of 50m using Green’s Law. Maxima are com-
puted over the full 24 hour time frame of the simulation. The colour scale is
that described in Table 1.
17

inundation modelling it is not possible to assert that tsunami of these amplitudes are not
of concern. This would depend critically on the local bathymetry and topography.
Events from Suite 1 in the subduction zones of the east, north and northwest rim
of the Pacific have less effect on the region, either because of their distance, because the
orientation of the fault lines acts to direct tsunami energy away from the region, or because
of intervening bathymetric features. They rarely produced normalised amplitudes greater
than Category 1, and never greater than Category 2.
Nation Maximum Am-
plitude for all
Tide Gauges
for all Mw 8.5
Tsunami (cm)
Most Significant Source
Regions (amplitude
greater than 75cm at
50m depth or single most
significant source region
if no amplitude exceeds
75cm )
American Samoa 92 Tonga
Cook Islands 44 Tonga
Fiji 140 Tonga, Kermadec, New He-
brides
French Polynesia 50 Tonga
Guam 300 Mariana, Izu-Bonin
Kiribati 49 Peru
Marshall Islands 40 New Hebrides
F.S. of Micronesia 160 Mariana, New Guinea, Philip-
pines, South Solomons
Nauru 20 South Solomons
New Caledonia 160 New Hebrides, South
Solomons
Niue 110 Tonga
Palau 130 Philippines, New Guinea
Papau New Guinea 310 South Solomons, New Guinea,
Mariana
Samoa 160 Tonga
Solomon Islands 290 South Solomons, New He-
brides
Tonga 260 Tonga
Tuvalu 57 New Hebrides
Vanuatu 380 New Hebrides, South
Solomons, Tonga, Kermadec
Table 3: Most significant source regions for each nation, based on model out-
put points that recorded a maximum amplitude exceeding 75 centimetres for
Suite 1 (Mw 8.5).
18

Figure 11: Maximum amplitudes of simulated tsunami at model output points
for nations in the (a) north and east and (b) south and west sections of the
study area. The maxima are taken over all 187 Mw 8.5 tsunami in Suite 1.
The amplitudes have been normalised to a standard depth of 50 metres us-
ing Green’s Law, and are computed over the three-hour period following first
arrival. The colour scale is that described in Table 1.
19

4.2 Suite 2: Mw 9.0
Like Suite 1, the nations most affected by the Suite 2 events were those in the south and
west of the study area, as a result of the subduction zones in that region. However the
plots in Appendix A show that Suite 2 events in the Chile-Peru, Cascadia, Aleutians and
Kuril subduction zones produced significant (Category 3 or above) normalised amplitudes
for some nations. For example the simulations indicate that the Chile-Peru zone is a
significant source of hazard from Mw 9.0 events for Fiji and French Polynesia, as are the
Aleutian and Kuril subduction zones for Guam, Federated States of Micronesia, Papua
New Guinea, the Solomon Islands and Vanuatu. Significant normalised amplitudes were
produced in Papua New Guinea and the Solomon Islands from modelled events in the
Cascadia Subduction Zone.
Figure 12: Maximum wave heights for the 39 tsunami of Suite 2 (Mw 9.0)
normalised to a standard depth of 50 metres using Green’s Law. These maxima
are computed over the full time frame of the simulation.
This modelling pays no regard to the probability of events of various magnitudes
occurring on any of these subduction zones. While there is no doubt that the Chile sub-
duction zone can host an earthquake exceeding Mw 9.0 (the 1960 event was magnitude 9.5)
there is as yet no consensus on a reliable method of determining the absolute maximum
magnitude on any given subduction zone. We believe that most seismologists would agree
that a magnitude 8.5 event is plausible on any of the subduction zones considered here,
and that a 9.0 event is impossible to rule out. Hence we consider both magnitudes. A
probabilistic tsunami hazard study would consider a range of earthquake magnitudes and
weight them according to estimates of their likelihood, in a similar way to the method
described in Burbidge et al, (2007).
20

Figure 13: Maximum amplitudes of simulated tsunami at model output points
for nations in the (a) north and east and (b) south and west sections of the
study area. The maxima are taken over all 39 Mw 9 tsunami in Suite 2. The
amplitudes have been normalised to a standard depth of 50 metres using Greens
Law, and are computed over the three-hour period following first arrival.
21

Nation Maximum Am-
plitude for all
Tide Gauges
for all Mw 9
Tsunami (cm)
Most Significant Source Re-
gions (amplitude greater than
75cm at 50m depth or single
most significant source region if
no amplitude exceeds 75cm )
American Samoa 140 Tonga
Cook Islands 160 Tonga
Fiji 390 Tonga, Kermadec, New Hebrides,
South Solomons, Aleutian, Peru,
Chile
French Polynesia 120 Tonga, Kermadec, Peru, Chile,
Aleutian
Guam 430 Mariana, Philippines, Ryukyu,
Nankai, New Guinea, Aleutian,
Izu-Bonin
Kiribati 99 Peru
Marshall Islands 110 Kuril, Mariana
Micronesia 230 Mariana, Philippines, New Guinea,
South Solomons, Aleutians, Nankai,
Ryukyu
Nauru 31 South Solomons
New Caledonia 240 New Hebrides, South Solomons,
Tonga, Kermadec
Niue 210 Tonga
Palau 240 Philippines, Mariana, Ryukyu,
Nankai, New Guinea
Papau New Guinea 340 South Solomons, Mariana, New
Guinea, Nankai, Ryukyu, Aleu-
tians, Kuril, New Hebrides, Philip-
innes
Samoa 190 Tonga
Solomon Islands 310 South Solomons, New Hebrides,
Aleutians, Mariana, Ryukyu,
Nankai
Tonga 330 Tonga, New Hebrides, Kermadec
Tuvalu 88 New Hebrides
Vanuatu 450 New Hebrides, Tonga, Aleutians,
South Solomons, Kermadec, Kuril,
Nankai-Ryukyu
Table 4: Most significant source regions for each nation, based on model out-
put points that recorded a maximum amplitude exceeding 75 centimetres for
Suite 2 (Mw 9).
22

5 Future Work
Future work should focus on the following areas:
•Validation and Sensitivity Analysis. While some preliminary work on valida-
tion of the model has been considered here (Appendices B and C), more extensive
validation and sensitivity analysis should be undertaken. This should include:
– Bathymetry. A comparison of different bathymetric models should be under-
taken to establish which model gives the best accuracy for the calculation of
tsunami propagation in the southwest Pacific. This validation should make use
of detailed earthquake rupture models and tide gauge data, although, as dis-
cussed above, such data will have to be chosen with care in order to minimize
the effects of complex (and in some cases poorly known) shallow bathymetry.
– Historical Data. Comparison of model results with observations for historical
events is important for assessing the usefulness of the categorization of offshore
tsunami heghts in terms of impact in Table 1.
•Probabilistic Hazard Assessment. In order to properly inform disaster man-
agement and planning officials about the level of hazard, some assessment of the
probability of occurrence of tsunami is called for. Such a probabilistic assessment
can follow methodology similar to that in Burbidge et al (2007), and is intended to
be undertaken as part of the current project.
•Inundation Modelling. Damage and fatalities caused by tsunami are generally
associated with the height of the wave onshore, and the extent of inundation, rather
than on the offshore tsunami heights considered here. Inundation modeling is re-
quired to elucidate the relationship between the normalised maximum amplitudes
considered here (in water of 50 metres deep) and the effects of the tsunami when
it reaches the shore. Good estimates of the extent of inundation can only be made
using high resolution bathymetric and topographic data and very computer intensive
simulations. As part of the current project, an inventory will be made of bathymetry
and topography datasets held by SOPAC. It is intended that inundation modeling
will be consdidered in a later phase of this project, targeted at areas where data
suitable for inundation modeling exists in an area that the current hazard mapping
project has determined is vulnerable to tsunami.
•Exposure/Risk Assessment. Finally, the tsunami hazard mapping considered in
this study should be combined with information on exposure - population, siting
of critical facilities such as hospitals and airports - to determine the level of risk
the Pacific countries face from tsunami. Exposure information will be included in
inventory of SOPAC data holdings to be made as a component of the current project.
As with inundation modeling, however, a comprehensive assessment of risk is left to
a later phase of this project.
23

References
Allport, J.K. and R.J. Blong, 1995. The Australian Tsunami Database (ATDB),
Sydney: School of Earth Sciences, Macquarie University.
Atwater, B.F., Musumi-Rokkaku, S., Satake, K., Tsuji, Y., Ueda, K. and D.K. Ya-
maguchi, 2005. The Orphan Tsunami of 1700: Japanese Clues to a Parent
Earthquake in North America, Seattle: UW Press.
Barrientos, S.E. and S.N. Ward, 1990. The 1960 Chile earthquake: inversion for slip
distribution from surface deformation. Geophys. J. Int. 103, pp.589-598.
Bird, P., 2003. An updated digital model of plate boundaries. Geochem. Geophys.
Geosys. 4(3), 1027, doi:10.1029/2001GC000252.
Bird, P. and Y.Y. Kagan, 2004. Plate-tectonic analysis of shallow seismicity: Ap-
parent boundary width, beta, corner magnitude, coupled lithosphere thickness,
and coupling in seven tectonic settings. Bull. Seism. Soc. Am. 94, pp.2380-
2399.
Burbidge, D., Cummins, P. and R. Mleczko, 2007. A Probablistic Tsunami Hazard
Assessment for Western Australia: Report to the Fire and Emergency Services
Authority of Western Australia, Geoscience Australia.
Cox, D.C. and J.F. Mink, 1963. The tsunami of 23 May 1960 in the Hawaiian Islands.
Bull. Seism. Soc. Am. 53(6), pp.1191-1209.
Gusiakov, S., 2005. Integrated Tsunami Database for the World Ocean, Siberian
Division Russian Academy of Sciences.
ICMMG - see Institute of Computational Mathematics and Mathematical Geophysics.
Institute of Computational Mathematics and Mathematical Geophysics, 2006. His-
torical Tsunami Database for the Pacific, 47 B.C. to Present, Siberian Division
Russian Academy of Sciences, Novosibirsk, viewed 23 May, 2007,
<http://tsun.sscc.ru/htdpac>.
Johnson, R. W., 1987. Large-scale volcanic cone collapse: the 1888 slope failure of
Ritter volcano, and other examples from Papua New Guinea. Bull. Volcanol.
49, pp.669-679.
Keys, J.G., 1963. The tsunami of 22 May 1960 in the Samoa and Cook Islands. Bull.
Seism. Soc. Am. 53(6), pp.1211-1227.
Mader, C.L., 2002. Modeling the 1958 Lituya Bay mega-tsunami, II. Sci. of Tsun.
Haz. 20(5), pp.241-250.
Meckel T.A., Mann P., Sharon M. and M.F. Coffin, 2005. Influence of cumulative
convergence on lithospheric thrust fault development and topography along the
Australia-Pacific plate boundary south of New Zealand. Geochem. Geophys.
Geosys. 6, Q09010, doi:10.1029/2005GC000914.
Mei, C., Stiassnie, M. and D. Yue, 2005. Theory and Applications of Ocean Surface
Waves. Singapore: World Scientific.
Myles, D., 1986. The Great Waves. London: Robert Hale.
Narayan, J. P., Sharma, M. L. and B.K. Maheshwari, 2005. Effects of Medu and
coastal topography on the damage pattern during the recent Indian Ocean
tsunami along the coast of Tamilnadu. Sci. Tsunami Haz. 23(2), pp.9-18.
24

National Geophysical Data Centre, 2007. Global Tsunami Database, National Oceanic
and Atmospheric Administration, Washington, viewed 23 May 2007,
<http://ngdc.noaa.gov/seg/hazard/tsu db.shtml>.
NGDC - see National Geophysical Data Centre.
Okal, E.A., Borrero, J. and C.E. Synolakis, 2004. The earthquake and tsunami of
1865 November 17: evidence for far-field tsunami hazard from Tonga. Geophys.
J. Int. 157, pp.154-174.
Song, Y.T., Ji, C., Fu, L.-L., Zlotnicki, V., Shum, C.K., Yi, Y. and V. Hjorleifs-
dottir, 2005. The 26 December 2004 tsunami source estimated from satel-
lite radar altimetry and seismic waves. Geophys. Res. Lett. 32, L20601,
doi:10.1029/2005GL023683.
Taylor, F.W., Edwards, R.L., Wasserburg, G.J. and C. Frohlich, 1990. Seismic recur-
rence intervals and timing of aseismic subduction inferred from emerged corals
and reefs of the Central Vanuatu (New Hebrides) Frontal Arc. J. Geophys.
Res. 95(B1), pp.393-408.
Vitousek, M.J., 1963. The tsunami of 22 May 1960 in French Polynesia. Bull. Seism.
Soc. Am. 53(6), pp.1229-1236.
Wang, R., Martin, F.L., and F. Roth, 2006. Computation of deformation induced by
earthquakes in a multi-layered crust – FORTRAN programs EDGRN EDCMP.
Comp. and Geosc. 29, pp.195-207 (inc. erratum).
25

Appendix A Significant Sources by Nation
The maps in this appendix show, for each nation, the maximum (normalised) tsunami
wave height that every earthquake source modelled in this study produced for that nation.
The length and colour of each vertical bar indicate the maximum (normalised) amplitude
recorded at any of the output points of that nation for the tsunami generated by the source
at the base of the bar. The colour scale is as in Table 1. The purple cross indicates the
location of the nation. For example, Figure 14(a) shows that at American Samoa all Suite
1 sources produced only Category 1 tsunami with the exception of two faults on the Tonga
Trench, one of which produced a Category 3 tsunami, the other a Category 2.
A.1 Suite 1: Mw 8.5
Figure 14: (a – f ) Maximum amplitude over all model output points by nation
and source for Suite 1. The purple cross indicates the location of the nation.
26

Figure 14: (g – n) Maximum amplitude over all model output points by nation
and source for Suite 1. The purple cross indicates the location of the nation.
27

Figure 14: (o – r) Maximum amplitude over all model output points by nation
and source for Suite 1. The purple cross indicates the location of the nation.
28

A.2 Suite 2: Mw 9
Figure 15: (a – h) Maximum amplitude over all model output points by nation
and source for Suite 2. The purple cross indicates the location of the nation.
29

Figure 15: (i – p) Maximum amplitude over all model output points by nation
and source for Suite 2. The purple cross indicates the location of the nation.
30

Figure 15: (q – r)s Maximum amplitude over all model output points by nation
and source for Suite 2. The purple cross indicates the location of the nation.
31

Appendix B Validation: Kuril Islands, 15 November, 2006
On 15 November 2006 a magnitude 8.3 earthquake occurred at 153.230◦E 46.607◦N off the
Kuril Islands (Figure 16) which produced a tsunami across the Pacific with measured wave
heights of (for example) 88 cm at Hawaii, 176 cm at Crescent City, California, and 57 cm
at Samoa (according to the USGS). The modelling procedures used in this study have been
validated against a detailed finite fault model for this earthquake published by the USGS,
and data from ocean bottom pressure gauges deployed by the US Government’s National
Oceanic and Atmospheric Administration (DART gauges) and by the Japanese Agency
for Marine-Earth Science and Technology (JAMSTEC). The locations of these gauges are
shown in Figure 16.
Figure 16: Numerical model of the Kuril Islands earthquake and tsunami of
15 November 2006, with the locations of the DART (circles) and JAMSTEC
(triangles) ocean bottom pressure gauges.
Using the techniques discussed in Sections 3.3 and 3.4, the finite fault model pub-
lished by USGS was used to compute an estimate of the sea floor deformation due to the
Kuril Islands event and the propagation of the resulting tsunami was modelled. Results
of the numerical simulations were calculated at the location of the ocean bottom pressure
gauges for comparison with the actual water levels recorded by the gauges. It was necessary
to filter the pressure gauge data to remove extraneous signals, primarily ocean tides and
high frequency components that are not part of tsunami waveforms. The comparisons are
presented on the following pages, with the filtered pressure gauge signals in blue and the
model results in red. Overall the model results agree well with the pressure gauge data,
particularly for the first peak and trough of the waveform. While there are some differences
in phase, the amplitudes and arrival times are in good agreement. On average the maxi-
mum amplitudes agree to within 23%. As well as possible shortcomings in the modelling
procedure (including limitations in the bathymetric model used), such differences as there
are can be attributed to a disparity between the actual earthquake rupture and the USGS
finite fault model, and the filtering process used for removing tidal components from the
observed pressure data.
32

Figure 17: (a – f) Comparison of ocean pressure gauge data (blue) and model
results (red) for the Kuril earthquake and tsunami of 15 November 2006.
33

Figure 17: (g – l) Comparison of ocean pressure gauge data (blue) and model
results (red) for the Kuril earthquake and tsunami of 15 November 2006.
34

Figure 17: (m) Comparison of ocean pressure gauge data (blue) and model
results (red) for the Kuril earthquake and tsunami of 15 November 2006.
The spike in the signal from DART buoy 46402 is likely to be a data error and is
probably not part of the tsunami signal.
The DART and JAMSTEC buoys are in water several thousand metres deep. It has
not been possible to validate the modelling procedure against data from shallower water
because such data come from instruments such as harbour tide gauges that typically are
in water only a few metres deep, where the model is not valid. There is some concern that
the bathymetric grid resolution employed may not be high enough to always adequately
represent the waveforms, particularly in shallower water where their wavelength will be
shorter. A preliminary investigation has been made into this potential problem and a
more detailed analysis will be carried out as part of future work.
35

Appendix C Validation: Chile, 22 May, 1960
On 22 May 1960 the largest earthquake ever recorded with modern seismographs (Mw 9.5)
occurred off the coast of Chile (at approximately 286.5◦E, 41◦S). This produced a Pacific
wide tsunami that caused widespread damage, particularly along the coasts of Chile, Hawaii
and Japan.
The techniques of Section 3 have been used to estimate the sea floor deformation
produced by this event and to model the resulting tsunami, based on a uniform slip model
for the earthquake presented by Barrientos and Ward (1990). Figure 18 shows the results
of this modelling.
In Western Samoa the tsunami was most pronounced at Fagaloa Bay (Upolu) where
the maximum run-up (the highest point above sea level reached by the wave) was estimated
to be about 2.5 metres (Keys, 1963). Minor damage to buildings was sustained and it
was reported that the waves carried fuel drums 73 metres inland. Residents, who had
been forewarned by announcements on the local radio station, had taken refuge on higher
ground and no loss of life occurred. The rest of Western Samoa appears to have escaped
undamaged, probably because of screening by offshore reefs, which are absent from Fagaloa
Bay (Keys, 1963). In American Samoa the tsunami reached a maximum run-up height of
over three metres at Pago Pago village (Allport and Blong, 1995). Buildings were moved
off their foundations and a house washed into the bay. No loss of life was reported.
In Fiji reports appear to be confined to the effects in Suva harbour. The maximum
runup was reported to be about 0.5 metres, and the tsunami induced a powerful surge in
the harbour. Many boats sustained damage, but no loss of life was recorded (Allport and
Blong, 1995).
In French Polynesia many of the islands are protected by outer reefs and deep lagoons,
with rather steep bathymetry offshore, and in most cases only slight damage was sustained.
No loss of life was recorded. The average runup surveyed in Tahiti was 1.7 metres. Larger
runups, up to 3.4 metres, were recorded along the north shore of the island which is more
exposed to the open ocean (Vitousek, 1963). The greatest effects in French Polynesia
were felt in the Marquesas Islands which have few outer reefs and more gradual changes
in offshore bathymetry. Runups of at least 4.5 metres (possibly up to nine metres) were
observed. Destruction of buildings near the shore was reported (Vitousek, 1963).
The Hawaiian islands suffered extensive damage and 61 deaths. The island of Hawaii
bore the brunt of the damage, mostly around Hilo where all the deaths and most of the
damage occurred. Damage to buildings was extensive in this area, with almost total
destruction in an area of nearly 300 acres. Rocks weighing up to 22 tonnes were carried
180 metres inland. There was considerable damage to houses and commercial buildings on
Maui, with some being destroyed. The other islands suffered only minor damage (Cox and
Mink, 1963).
In Japan, run-ups of up to 6.4 metres were recorded (ICMMG, 2006), causing
widespread flooding and damage. About 5000 homes were lost leaving 50,000 people home-
less, and between 180 and 190 lives were lost (Myles, 1986).
The effects of the tsunami were felt in Australia, with boats removed from their
moorings in Sydney, Brisbane, Newcastle and Evans Head. Minor damage and flooding
was reported in New Zealand and Papua New Guinea (Allport and Blong, 1995).
In Chile, of course, damage and loss of life were extensive. A maximum run-up of
36

Figure 18: Normalised modelled maximum wave heights of the 1960 Chilean
tsunami based on uniform slip model given by Barrientos and Ward (1990).
Wave heights have been normalised to 50 metres depth and the maximum is
taken over the full time period of the simulation.
37

25 metres was recorded (NGDC, 2007) and thousands of people drowned. Towns were
completely obliterated and debris was carried more than three kilometres inland (Myles,
1986).
These historical accounts should be read in conjunction with Figure 18. Little damage
and no loss of life was reported for coastlines where the incident waves were of Categories
1 or 2 (for example Australia, Papua New Guinea and, for the most part, New Zealand).
However the magnitude of the effect of the tsunami was critically dependent on local fea-
tures of the coastline, the height of the tide when the tsunami arrived and the density of
the population in vulnerable areas. Fiji, Samoa and parts of French Polynesia received
Category 3 (even 4) amplitudes and sustained only minor damage, whereas similar am-
plitude waves in Hawaii and Japan caused extensive damage and loss of life. Category 5
amplitudes, such as those received along the coast of Chile, resulted in extensive damage
and loss of life.
38